KR102560437B1 - Artificial intelligence based digital heathcare apparatus, and computer-implemented method for the same - Google Patents

Abstract

Disclosed are an artificial intelligence-based digital healthcare apparatus and a computer-implemented method for the same, which can greatly shorten medical treatment time while maintaining or improving medical treatment accuracy by a hospital (doctor). According to the present invention, the computer-implemented smart healthcare method comprises: a step of using user authentication information of a user to register device information of a first device to measure a biosignal of the user; a step of determining whether biometric data including a measurement value of the biosignal are collected from the first device within a measurement period set for the biosignal; and a step of transmitting a measurement request for encouraging the user to measure the biosignal using the first device to the first device and a second device which is a communication terminal of the user capable of communicating with the smart healthcare apparatus if it is determined that the biometric data are not received from the first device within the measurement period.

Description

Artificial intelligence-based digital healthcare device and computer-implemented method for it

The present invention relates to an artificial intelligence-based digital healthcare device and a computer-implemented method therefor, and more particularly, to effectively collect and manage personal health-related data from a user device, and to use an artificial intelligence-based model to It relates to a technology that automatically analyzes and monitors a person's health condition to support personal health management.

Globally, the demand for non-face-to-face medical services including telemedicine and interest in personal health management are increasing significantly with the recent corona pandemic as a catalyst, and blood pressure, heart rate, electrocardiogram (ECG), body composition, etc. Dissemination of various measuring devices (e.g., smart watches as wearable devices) that can easily and quickly measure various types of sensor signals is spreading.

As the bio-signal measurement technology progresses by leaps and bounds, the creation and management of bio-data (personal health data), also called PGHD (Patient-Generated Health Data) or PHR (Personal Health Record), is also becoming easier than in the past.

However, personal health data measured (collected) by or from an individual simply remains at the role of providing extremely fragmentary information related to the individual's health condition. A significant number of individuals neglect active health management, such as visiting a hospital or seeing a doctor, unless they are aware that there is a serious abnormality in their health condition. am. To name a few representative reasons why individuals fail in managing their own health, the absence of a hospital (primary doctor) that has built up medical trust through long-term use, difficulties or annoyance in searching for information on appropriate hospitals, and reservations over the phone, etc. There are difficulties in the process, etc.

Republic of Korea Patent Publication No. 10-2019-0107905 (2019.09.23)

The present invention has been made to solve the above problems, and sets a regular or non-regular appropriate measurement period for each type of biosignal, thereby inducing the user to measure his or her biosignal based on artificial intelligence. An object of the present invention is to provide a healthcare device and a computer-implemented method therefor.

In addition, the present invention automatically collects measurement values of biosignals of each user from at least one first device, analyzes a time series of biosignals collected through artificial intelligence-based logic, and learns the pattern, thereby improving the health of each user. An object of the present invention is to provide an artificial intelligence-based digital healthcare device for monitoring symptoms (possibility) and a computer-implemented method therefor.

In addition, the present invention is an artificial intelligence-based digital healthcare device for notifying any medical staff (eg, hospital, primary care physician) of personal health data (or analysis results thereof) of a user determined to have health abnormalities, and a computer for the same. It aims to provide an implementation method.

In addition, the present invention collects medical examination information through communication with a communication terminal (eg, mobile phone) for each user in parallel with the process of collecting biometric data for each user, and processes the collected medical examination information to obtain health based on time-series data of the biometric data. An object of the present invention is to provide an artificial intelligence-based digital healthcare device and a computer-implemented method therefor, which are utilized for the purpose of verifying anomalies or correcting measurement cycles of vital signs.

Other objects and advantages of the present invention can be understood by the following description, and will be more clearly understood by the examples of the present invention. Furthermore, it will be readily apparent that the objects and advantages of the present invention may be realized by means of the instrumentalities and combinations thereof set forth in the claims.

A computer-implemented intelligent healthcare method according to an aspect of the present invention includes the steps of registering device information of a first device for measuring the user's bio-signal using user authentication information of the user; determining whether biometric data, including the measured value of the biosignal, is collected from the first device within a measurement period set for the measurement period; and determining whether the biometric data is collected from the first device within the measurement period. When it is determined that no reception has been received, at least one of the first device and the second device, wherein the second device is the user's communication terminal capable of communicating with the intelligent healthcare device, is sent to the biological body using the first device. and transmitting a measurement request for encouraging the user to measure a signal.

Registering the device information may include issuing, to the first device, an access token in which the user authentication information is combined with the device information of the first device. The access token is used for encryption of the measurement value of the biosignal by the first device.

The computer-implemented intelligent healthcare method, upon receiving the biometric data received from the first device in a state encrypted by the access token, determining validity of the access token, determining that the access token is valid. When determining, decrypting the biometric data received from the first device using a decryption key corresponding to the access token, and using the decrypted biometric data, a first personal health time-series data set of the user - the above The method may further include updating the first personal health time series data set including a history of change over time of the measured value.

The computer-implemented intelligent healthcare method may further include modifying the measurement period set for the bio-signal based on at least one of the first personal health time-series data set and external environment information.

The computer-implemented intelligent healthcare method includes generating an artificial intelligence model for processing the first personal health time-series data set and using the artificial intelligence model, based on the first personal health time-series data set , calculating an abnormality score related to the user's health condition.

The generating of the artificial intelligence model may include performing a comparison between at least one normal range determined for the bio-signal and at least one measurement value included in the first personal health time-series data set to obtain a first data related to the bio-signal. Generating a first model that outputs a score of 1, learning with a plurality of reference health time-series data sets for the physiological signals previously acquired from a plurality of individuals other than the user, and processing the first personal health time-series data set generating a second model that outputs a second score related to the bio-signal, and processing the first personal health time-series data set as learned with the first personal health time-series data set to generate a second model related to the bio-signal Generating a third model that outputs 3 scores, and outputting a weighted average of the first score, the second score, and the third score as the ideal score, the first model, the second model, and the It may include combining a third model.

The generating of the second model may include supervised learning using the plurality of reference health time-series data sets, wherein each measurement value included in the plurality of reference health time-series data sets is classified as a normal measurement value or an abnormal measurement value. generating a first candidate model through learning), when the ratio of abnormal measurement values of the plurality of reference health time series data sets is greater than or equal to a first reference ratio and less than a second reference ratio, of the plurality of reference health time series data sets Processing the plurality of reference health time series data sets through execution of at least one of down sampling for normal measurement values and up sampling for abnormal measurement values, and through supervised learning using the plurality of processed reference health time series data sets. Generating a second candidate model, when the ratio of abnormal measurement values of the plurality of reference health time series data sets is less than the first reference ratio, unsupervised learning using only normal measurement values of the plurality of reference health time series data sets ( generating a third candidate model through unsupervised learning; evaluating performance of each of the first candidate model, the second candidate model, and the third candidate model using a first test data set; and and selecting one of the first candidate model, the second candidate model, and the third candidate model evaluated to have the highest performance as the second model.

The generating of the third model may include generating a fourth candidate model through supervised learning using the first personal health time series data set; When the ratio is greater than or equal to the third reference ratio and less than the fourth reference ratio, the first individual through execution of at least one of down-sampling for normal measurement values and up-sampling for abnormal measurement values of the health time series data set of the first individual Processing the health time-series data set and generating a fifth candidate model through supervised learning using the processed first personal health time-series data set, wherein the ratio of abnormal measurement values in the first personal health time-series data set is If the ratio is less than 3, generating a sixth candidate model through unsupervised learning using only normal measurement values of the first personal health time series data set; using a second test data set; Evaluating performance of each of the fifth candidate model and the sixth candidate model, and selecting one of the fourth candidate model, the fifth candidate model, and the sixth candidate model evaluated as having the highest performance in the third candidate model. It may include a step of selecting a model.

In the combining of the first model, the second model, and the third model, a first weight, a second weight, and a third weight are adjusted according to the number of measurement values recorded in the first personal health time series data set. steps may be included. The sum of the first weight, the second weight, and the third weight is 1, and whenever the number of measurement values recorded in the first personal health time series data set increases by the reference number, the first weight The second weight may decrease by a first adjustment value within a range of 1, the second weight may decrease by a second adjustment value within a second range, and the third weight may increase by a third adjustment value within a third range.

The computer-implemented intelligent healthcare method, when it is determined that a medical examination for the user is necessary, a digital medical examination to be presented to the user - the digital medical examination includes a plurality of medical examination items, each of which is a multiple choice or short answer type generating -, sending a medical examination request for the digital medical questionnaire to the second device in order to call an intelligent active dialogue agent installed in the second device, through interaction with the intelligent active dialogue agent Receiving a medical examination result for the digital medical examination form from the second device, and processing the medical examination result to obtain a second personal health time series data set of the user, wherein the second personal health time series data set is at least different from the biosignal. A step of updating - including a history of change over time of one health-related item may be further included.

According to at least one of the embodiments of the present invention, a user may be induced to measure his or her own biosignal by setting a regular or non-regular appropriate measurement period for each type of biosignal. Accordingly, it is possible to solve the user's trouble of remembering when and how much to measure his or her own bio-signals.

In addition, according to at least one of the embodiments of the present invention, biometric data (including measured values of biometric signals) for each user is automatically collected from at least one first device, and biometric data collected through artificial intelligence-based logic is analyzed. By analyzing the time series and learning the patterns, it is possible to monitor the symptoms (probability) of health abnormalities for each user. Since these abnormal health symptoms are provided to a hospital (doctor) during a hospital visit or remote medical treatment, treatment time can be significantly reduced while maintaining or improving the accuracy of treatment by the hospital (doctor).

In addition, according to at least one of the embodiments of the present invention, medical examination information is collected through communication with a communication terminal (eg, mobile phone) for each user in parallel with the process of collecting biometric data for each user, and the collected medical examination information is processed It can be used to verify health abnormalities based on time-series data of biosignals or to modify the measurement cycle of biometric data.

The effects of the present invention are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the description of the claims.

The following drawings attached to this specification illustrate preferred embodiments of the present invention, and together with the detailed description of the present invention serve to further understand the technical idea of the present invention, the present invention is the details described in such drawings should not be construed as limited to
1 is a diagram showing the configuration of a remote health care system according to the present invention by way of example.
2 is a diagram showing the configuration of the intelligent healthcare device shown in FIG. 1 by way of example.
FIG. 3 is a flowchart illustrating a method for registering a user device in the intelligent healthcare device shown in FIG. 1 by way of example.
4 to 6 are flowcharts illustrating a method of collecting measurement values of biosignals from the first device by way of example.
FIG. 7 is a flowchart illustrating a method for the intelligent healthcare device shown in FIG. 1 to collect user medical examination information through interaction with a user device.
FIG. 8 is a flowchart illustrating a method of modifying a measurement period for each type of biosignal executed by the intelligent healthcare device shown in FIG. 1 by way of example.
10 is a flowchart exemplarily illustrating a method of generating the second model described above with reference to FIG. 9 .
FIG. 11 is a flowchart exemplarily illustrating a method of generating the third model described above with reference to FIG. 9 .
12 includes an apparatus for recommending a visit by a primary care physician according to real-time monitoring of a patient's personal health data according to a first embodiment of the present invention or a hospital recommendation apparatus according to real-time monitoring of a patient's personal health data according to a fourth embodiment of the present invention. It is a drawing showing the overall configuration of
13 is a flowchart illustrating representative steps of a method of actively reserving a patient's visit schedule according to a second embodiment of the present invention.
14 is a flowchart embodying step S1340 in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention.
FIG. 15 is a flow chart further including an additional embodiment related to a measurement request of personal health data after step S1360 in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention.
16 is a flow chart further including an additional embodiment related to calculating treatment time after step S1360 in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention.
17 is a flowchart illustrating representative steps of a method for recommending a hospital for a patient-supervising doctor relationship according to a fourth embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Advantages and features of the present invention, and methods for achieving them, will become clear with reference to the embodiments described below in detail in conjunction with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the present embodiments make the disclosure of the present invention complete, and common knowledge in the art to which the present invention belongs. It is provided to fully inform the holder of the scope of the invention, and the present invention is only defined by the scope of the claims. Like reference numbers designate like elements throughout the specification.

Unless otherwise defined, all terms (including technical and scientific terms) used in this specification may be used in a meaning commonly understood by those of ordinary skill in the art to which the present invention belongs. In addition, terms defined in commonly used dictionaries are not interpreted ideally or excessively unless explicitly specifically defined.

Meanwhile, terms used in this specification are for describing embodiments and are not intended to limit the present invention. In this specification, the singular form also includes the plural form unless otherwise specified in the phrase.

As used herein, "comprises" and/or "comprising" means that a stated component, step, operation, and/or element is one or more other components, steps, operations, and/or elements. Existence or additions are not excluded.

1 is a diagram showing the configuration of a remote healthcare system according to the present invention by way of example, and FIG. 2 is a diagram showing the configuration of the artificial intelligence-based digital healthcare device shown in FIG. 1 by way of example.

Referring to FIG. 1, the remote healthcare system 10 includes at least one user device 1, at least one medical person device 2, and an artificial intelligence-based digital healthcare device (3, hereinafter referred to as 'intelligent healthcare device'). '). The intelligent healthcare device 3 may be provided in the form of an external server or platform. In the remote healthcare system 10 according to the present invention, multiple users and multiple medical personnel can participate as medical consumers and medical providers. .

The user device 1 may be classified into a first device 1a, a second device 1b, or a third device 1c.

The first device 1a is a measurement device (eg, point of care testing (POCT) equipment) configured to measure at least one kind of biosignal and store the measured value of the biosignal. Biometric signals include, for example, blood pressure, blood pressure, heart rate, electrocardiogram, blood sugar, and body composition. An arbitrary user may simultaneously or temporally measure two or more biosignals using two or more first devices 1a. The measurement value of the bio-signal (which may be referred to as personal health recorded (PHR)) further includes the time and/or location at which the bio-signal was measured (eg, GPS information) as well as the characteristics (eg, strength) of the bio-signal itself. can include The first device 1a may transmit measurement values of biosignals for each user to the intelligent healthcare device 3 through a wired/wireless network such as the Internet.

The second device 1b is a communication terminal that is connected to the intelligent healthcare device 3 through a wired or wireless network such as the Internet and performs two-way communication with the intelligent healthcare device 3 . The second device 1b may transmit health-related data for each user in addition to bio-signals to the intelligent healthcare device 3 and receive data related to the user's health management from the intelligent healthcare device.

An intelligent active dialog agent may be installed in at least one of the first device 1a and the second device 1b. The intelligent active conversation agent can be referred to as IPCA (Intelligent Proactive Conversational Agent), and is a chatbot that runs on the basis of artificial intelligence and supports users to conveniently use the health management service provided by the intelligent healthcare device (3). It is one of a kind.

The third device 1c is a type of user device having both a function of measuring bio-signals and a function of communicating with the intelligent healthcare device 3, such as a smart watch. The second device 1b may refer to the third device 1c.

The medical device 2 encompasses the terminal of the attending physician, which will be described later, and is connected to the intelligent healthcare device 3 through a wired or wireless network such as the Internet to perform two-way communication with the intelligent healthcare device 3. The medical device 2 may display the personal health data (or analysis results thereof) for each user provided from the intelligent healthcare device 3 on a graphic maintenance interface. The medical device 2 transmits the need for face-to-face or non-face-to-face treatment of a specific user to the intelligent healthcare device 3 in response to a request from a medical person (such as a doctor), and the intelligent healthcare device 3 informs the medical person Reservation guide information for face-to-face treatment or non-face-to-face treatment is transmitted to the user device 1 of a specific user requested by the user. The intelligent health care device 3 mediates a reservation process (eg, search for a time available for reservation, selection of a doctor available for reservation) between the user device 1 and the medical device 2, and transfers reservation completion information to the user device 1 and Notifies both medical personnel devices 2.

Referring to FIG. 2 , the intelligent healthcare device 3 includes a storage unit 11, a communication unit 12 and a control unit 13.

The storage unit 11 includes read only memory (ROM), random access memory (RAM), erasable programmable read only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, and static SRAM (SRAM). RAM), a hard disk drive (HDD), a solid state drive (SSD), and the like. Also, the storage unit 11 may include at least one database for accumulatively recording the first personal health time-series data set and/or the second personal health time-series data set for each user. The storage unit 11 may include a computer-readable recording medium in which a computer program for executing a computer-implemented method to be described later is stored.

The communication unit 12 is provided to perform two-way communication by connecting to the user device 1 and the medical worker device 2 through a wired or wireless network. The communication unit 12 may receive biometric data including measured values of biometric signals for each user from the user device 1 , and the received biometric data may be stored in the storage unit 11 .

The control unit 13 is operably coupled to the storage unit 11 and the communication unit 12, and the storage unit 11 and the communication unit 12 to perform each function (operation) of the intelligent health care device 3 ) can be controlled. Here, that any two components are operably coupled means that they are interconnected to enable data (signal) transmission from one of the two components to the other.

The control unit 13 may be implemented by hardware, firmware, software, or a combination thereof. When implemented using hardware, an application specific integrated circuit (ASIC) or digital signal processor (DSP) ), DSPD (digital signal processing device), PLD (programmable logic device), FPGA (field programmable gate array), etc., when implemented using firmware or software, a module, procedure, or function that performs the above functions or operations Firmware or software may be configured to include the like.

The controller 13 executes a computer program stored in the storage unit 11, and when the computer program is executed, the controller 13 induces the controller 13 to perform operations for health management for each user.

Hereinafter, various computer-implemented methods executed by the intelligent healthcare device 3 will be described with reference to FIGS. 3 to 11 .

FIG. 3 is a flowchart illustrating a method for registering a user device 1 in the intelligent healthcare device 3 shown in FIG. 1 by way of example.

Referring to FIG. 3 , in step S310, the controller 13 receives a device registration request from the first device 1a connected to the intelligent healthcare apparatus 3 through a wired/wireless network. The device registration request may include device information of the first device 1a and user authentication information of a user having permission to use the first device 1a (eg, ownership). User authentication information may be, for example, a user ID and password assigned to each user.

In step S320, the controller 13 determines whether the user authentication information is valid in response to the device registration request. The validity of the user authentication information may be verified by comparison with user authentication information pre-recorded in the storage unit 11 through a user's member sign-up procedure, etc., and/or by an authentication protocol such as Oauth 2.0. If the value of step S320 is "yes", it proceeds to step S330. If the value of step S320 is "no", it proceeds to step S350. Authentication for the first device 1a may be performed only once for each user. A device agent (device manager) may be installed in the first device 1a, and two or more users may be registered as having normal use rights for the first device 1a by the device agent. The device agent may be a computer program that manages (eg, adds, deletes, modifies, etc.) a list of users having permission to use the first device 1a. Although not shown, each user may be registered as a user having permission to use two or more first devices, and each first device may be registered for measuring bio-signals of two or more users.

In step S330, the control unit 13 maps the device information of the first device 1a to the user authentication information of the user and records it in the storage unit 11.

In step S350, the controller 13 issues an access token in which user authentication information of a user is combined with device information of the first device 1a to the first device 1a. At this time, the controller 13 may generate a decryption key corresponding to the access token. As a result, the first device 1a is registered in the intelligent healthcare device 3 as a device for measuring the user's biological signal.

The access token may be a token encrypted by a one-way encryption algorithm such as bcrypt, and may be used to encrypt the measured value of the biosignal by the first device 1a. That is, the first device 1a uses the access token issued from the intelligent healthcare device 3 as an encryption key to encrypt the measurement value of the biosignal, and then converts the encrypted measurement value into intelligent biometric data including the access token. It can be transmitted to the healthcare device (3). Access tokens are issued one by one for each first device 1a-user pair, and may be given a valid period. Even if the access token is exposed to the outside, it is impossible to infer device information and user authentication information included therein due to one-way encryption characteristics. The validity period of the access token can be changed by the user.

Once the access token is issued, until the validity period of the access token (eg, 3 months) elapses, the first device 1a uses the access token itself without a login procedure based on user authentication information to provide intelligent healthcare services. It is possible to automatically connect to the device 3, and can transmit the measured value(s) recorded in the first device 1a from the last transmission time of the biometric data to the present time to the intelligent healthcare device 3.

The control unit 13 may automatically renew the validity period of the corresponding access token (or reissue the access token) when user authentication information associated with the access token is changed or the validity period of the access token expires, and the decryption key accordingly. It can also auto-renew.

As described above, when it is determined that the user authentication information is invalid, step S350 proceeds. In step S350, the controller 13 transmits an authentication failure notification for the device registration request to the first device 1a.

4 to 6 are flowcharts illustrating a method of collecting measurement values of biosignals from the first device 1a by way of example. The methods of FIGS. 4 to 6 may be executed for the first device 1a registered to the intelligent healthcare device 3 according to the method of FIG. 3 .

Referring to FIG. 4 , in step S410, the controller 13 determines whether biometric data from the first device 1a has been received. The biometric data includes a measurement value of a biosignal measured from a user by the first device 1a, the measurement date and/or measurement location of the measurement value (location of the first device 1a), and the first device It may further include the access token issued in (1a). If the value of step S410 is "yes", it proceeds to step S420. If the value of step S410 is "No", the process proceeds to step S510.

In step S420, the control unit 13 determines the validity of the access token included in the biometric data received from the first device 1a. If the value of step S420 is "yes", it proceeds to step S430. If the value of step S420 is "No", it may mean that the validity period of the previously issued access token for the first device 1a has expired or that user authentication information mapped to device information of the first device 1a has been changed. there is. If the value of step S420 is "No", the process proceeds to step S610.

In step S430, the controller 13 decrypts the biometric data (encrypted with the access token) received from the first device 1a using the decryption key corresponding to the access token.

In step S440, the controller 13 updates the first personal health time-series data set of the user by using the measured value of the decoded biometric data. The first personal health time-series data set includes a history of changes over time in measurement values of the user's bio-signals, and may be referred to as a PHR time-series data set.

Referring to FIG. 5 , when the value of step S410 is “No”, step S510 proceeds. In step S510, the control unit 13 determines whether or not the measurement period set for the biosignal has elapsed. That is, in step S510, it is determined whether a measurement period (time) has elapsed since the last measurement of the user's biological signal by the first device 1a. If the value of step S510 is "yes", it proceeds to step S520. If the value of step S510 is “No”, it means that the biometric data from the first device 1a is received within the measurement period, and in this case, step S410 may be returned.

In step S520, the control unit 13 transmits, to at least one of the first device 1a and the second device 1b, a measurement request for encouraging the user to measure a physiological signal using the first device 1a. do. At least one of the first device 1a and the second device 1b, in response to the measurement request, calls (activates) the intelligent active conversation agent installed therein, and measures the biosignal to the user in the form of a pop-up message. Interaction with the user may be executed, such as activating a conversation interface for notifying necessity and proceeding with measurement of a biosignal.

Referring to FIG. 6 , when the value of step S420 is “No”, step S560 proceeds.

In step S610, the control unit 13 updates the previously issued access token for the first device 1a and a decryption key corresponding to the access token.

In step S620, the controller 13 transmits an approval request for biometric data (received in step S410) to the second device 1b. The approval request may be a request asking the user whether or not to approve the biometric data from the first device 1a previously received in step S410 as indicating the measured value of the user's biosignal.

In step S630, the controller 13 determines whether or not the biometric data (received in step S410) has been approved by the second device 1b within a predetermined period of time. If the value of step S630 is "yes", the process proceeds to step S640. If the value of step S630 is "No", the process proceeds to step S650.

In step S640, the controller 13 issues the access token updated in step S610 to the first device 1a. That is, the previously issued access token to the first device 1a is replaced with a new access token. The first device 1a may retransmit the biometric data transmitted to the intelligent healthcare device 3 in step S410 in response to a new access token being issued through step S640.

In step S650, the controller 13 sets the registration state of the first device 1a to an approval rejection state. When the first device 1a is set to an approval rejection state, the controller 13 may continue to block biometric data from the first device 1a.

The user sends a release request for the approval rejection state of the first device 1a to the intelligent healthcare device 3 through manipulation of the device agent installed in at least one of the first device 1a and the second device 1b. , and the intelligent healthcare device 3 may release the approval rejection state for the first device 1a in response to the release request. The user receives device IDs, types, communication protocols, and access token information (e.g., issuance date and expiration date) of the user devices 1 registered in the device agent installed on at least one of the first device 1a and the second device 1b. ), measurable bio-signal items, location (by hour, GPS-enabled devices are automatically set), and information such as device usage can be checked and managed (changed, deleted).

FIG. 7 is a flowchart illustrating a method for the intelligent healthcare device 3 shown in FIG. 1 to collect user medical examination information through interaction with the user device 1 by way of example. The intelligent healthcare device 3 may collect the user's medical examination information in parallel with the user's biometric data collection process.

Referring to FIG. 7 , in step S710, the control unit 13 determines whether a medical examination is necessary for the user. The medical examination of the user is conducted regularly or irregularly, and when the user directly requests the intelligent healthcare device 3 through the second device 1b to proceed with the medical examination, or the doctor as the attending physician uses the medical device 2 When a medical examination of the user is requested to the intelligent healthcare device 3 through, it may be determined that the medical examination of the user is necessary. In addition, the need for medical examination is based on the elapsed time from the date and time of the last medical examination of the user, the first personal health time series data set to date, the user's hospital reservation information, and / or external environment information. can be determined by the model. For example, there is a possibility that the health status of the user is degraded regularly or by at least one of the first personal health time series data set and external environment information (eg, temperature, humidity, wind speed, fine dust) every time a predetermined period elapses If the daily temperature difference is high (eg, the daily temperature difference between an area where the user is estimated based on the GPS information of the first device 1a or an area of interest registered by the user is greater than an allowable value), it may be determined that medical examination is necessary.

In step S720, the control unit 13 generates (creates) a digital medical questionnaire to be presented to the user. The digital medical questionnaire may include a plurality of questionnaire items, each of which is a multiple choice or subjective questionnaire. The control unit 13 uses an artificial intelligence-based questionnaire content generation model, based on at least one of the user's general health information, the first personal health time series data set, and external environment information, at least one questionnaire item of the digital medical questionnaire. can be automatically generated. General health information is pre-entered by the user or pre-collected from the medical information server, such as the user's gender, weight, height, blood type, age, eyesight, past/current medications, past/current symptoms, past/current symptoms, past/current health information. Current medical history (e.g., diagnosed disease, surgical history), etc. The storage unit 11 of the intelligent healthcare device 3 may include a paperweight database, and a paperweight template in which at least two or more paperweight items are aligned may be recorded in the paperweight database. The controller 13 configures a digital medical questionnaire based on measured value(s) for each type of biosignal being collected and managed for the user through the first device 1a and external environment information such as changing seasons and pandemics. It is possible to automatically select an item and create the contents (question) of the selected medical examination item.

In step S730, the control unit 13 transmits a medical examination request for the digital medical examination generated in step S720 to the second device 1b to call the intelligent active conversation agent installed in the second device 1b. The intelligent active conversation agent of the second device 1b outputs an interface for conducting a medical examination process on the digital medical examination form from the second device 1b in response to a request for medical examination from the intelligent healthcare device 3. , induces user participation in the paperweight.

In step S740, the control unit 13 receives the medical examination result (medical examination data) for the digital medical examination form from the second device 1b through interaction with the intelligent active conversation agent of the second device 1b. The medical examination results may be collected for each medical examination item of the digital medical examination table, or may be collectively collected after the medical examination for all medical examination items of the digital medical examination table is completed.

According to papers and research results known before the filing date of the present invention (Cartreine et al., 2010; Dayan et al., 2007, Locke et al., 1992, Richman et al., 1999), people do not share sensitive information with others. When it comes to disclosure to humans, they tend to be more comfortable with computer agents who have fewer privacy concerns and less fear of negative reviews than with real people, and researchers in various fields, such as healthcare and marketing, are more comfortable with face-to-face interviews with individual respondents. In contrast, it was found that computer-based surveys or interviews using computers are useful for increasing the response rate and efficiently collecting personal sensitive information. The intelligent active conversation agent is a kind of software provided to replace the face-to-face interview between the user and the medical personnel. For a user scheduled to visit a hospital, a medical questionnaire is filled out in advance through the intelligent active conversation agent, and the medical personnel device (2) By sharing, there is an advantage in that the accuracy of the answer is increased and the treatment time is shortened compared to the patient directly answering the doctor's question during treatment hours at the hospital on the same day.

From now on, the interview process through interaction with an intelligent active conversation agent will be described.

In response to the request for medical examination, the intelligent active conversation agent provides text information and basic information about the digital medical examination, such as the reason why the user needs the medical examination, the purpose of the medical examination, and the response method (this information is pre-recorded for each medical examination in the medical examination database) into text information and At least one of the audio information is guided to the user. The audio information is output through text-to-speech (TTS) for text information, and may be selectively executed according to a user's manipulation of the second device 1b.

The content of each medical examination item of the digital medical questionnaire is also presented to the user in the form of at least one of text and audio by an intelligent active conversation agent, through which the user can easily grasp the contents of each medical examination item.

The intelligent active conversation agent measures the elapsed time from the time when a specific medical examination item is output from the second device 1b during the medical examination of the user, and the user is found to have difficulty in answering the specific medical examination item. In this case, additional explanations and examples pre-mapped to the corresponding paperweight items are automatically provided to the user. In addition, for the questionnaire items identified by the user as difficult to answer immediately, according to the user's request, they are classified as those scheduled to be answered later and recorded in the list of unanswered questionnaire items, and the list of unanswered questionnaire items is stored for a predetermined period of time (for example, After one hour) has elapsed or before the hospital appointment date arrives, a separate interview may be conducted.

Intelligent Active Conversation Agents can exclude from the digital questionnaire items that have already been answered by the user and are not likely to change (e.g. What is your STD?) Have you ever had a disease or died of that disease?, What was the highest amount of alcohol you drank per day?), information about the user's answers to other past questionnaire items related to the heading questionnaire item (e.g., date and time of response, response content) visually/auditically, it is possible to prevent unnecessary medical examination for the same questionnaire item and induce active participation of the user in the questionnaire item requiring an answer.

The intelligent health care device 3 uses artificial intelligence technology (e.g., context-based questionnaire generation) through a digital questionnaire including at least one questionnaire item (multiple choice) dynamically generated according to the user's condition or situation. It is possible to perform a customized questionnaire for each user. For example, the control unit 13 detects an abnormal behavior in the measured value pattern of the user's recently measured biosignal (eg, blood pressure, heart rate), changes the surrounding environment, such as a change of season when the diurnal temperature difference is 10 degrees or more, and the user's biological body. When detecting a characteristic change (e.g., a sudden increase/decrease in weight of more than 3 kg in the last month), a natural language generation (NLG) deep learning model is used to automatically generate questionnaire items related to the context. can do. To this end, the platform continuously learns and updates deep learning models for automatic generation of paperweight items.

If the user checks (responses) that the questionnaire item dynamically generated by the intelligent healthcare device 3 through the intelligent active conversation agent is inappropriate, the control unit 13 determines the user's answer to the corresponding questionnaire item. is excluded from learning to increase the performance of the deep learning model.

The initial learning of the deep learning model for dynamically generating questionnaire items is performed using automated techniques such as crawling or manually collected data from a group of previously secured medical experts (e.g., doctors, health examination centers, other medical professional literature, etc.). It is conducted using a natural language processing (NLP) deep learning model to analyze subjective items (e.g., feel free to write if you have any uncomfortable body parts or special symptoms) rather than multiple choice, and allow multiple users to It is built as learning data to create new questions based on the answers commonly entered for (collecting learning data at low cost), and deep learning models can be trained regularly or irregularly using this.

When the second device 1b transmits the medical examination data including the answers entered by the user for each medical examination item to the intelligent healthcare device 3, it encrypts using a unique access token issued separately like biometric data, can transmit The intelligent healthcare device 3 may generate and store a decryption key corresponding to an access token associated with the paperweight data, and may use this to decrypt the paperweight data from the second device 1b.

In step S750, the controller 13 processes the result of the medical examination received in step S740 to update the second personal health time series data set of the user. The second personal health time series data set includes a history of change over time of at least one health-related item different from the user's bio-signal measured by the first device 1a. Health-related items are items that change over time (e.g., weight, height, amount of alcohol consumed in the past month, exercise time in the past week, medications taken in the past week, recent diet, pain level in a specific body part, etc.) It may be, and like biometric data, it may be stored in a paperweight database in the form of a time series. The second personal health time series data set is used as additional feature data (for learning) when learning using an artificial intelligence model to monitor (detect) abnormal patterns of biosignals, and thereby artificial intelligence for monitoring (detection) of abnormal patterns improve the accuracy of the model.

FIG. 8 is a flowchart illustrating a method of modifying a measurement period for each type of biosignal executed by the intelligent healthcare device 3 shown in FIG. 1 by way of example. As described above, the user's first device 1a can measure one or more types of biosignals, and the measurement period can be individually set and adjusted for each biosignal measurable by the first device 1a. In a specific user-first device 1a pair, the initial value of the measurement period for each bio-signal may be predetermined as a predetermined value, and may be adjusted regularly or irregularly while monitoring the user's personal health-related information. there is.

Referring to FIG. 8 , in step S810, the control unit 13 acquires at least one period adjustment parameter for the biosignal. The period adjustment parameter refers to a parameter referred to for adjusting a measurement period of a biosignal associated with a specific user-first device 1a pair. For example, when the biosignal is blood pressure, parameters related to personal health information (e.g., height, weight, previously diagnosed diseases, medical examination data (contents of answers), user location, movement route, and sudden change in user location by 1 km or more within unit time) Etc.), parameters representing external environment information (e.g., temperature, season, change of season, epidemic epidemic, etc.) and parameters representing the analysis result of the first personal health time series data set (e.g., anomalies in measured values of biosignals) /pattern, etc.) may be obtained as a period adjustment parameter. External environment information may be periodically (eg, every hour) collected by the control unit 13 by accessing an external environment monitoring server (eg, a public portal) through the communication unit 12 .

In step S820, the control unit 13 calculates a correction amount for the bio-signal measurement period based on the at least one period adjustment parameter obtained in step S810. Here, the calculated correction amount has 0, a positive value, or a negative value, and when the periodic adjustment parameter represents the user's health improvement (possibility), the absolute value of the correction amount increases to a positive value, and the periodic adjustment parameter In the case that indicates deterioration (possibility) of the user's health, the correction amount may increase in absolute value to a negative value, and in the case where the period adjustment parameter indicates the maintenance of the user's health, the correction amount may be set to zero.

In step S830, the control unit 13 corrects (renews) the measurement period by applying (summing up) the correction amount calculated in step S820 to the current measurement period determined for the biosignal. The measurement period corrected through step S830 is used as a measurement period of the user's bio-signal until the next meeting is corrected. That is, the modified measurement period may be used in step S510 described above with reference to FIG. 5 .

The control unit 13 may transmit the modified measurement period to the second device 1b through the communication unit 12 . A calendar application for measuring and managing bio-signals provided by the intelligent healthcare device 3 may be installed in the second device 1b. A measurement execution alarm may be activated through a screen and/or a speaker of the device 1b. In addition, the second device 1b may record the measurement date and time of the biosignal by the first device 1a in a calendar application, and automatically or manually inform the user of the biosignal measurement history up to now. can In addition, the calendar application may synchronize information related to the measurement period and measurement result of the biosignal with the third party calendar application through interworking with the third party calendar application.

Simple notification information transmitted from the intelligent healthcare device 3 to the second device 1b may be delivered to the user through a messenger service and/or SNS provided by a third party instead of an intelligent active conversation agent.

FIG. 9 is a flowchart exemplarily illustrating a method of monitoring a user's health abnormality executed by the intelligent healthcare device 3 shown in FIG. 1 .

Referring to FIG. 9 , in step S910, the controller 13 generates an artificial intelligence model for processing the first personal health time-series data set (which may be updated in step S440). Step S910 may include step S912, step S914, step S916 and step S918.

In step S912, the controller 13 performs a comparison between at least one normal range determined for the biosignal and at least one measured value included in the first personal health time series data set, and obtains a first score related to the biosignal. A first model to be output is created. The normal range (or other risk range) for each type of bio-signal is established in the database in the storage unit 11 through consultation with medical experts.

For example, when the biosignal is blood pressure, a systolic blood pressure of less than 120 mmHg and a diastolic blood pressure of less than 80 mmHg may be determined as normal ranges, and the first model converts each of these normal ranges into the user's first personal health time series data set (in step S440). It can be compared with the measured value (which may be the one before being updated by ), and the user's degree of blood pressure abnormality can be digitized and output as a score.

Alternatively, the normal range for each bio-signal may be subdivided into multiples according to individual physical factors such as gender, age, height, weight, etc., and/or external environmental factors such as season, temperature, and measurement time, instead of being determined singly. , The control unit 13 may identify a normal range corresponding to the user's individual physical element and/or external environment factor among a plurality of normal ranges for a specific bio-signal, and apply the identified normal range to the first model. Step S914 may be performed only once for the first time.

In step S914, the control unit 13 determines the user's first personal health time series data set (by step S440), as learned from a plurality of reference health time series data sets for the physiological signals previously acquired from a plurality of individuals other than the user. may be before updating) to generate a second model that outputs a second score related to a biosignal. The reference health time-series data set relates to the same biosignal as the first personal health time-series data set, and includes a change history of measurement values of another person's biosignal over time. Since the size (amount) of the first personal health time series data set is not sufficient at the beginning when the user signs up as a member of the intelligent healthcare device 3, a pre-learned second model is built with the data set collected from other users. . Multiple reference health time series data sets can be collected from public data set repositories such as data hubs or obtained through paid purchases.

Some biosignals can be judged as abnormalities by measuring values at the moment of measurement, such as blood pressure or body mass index (BMI). Since there are types, the controller 13 can select a model that matches the characteristics of each type of biosignal and perform learning on the selected model.

In step S916, the control unit 13 determines the first personal health time-series data set (which may be the one after being updated in step S440) as learned from the first personal health time-series data set (which may be the one before being updated by step S440). may be) to generate a third model that outputs a third score related to the biosignal. Generation of the third model in step S916 may be performed when the size (amount) of the first personal health time-series data set (which may be before being updated in step S440) is equal to or greater than the reference data amount. In addition, in the learning of the third model in step S916, in addition to the first personal health time series data set, a second personal health time series data set previously obtained according to the method of FIG. 7 may be used as learning data. Creation of an optimally personalized second model is possible.

In step S918, the controller 13 may combine the first model, the second model, and the third model so as to output a weighted average of the first score, the second score, and the third score as an ideal score. An ensemble technique may be applied to a combination of the first model, the second model, and the third model.

Specifically, the controller 13 may determine (adjust) the first weight, the second weight, and the third weight according to the number of measurement values recorded in the first personal health time-series data set. The first range, the second range, and the third range corresponding to the first weight, the second weight, and the third weight on a one-to-one basis may be adjusted if necessary by the manager of the intelligent healthcare device 3 in consideration of expert advice. .

An example of an artificial intelligence model generated through a combination of the first model, the second model, and the third model will now be described with reference to the following formula.

<Equation 1>

S(x) = M1(x)ХW1 + M2(x)ХW2 + M3(x)ХW3

In Equation 1 above, x is all of the first personal health time series data set (which may be updated by step S440), two or more measured values measured in a specific period (eg, the last month), or measured at a specific time is a single measured value. M1(x) is the first score output from the first model for x, M2(x) is the second score output from the second model for x, M3(x) is the output score from the third model for x The third score, W1 is the first weight, W2 is the second weight, W3 is the third weight, and S(x) is the ideal score output from the artificial intelligence model. In this case, each of the first score, the second score, and the third score may be in the range of 0 to 1, and the sum of the first weight W1, the second weight W2, and the third weight W3 may be 1. , and therefore S(x) may also be in the range of 0 to 1.

The first weight W1 may be adjusted in a first range (eg, 0.8 to 0.5), the second weight W2 may be adjusted in a second range (eg, 0.2 to 0.1), and the third weight W3 may be adjusted in a third range (eg, 0 to 0.4). The controller 13 first adjusts the first weight W1 within a first range whenever the number of measurement values recorded in the first personal health time series data set increases by the reference number (eg, 100). value, the second weight W2 may be decreased by a second adjustment value within a second range, and the third weight W3 may be increased by a third adjustment value within a third range. For example, when the number of measurement values recorded in the first personal health time series data set (which may be before updating by step S440) is less than 100, the first weight W1 is 0.8, the upper limit of the first range, The second weight W2 may be set to 0.2, the upper limit of the second range, and the third weight W3 may be set to 0, the lower limit of the third range. Thereafter, when the number of measurement values recorded in the first personal health time series data set (which may be before being updated by step S440) is 100 to 199, the first weight W1 is reduced by the first adjustment value to 0.8 , the second weight W2 may be set to 0.2 reduced by the second adjustment value, and the third weight W3 may be set to 0.01 increased by the third adjustment value (predetermined).

The first adjustment value and the second adjustment value may be determined by Equation 2 below whenever the third weight W3 increases by the third adjustment value.

<Equation 2>

First adjustment value = Third adjustment value Х Width of the first range / (width of the first range + width of the second range)

Second adjustment value = Third adjustment value Х Width of the second range/(width of the first range + width of the second range)

Alternatively, the first weight W1, the second weight W2, and the third weight W3 may be determined by Equation 3 below.

<Equation 3>

W1 = U1 - {W3 Х first range width / (first range width + second range width)}

W2 = U2 - {W3 Х Width of second range/(Width of first range + Width of second range)}

W3 = L3 + {third adjustment value Х D(N/M)}

In Equation 3, U1 is the upper limit of the first range, U2 is the upper limit of the second range, L3 is the lower limit of the third range, M is the reference number, N is the number of measured values recorded in the first personal health time series data set , D(N/M) is the largest integer not exceeding N/M. For example, when N = 911, M = 100, and the third adjustment value = 0.01, W3 = 0.09. Here, U1 + U2 + L3 = 1. Also, when L1 is the lower limit of the first range, L2 is the lower limit of the second range, and U3 is the upper limit of the third range, L1 + L2 + U3 = 1.

In step S920, the controller 13 calculates an abnormality score related to the user's health condition based on the first personal health time series data set (which may be updated in step S440) using the artificial intelligence model. do. The user's abnormal score for the physiological signal calculated by the artificial intelligence model may be mapped to the first personal health time series data set and recorded in the storage unit 11 .

The intelligent healthcare device 3 may transmit the abnormal score calculated through the method of FIG. 9 to the medical device 2 of a medical person who is registered as the user's primary care physician (maintenance hospital) or who is treating (reserving) the user as a patient. there is.

On the other hand, the artificial intelligence model has optimal performance when learning with training data (eg, multiple reference health time series data sets) in which positive data and negative data are distributed in equal proportions. However, in general, there is a case of class imbalance in which abnormal data is much smaller than normal data, and the anomaly detection performance of the trained model is inevitably lowered due to a low ratio of normal data. Thus, the inventor of the present invention confirmed that the performance of the artificial intelligence model can be improved through the method according to FIG. 10 .

10 is a flowchart exemplarily illustrating a method of generating the second model described above with reference to FIG. 9 .

Referring to FIG. 10 , in step S1010, the controller 13 generates a first candidate model through supervised learning using a plurality of reference health time-series data sets. As described above, the plurality of reference health time series data sets are previously obtained from individuals other than the user, and each measurement value included therein is classified as a normal measurement value (normal data) or an abnormal measurement value (abnormal data). The normal measurement value may be a measurement value indicating that the health state is normal, and the abnormal measurement value may be a measurement value indicating that the health state is abnormal.

In step S1020, the controller 13 processes the plurality of reference health time series data sets by executing at least one of down sampling of normal measurement values and up sampling of abnormal measurement values of the plurality of reference health time series data sets; , a second candidate model is generated through supervised learning using a plurality of processed reference health time series data sets. Step S1020 may be selectively executed only when the ratio of abnormal measurement values in the plurality of reference health time series data sets is greater than or equal to the first reference ratio (eg, 10%) and less than the second reference ratio (eg, 30%). For example, if the number of measurement values in the plurality of reference health time series data sets is greater than or equal to a predetermined number (eg, 10,000), downsampling of normal measurement values is performed, and if the number is less than the predetermined number, upsampling of abnormal measurement values is performed. this can be executed. As another example, the control unit 13 repeatedly performs down-sampling of normal measurement values and up-sampling of abnormal measurement values of the plurality of reference health time-series data sets at predetermined ratio intervals to obtain two or more processed plurality of reference health time-series data. After generating the set, a second candidate model trained with two or more processed reference health time series data sets may be generated. In the processed plurality of reference health time series data sets, the ratio of abnormal measurement values is greater than or equal to the second reference ratio.

In step S1030, the control unit 13 generates a third candidate model through unsupervised learning using only normal measurement values of a plurality of reference health time-series data sets as training data. In other words, when normal measurements occupy most of the multiple reference health time series data sets, performance is limited even when supervised learning is performed using upsampling and/or downsampling techniques, so unsupervised learning is used instead of supervised learning. Performance of the third candidate model may be improved by generating the third candidate model learned through the process. Step S1030 may be selectively executed only when the ratio of abnormal measurement values of the plurality of reference health time series data sets is less than the first reference ratio.

The third candidate model outputs a score corresponding to each measured value of a plurality of reference health time series data sets, classifies a measured value corresponding to a score below a threshold value as a normal measurement value, and measures a score corresponding to a score above the threshold value A value can be classified as an outlier measurement. The output data of the third candidate model is compared with training data (a plurality of reference health time series data), and learning may be performed several times so that the threshold value of the third candidate model is automatically adjusted according to the comparison result. For example, when the third candidate model outputs the specific measurement value as a normal measurement value even though the specific measurement value of the training data is an abnormal measurement value, the threshold value is automatically adjusted so that the specific measurement value is identified as an abnormal measurement value. can

In step S1040, the control unit 13 evaluates performance of each of the first candidate model, the second candidate model, and the third candidate model using the first test data set. Here, the first test data set may be a plurality of reference health time-series data sets, or may be a previously obtained data set separately from the plurality of reference health time-series data sets.

In step S1050, the controller 13 selects one of the first candidate model, the second candidate model, and the third candidate model evaluated as having the highest performance as the second model. This completes step S914.

FIG. 11 is a flowchart exemplarily illustrating a method of generating the third model described above with reference to FIG. 9 .

Referring to FIG. 11 , in step S1110, the controller 13 generates a fourth candidate model through supervised learning using the first personal health time-series data set (which may be the one before being updated in step S440). Each measurement value of the first personal health time series data set may be classified as an abnormal measurement value or a normal measurement value by the above-described second model or another model.

In step S1120, the controller 13 processes the first personal health time-series data set by performing at least one of down-sampling normal measurement values and up-sampling abnormal measurement values of the first personal health time-series data set; , a fifth candidate model is generated through supervised learning using the processed first personal health time series data set. Step S1120 may be selectively executed only when the ratio of abnormal measurement values in the first personal health time series data set is greater than or equal to the first reference ratio and less than the second reference ratio. For example, if the number of measurement values in the first personal health time series data set is greater than or equal to a predetermined number, downsampling of normal measurement values may be performed, and if the number of measurement values of the first personal health time series data set is less than the predetermined number, upsampling of abnormal measurement values may be performed. As another example, the controller 13 repeatedly performs down-sampling of normal measurement values and up-sampling of abnormal measurement values of the first personal health time-series data set at intervals of a predetermined ratio to obtain two or more processed first personal health time-series data After generating the set, a fifth candidate model trained with two or more processed first personal health time series data sets may be generated. In the processed first personal health time series data set, a ratio of abnormal measurement values is greater than or equal to a second reference ratio.

In step S1130, the controller 13 generates a sixth candidate model through unsupervised learning using only the normal measurement values of the first personal health time-series data set as training data. Step S1130 may be selectively executed only when the ratio of abnormal measurement values in the first personal health time series data set is less than the first reference ratio.

The sixth candidate model outputs a score corresponding to each measurement value of the first personal health time series data set, classifies a measurement value corresponding to a score below a threshold value as a normal measurement value, and measures a score corresponding to a score above the threshold value A value can be classified as an outlier measurement. The output data of the sixth candidate model is compared with the learning data (first personal health time series data set), and learning may be performed several times so that the threshold value of the sixth candidate model is automatically adjusted according to the comparison result.

In step S1140, the controller 13 evaluates performance of each of the fourth candidate model, the fifth candidate model, and the sixth candidate model by using the second test data set. Here, the second test data set may be a first personal health time-series data set or at least one of the plurality of reference health time-series data sets described above.

In step S1150, the control unit 13 selects one of the fourth candidate model, the fifth candidate model, and the sixth candidate model evaluated as having the highest performance as the third model. This completes step S916.

FIG. 12 is a diagram showing the overall configuration included in the apparatus 100 for recommending a visit by a primary care physician based on real-time monitoring of patient's personal health data according to the first embodiment of the present invention.

However, this is only a preferred embodiment for achieving the object of the present invention, and some components may be added or deleted as necessary, and other components may also perform the role played by one component.

According to a first embodiment of the present invention, an apparatus 100 for inviting a doctor to lead a doctor's visit according to real-time monitoring of personal health data of a patient includes a processor 10, a network interface 20, a memory 30, a storage 40, and connecting them. It may include a data bus 50 that can be implemented as an independent device by itself or implemented in the form of a physical service server such as an in-house system and a space rental system, or an intangible cloud service server. there will be The intelligent healthcare device 3 shown in FIG. 1 may be used as the device 100 for inviting the attending physician to visit.

The processor 10 controls the overall operation of each component. The processor 10 may be any one of a central processing unit (CPU), a micro processor unit (MPU), a micro controller unit (MCU), or a type of processor widely known in the art to which the present invention belongs, and a machine learning model processor. Alternatively, it can be implemented as an artificial intelligence model processor such as a deep learning model processor. In addition, the processor 10 may perform an operation for at least one application or program for executing the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention.

The network interface 20 supports wired/wireless Internet communication of the device 100 leading the doctor to visit a hospital according to real-time monitoring of personal health data of a patient according to the first embodiment of the present invention, and may support other known communication methods. . Accordingly, the network interface 20 may include a communication module according to it.

The memory 30 stores various types of information, commands and/or information, and one or more computer programs ( 41) can be loaded. Although RAM is shown as one of the memories 30 in FIG. 1 , it goes without saying that various storage media can be used as the memory 30 .

Storage 40 may non-temporarily store one or more computer programs 41 and mass network information 42 . The storage 40 may be a non-volatile memory such as a read only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, a hard disk, a removable disk, or a It may be any one of widely known computer-readable recording media in any form.

The computer program 41 is loaded into the memory 30 so that one or more processors 10 (A) receive the patient's personal health record (PHR) measurement value from the patient terminal based on the patient's PHR. An operation for calculating a PHR evaluation value including at least one of an abnormality score and a PHR abnormality grade in real time, (B) an operation for determining whether the patient's PHR evaluation value calculated in real time exceeds a predetermined threshold value, (C) An operation for sending a real-time notification to the patient's doctor's terminal if the result of the determination of the (B) operation is exceeded; an operation of displaying two candidate schedules on a patient calendar and outputting them to the patient terminal; (E) receiving selection information for one of the plurality of candidate schedules from the patient terminal to complete a visit reservation; and (F) an operation of displaying the completed hospital visit reservation on the doctor's calendar and outputting it to the doctor's terminal.

The operation performed by the computer program 41 briefly mentioned above can be regarded as one function of the computer program 41, and a more detailed description is to actively reserve a patient's visit schedule according to the second embodiment of the present invention. It will be described later in the description of the method.

The data bus 50 serves as a transfer path for commands and/or information between the processor 10 , the network interface 20 , the memory 30 and the storage 40 described above.

If the apparatus 100 for recommending a patient's personal health data real-time monitoring according to the first embodiment of the present invention as described above is implemented in the form of an independent device, for example, a server, it is connected to the personal health data A patient terminal (not shown) that transmits may be a terminal equipped with various sensors capable of measuring a patient's bio-signal, and any specific type of terminal may be used.

On the other hand, the doctor's terminal (not shown) does not need to measure the doctor's vital signals, so it does not have to be a terminal equipped with a sensor for this purpose. (not shown) will be a representative figure, but since the case where the attending physician becomes the patient's position may occur, it will not exclude the case of a terminal of the same type as the patient terminal (not shown).

On the other hand, both the patient terminal (not shown) and the doctor's terminal (not shown) access the device 100 for leading the doctor's visit according to real-time monitoring of the patient's personal health data according to the first embodiment of the present invention through a network, A method of actively reserving a patient's visit schedule may be provided, and for this purpose, dedicated applications may be installed in both the patient terminal (not shown) and the doctor's terminal (not shown). That is, the dedicated application itself provides a specific function provided by the device 100 for recommending a patient's personal health data according to real-time monitoring of the patient's personal health data according to the first embodiment of the present invention to a patient terminal (not shown) and a doctor's terminal ( It will be referred to as a configuration that serves as a medium for providing to (not shown) or receiving various information from them.

Hereinafter, a method for actively reserving a patient's visit schedule according to a second embodiment of the present invention will be described with reference to FIGS. 13 to 16 .

13 is a flowchart illustrating representative steps of a method of actively reserving a patient's visit schedule according to a second embodiment of the present invention.

However, this is only a preferred embodiment in achieving the object of the present invention, and some steps may be added or deleted as necessary, and any one step may be included in another step and performed.

In addition, it has been previously shown that each step is performed by the attending physician-initiated visit recommendation device (100, hereinafter referred to as "device") according to the patient's personal health data real-time monitoring according to the first embodiment of the present invention, and the implementation form is the server form. It is assumed that the patient terminal (not shown) has a network function and at the same time, a smart watch equipped with various sensors that can measure biosignals, and the doctor terminal (not shown) is a desktop PC. .

First, based on the patient's personal health record (PHR) measurement value received by the device 100 from the patient terminal, a PHR evaluation including at least one of the patient's PHR abnormality score and PHR abnormality grade is obtained in real time. Calculate (S1310).

Here, the patient's personal health data received from the patient terminal (not shown) may be different depending on the sensor mounted in the patient terminal (not shown), so it is possible to receive all of the various types of personal health data from as many sensors as possible. However, receiving only the personal health data necessary for analyzing the patient's condition according to the patient's settings on the dedicated application or the doctor's setting is a shortcut to accurately analyze the patient's condition, prevent problems with storage space, and prevent adverse effects of data waste. will be.

For example, if personal health data that can be measured by a patient terminal (not shown) includes blood pressure, body temperature, heart rate, blood sugar level, and electrocardiogram, and a patient belonging to a diabetes risk group, only blood pressure and blood sugar level can be set to be received as personal health data. There will be.

The patient's personal health data received from the patient terminal (not shown) is the personal health data measured by the patient terminal (not shown) in real time immediately before sending, and the personal health data of the patient according to the patient settings on the dedicated application or the doctor's settings. Data can be set to be measured at regular intervals, and furthermore, it can be set to be measured based on certain conditions.

More specifically, in the former case, for example, the blood glucose level is measured and sent at intervals of 1 hour, 6 hours, 1 day, etc., or in the latter case, the blood glucose level is measured only when the blood pressure is above a certain value, for example The latter can be used more usefully when personal health data that can be used for analyzing the condition of a patient with a certain disease or disease is personal health data that is correlated with each other.

The apparatus 100 calculates a PHR evaluation value including at least one of a PHR abnormal score and a PHR grade of a patient in real time based on a personal health data measurement value received from a patient terminal (not shown).

Here, the patient's PHR abnormality score and PHR grade are the same as the previous description, so detailed explanations are omitted to prevent redundant description. It will be possible to calculate only points, only PHR ratings, or both.

On the other hand, when calculating both the PHR abnormality score and the PHR grade in calculating the PHR evaluation value, the processor 10 included in the device 100 is implemented as a processor capable of parallel processing or implemented as a multi-processor to improve the calculation speed. Might do.

If the PHR evaluation value is calculated in real time, the device 100 determines whether the patient's PHR evaluation value calculated in real time exceeds a predetermined threshold value (S1320).

Here, the predetermined threshold value may be different depending on the type of personal health data received in the previous step S1310. When the PHR evaluation value includes only the PHR or higher score, the threshold value is the score and the PHR evaluation value includes only the PHR or higher score. In this case, the threshold value may be set as a grade, and in the case of including both of them, each threshold value may be set as a score and a grade.

On the other hand, the threshold value may be individually set according to the patient on a dedicated application even if the attending physician is the same type of PHR evaluation value. For example, patient A is a hypertensive patient and patient B is a diabetic patient. Even if the PHR evaluation value for blood pressure was calculated from these two, since the blood pressure value is more important in hypertensive patients than the blood pressure value in diabetic patients (in the case of diabetic patients, blood sugar level corresponds to the most important personal health data), hypertension It means that the blood pressure threshold for the patient can be set lower than the blood pressure threshold for the diabetic patient. It should be set as a standard value to determine the degree to which the level is reached.

It is determined whether the PHR evaluation value exceeds a predetermined threshold value, and if the PHR evaluation value exceeds a predetermined threshold value, the device 100 sends a real-time notification to the patient's personal physician terminal (not shown) (S1330).

The real-time notification here is a notification on a dedicated application, but apart from this, notification through an interactive messenger or notification through a text message can be any kind of notification. It corresponds to an unspoken request to determine whether the condition is necessary.

Accordingly, the real-time notification may include predetermined information capable of determining whether the patient needs to visit the hospital, such as the measured value of the patient's personal health data received from the patient terminal (not shown), and the PHR calculated in real time. It may include any one or more of an evaluation value, a measurement value of personal health data within a predetermined period of time based on the current point of time, or a graph of changes in PHR evaluation value, and the patient's personal information. The patient's personal information includes the patient's name, age, and height. , weight, gender, occupation, and whether or not you have any diseases or conditions.

Furthermore, when it is difficult to determine whether the patient's condition requires a hospital visit only with the information included in the real-time notification, the device 100 transmits the patient's medical examination data or other personal health data through the attending physician's terminal (not shown). A request for additional transmission of the calculated PHR evaluation value or a graph of changes within a predetermined time may be received and provided to the doctor's terminal (not shown).

On the other hand, it is determined whether the PHR evaluation value exceeds a preset threshold value, and if the result of the determination is not exceeded, the process may end without reserving a visit schedule. Through the device 100, a message that a visit is not required but that special attention or careful management is required can be sent to the patient terminal (not shown) (S1335). In this case, the patient pays more attention to his or her health condition. you will be able to tilt

If a real-time notification is sent to the doctor's terminal (not shown), the patient needs to visit the hospital, so the device 100 receives the visit request information from the doctor's terminal and sets up a plurality of candidate schedules including possible dates and times for the patient's visit. It is displayed on the patient's calendar and output to the patient's terminal (not shown) (S1340).

This step S1340 can be seen as a step of initiating a process related to a visit reservation between a patient who needs to visit and a doctor, and will be described in detail with reference to FIG. 14 below.

14 is a flowchart embodying step S1340 in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention.

However, this is only a preferred embodiment in achieving the object of the present invention, and some steps may be added or deleted as necessary, and any one step may be included in another step and performed.

First of all, the device 100 receives visit request information including a plurality of candidate schedules for the chief physician selection including possible dates and times for visits of patients selected by the chief physician from the terminal (not shown) (S1340-1).

Since a visit is a visit by a patient to a hospital, it is common to follow the doctor's treatment schedule. will be able to do it

The primary care physician terminal (not shown) sends visit request information including a plurality of primary care physician selection candidate schedules including possible dates and times for the patient's visit to the device 100, and the visit request information outputs the primary care physician's calendar on a dedicated application. and can be created through selection on it.

More specifically, the attending physician's calendar automatically displays a schedule for which visits have been made through the device 100, and a reservation schedule according to a call received through the hospital rather than the device 100 is also displayed through the attending physician's terminal (not shown). It can be indicated by handwriting. The attending physician will be able to easily select the date and time available for the patient's visit by referring to the attending physician's calendar printed on the attending physician's terminal (not shown). Rather than selecting a plurality of times within a day, selecting a plurality of times from several days may be a method for giving the patient a degree of freedom in selecting a visit date and time.

On the other hand, the device 100 automatically searches for dates and times that can be visited on the attending physician's calendar and displays them separately on the attending physician's calendar so that the attending physician can easily select a date and time that can be visited by the attending physician. It may contribute to improving the convenience of

Thereafter, the device 100 selects a plurality of device selection candidate schedules including a plurality of device selection candidate schedules having different dates and times from the plurality of primary care physician selection candidate schedules included in the received visit request information (S1340-2).

In step S1340-1, if the attending physician has selected a possible date and time as the attending physician's selection candidate schedule by referring to the attending physician's calendar, in step S1340-2, the device 100 selects a different date and time by itself.

When the device 100 selects the device selection candidate schedule, the date and time that can be visited on the attending physician's calendar are automatically searched in step S1340-1, and among the dates and times separately displayed on the attending physician's calendar, those who have not been selected by the attending physician Not only can the date and time be selected, but also information about the patient's occupation among personal information of the patient can be used.

For example, if the patient's occupation is an office worker, he or she may choose a time between 12:00 and 1:00 p.m. on weekdays, which corresponds to a typical office worker's lunch break. You will be able to select a time between 10am and 12pm.

On the other hand, it would be desirable to select a plurality of available dates and times for visitation as well as the candidate schedule for selecting the device, similar to the candidate schedule for selecting the attending physician.

Finally, the device 100 displays a plurality of candidate schedules including a plurality of candidate schedules selected by the attending physician and candidate schedules selected by the device on the patient's calendar and outputs them to the patient terminal (S1340-3).

When displaying and outputting a plurality of candidate schedules on the patient terminal, the apparatus 100 may distinguish between candidate schedules selected by the attending physician and candidate schedules selected by the device, and display the date and time preferred by the attending physician and other dates and times. It is to be distinguished from, and it will be said that the same can be displayed simply as a date and time that can be visited without distinction.

Let's go back to the description of FIG. 13 again.

If a plurality of candidate schedules are displayed on the patient's calendar and output to the patient's terminal (not shown), the device 100 receives selection information for any one of the plurality of candidate schedules from the patient's terminal (not shown) and makes a visit reservation. is completed, displayed on the patient's calendar and output to the patient terminal (not shown) (S1350), and also displayed on the doctor's calendar and output to the doctor's terminal (not shown) (S1360).

If selection information for any one of a plurality of candidate schedules is received on the patient calendar, the visit reservation can be regarded as complete, and the device 100 updates the patient calendar and the attending physician's calendar to display the date and time at which the selection information was received to the patient. It can be output to a terminal (not shown) and a doctor's terminal (not shown).

On the other hand, in relation to the reservation completion, only when the primary care physician's reservation completion approval information is received on the dedicated application through the primary care physician's terminal (not shown), the date and time selected by the patient may be marked as reservation completed, and a plurality of candidate schedules may be displayed. is displayed on the patient calendar, if a personal schedule overlaps with a plurality of candidate schedules exists by searching or loading the patient's personal schedule, the patient's calendar is output by processing so that the patient cannot select the overlapping candidate schedule. You will be able to.

Steps S1350 and S1360 described above relate to a case in which the patient directly selects any one schedule among a plurality of candidate schedules first presented to the patient by the attending physician or the device 100, and other personal schedules that cannot be searched or loaded. Therefore, it may be impossible to select any of a plurality of candidate schedules. In this case, it is a problem how to proceed with a visit reservation.

In this case, the device 100 receives, from the patient terminal (not shown), visit request information including a plurality of patient selection candidate schedules including possible dates and times for the patient to visit the hospital selected by the patient himself, and the device 100 itself A plurality of device selection candidate schedules including a plurality of patient selection candidate schedules and a plurality of different dates and times may be selected, displayed on the doctor's calendar together with the patient selection candidate schedules, and output to the doctor's terminal (not shown). The attending physician can check the attending physician's calendar printed on the attending physician's terminal (not shown) and select a date and time. It can be treated as completed.

Meanwhile, when the device 100 outputs the patient selection candidate schedule selected by the patient and the device selection candidate schedule selected by the device 100 on the attending physician's calendar, the primary care physician's reservation completion schedule is searched for or loaded, and the patient selection candidate schedule selected by the patient is displayed. If there is a reservation completion schedule that overlaps with the device selection candidate schedule selected by the device 100, the attending physician may output the attending physician's calendar by processing so that the attending physician cannot select the overlapping candidate schedule.

That is, in the above case, it can be seen that the description of the patient or the patient terminal (not shown) and the description of the doctor or the doctor's terminal (not shown) are interchanged with each other in steps S1340 to S1360 described above, and after the visit reservation is completed, one side It can be seen as an explanation that can be applied even when a change of reservation is required due to a request from

According to steps S1310 to S1360 described above, the patient's visit schedule can be regarded as being reserved, and output on a patient calendar output on a dedicated application of a patient terminal (not shown) or output on a dedicated application of a doctor's terminal (not shown) below. The calendar, which is an important component for schedule management in the present invention, including the attending physician's calendar, will be described.

When a visit reservation is completed, the date and time when the reservation is completed are displayed on both the patient calendar and the attending physician's calendar. Due to space reasons, it is difficult to display a lot of text on the calendar itself, so only the date and time are displayed, and when the corresponding time is selected Detailed information on the corresponding visit reservation may be output in a pop-up format or the like.

In addition to basic information such as the patient's name, primary care physician's name, expected visit time, disease information, and patient's personal information, detailed information on the visit reservation here includes information such as notes entered by the primary care physician, additional medical questionnaires, or inquiries entered by the patient. can include

Meanwhile, the date and time at which personal health data should be measured may be automatically displayed by the device 100 in the patient calendar, in addition to information on appointment reservations, and the corresponding date and time may be based on the measurement cycle described above. Furthermore, it would be a matter of debate that the patient can mark the measurement date and time on the patient's calendar by himself, and in the case of the patient's personal schedule, it may not be displayed on the patient's calendar in the dedicated application, so linking with external calendar services such as Google Calendar and Outlook Calendar It will also be possible to synchronize the schedules displayed on each calendar through.

Apart from this, the patient will be able to search the calendar of the attending physician or the calendar of another doctor other than the attending physician through a dedicated application, and the date and time available for visit reservation will be displayed on the calendar. However, detailed information related to the information of other reservations regarding the date and time when the reservation was completed will not be confirmed. This is because it is related to personal information.

The doctor's calendar can be checked by only one doctor, who is the owner of a doctor's terminal (not shown) with a dedicated application installed. In addition, when multiple doctors work in one hospital, the calendar for the hospital itself is treated as the doctor's calendar Several doctors may share it through their terminals. In this case, visit reservation information displayed on the attending physician's calendar may be recorded on each doctor's personal external calendar through synchronization, but only visit reservation information displayed on the attending physician's own external calendar may be retrieved when the calendar is retrieved.

On the other hand, contrary to the case of the patient, the attending physician cannot inquire the patient's calendar even through a dedicated application, because the patient's personal information irrelevant to the visit reservation may be included.

In the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention, after step S1360, the device 100 sends a PHR to a doctor's terminal (not shown) among patients who have a patient-personal relationship with the doctor. It may further include a step of sending a list of patients who have a history of exceeding a predetermined threshold value within 24 hours a day (S1370), and from the point of view of the attending physician, visits to patients who need to visit at the end of the day's treatment It is possible to contribute to more faithful and perfect management of the health conditions of the patients in charge by allowing the user to check once again whether the patient has recommended the patient or whether the visit reservation has been completed.

Furthermore, in the method for actively reserving a patient's visit schedule according to the second embodiment of the present invention, after step S1360, the device 100 sends a visit request notification according to the visit reservation to the patient terminal (not shown). It may further include a step (S1380), and the visit request notification may set the time to send the notification by setting on the dedicated application, such as one day before the visit reservation, the morning time on the day of the visit, and one hour before the visit reservation, and the visit request notification Also, any type of notification through a push notification, an interactive messenger, or a text message on a dedicated application will be acceptable.

In addition, the device 100 sends a visit request notification together with information such as precautions before visiting, for example, that fasting is required 6 hours before the visit or that taking drugs such as aspirin is prohibited, Pre-visit notes may be input by the attending physician through a doctor's terminal (not shown) or automatically input by the device 100 based on visit reservation information or through a pre-established database of notes on visit.

On the other hand, in the case of the attending physician, since the possibility of the patient's visit not going to work on the scheduled date is slim, a notification is not sent to the patient terminal (not shown) as if a notification of a visit request is sent. By outputting a list of patients scheduled to visit that day on the dedicated application on the city), it will contribute to confirming the patient to be treated on the same day once more.

On the other hand, in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention, after step S1360 as shown in FIG. ) may further include a step (S1390) of sending a measurement request notification of personal health data that is the cause of the visit (S1390), wherein the personal health data measurement request notification includes a questionnaire and It is possible to include any one or more of the questionnaires regarding the items that the attending physician will ask the patient in the course of medical treatment, so that the treatment time can be shortened when the patient actually visits the hospital.

If the device 100 requests measurement of personal health data that causes the visit one day before the scheduled visit date, measurement at a time too far from the scheduled visit date is maximal because the condition of the patient may be significantly different from the day of visit. In order to accurately grasp the patient's condition by measuring it at a time close to the appointment date of the visit, it is of course possible to request a different measurement time before the visit according to the setting of the attending physician or the type of personal health data that causes the visit.

This time, additional functions provided by the device 100 for patient convenience will be described.

The device 100 can identify external environmental information such as temperature or today's weather in accordance with the morning wake-up time of the patient on the day of the visit reservation, and send a notification to the patient terminal (not shown) through a dedicated application along with the visit reservation information. there is. For example, it may be precautions such as bring an umbrella if it rains, use public transportation if it snows, leave earlier than usual because it snows, etc. etc. will be available.

On the other hand, the device 100 can calculate the travel route to the hospital and the estimated required time based on the current location of the patient terminal (not shown) through linkage with an external map service on the day of the appointment reservation, and the estimated required time is available to the public. It can be calculated by dividing it into the case of using transportation and the case of using one's own car, and if the travel route to the hospital is short, the estimated time required on foot can also be calculated and output through a dedicated application.

Such a movement route and estimated required time can be continuously recalculated at a predetermined time interval, for example, 30 minutes or 1 hour, and outputted through a dedicated application. By making it possible to send a notification to prepare for a visit through a dedicated application when there are more minutes left, it can contribute to helping the patient arrive at the appointment time without delay.

In addition, the device 100 calculates the remaining distance from the hospital using the GPS information of the patient terminal (not shown) 10 minutes before the scheduled visit time through a dedicated application, and when it is expected that the device will not arrive within 10 minutes, the doctor's terminal (not shown), a notification including an estimated time of arrival may be sent, and the attending physician who confirms this will be able to use the time efficiently by seeing other patients at the scheduled time of visit.

Meanwhile, in relation to the convenience function, in the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention, after step S1360 as shown in FIG. Receiving the time the patient arrives at the hospital for a visit and the time the patient leaves the hospital after receiving treatment from the hospital, calculating the treatment time, which is the time the patient stays in the hospital (S1396), Completing the step of calculating for all patients who visited the hospital (S1398) and learning the calculated treatment time and corresponding visit reservation information to predict the treatment time for each day and time zone for the hospital and output it to the patient terminal (not shown) A step (S1399) may be further included, and by providing the patient with treatment time information for each day and time of the corresponding hospital, it is possible to contribute to the selection of a hospital when reserving a visit in the future.

Here, the GPS information of the patient terminal (not shown) can be used for the time the patient arrives at the hospital for a visit and the time the patient leaves the hospital after receiving medical treatment. In addition, it is possible to calculate more accurate treatment time by allowing the patient to manually select the time of arrival and departure of the hospital through a dedicated application, and by assigning a process such as a kind of attendance check to the patient.

On the other hand, since the treatment time is calculated for all patients who have completed the visit reservation through the dedicated application and visited the hospital, the learning data for calculating the treatment time may be small in the initial service of the dedicated application. In this case, the specific day of the week Patient convenience may be improved by calculating and outputting the average treatment time for each time zone.

So far, a method for actively reserving a patient's visit schedule according to the second embodiment of the present invention has been described. According to the present invention, it is possible to easily complete a reservation for a visit based on a schedule displayed on a patient calendar and a doctor's calendar on a dedicated application without a complicated and cumbersome phone reservation process. In addition, if the attending physician monitors the patient's personal health data measurement value and determines that a visit is necessary, a visit is first recommended and a visit reservation is made. You can avoid the situation of delaying your visit. In addition, since the hospital's treatment time for each day and time is predicted and provided to the patient, the patient can not only reserve a visit schedule optimized for his/her schedule, but also help select the hospital he/she wants to visit. . Lastly, by having the patient measure personal health data and fill out a medical questionnaire the day before the visit, the treatment time can be shortened, and even traffic information is provided so that the patient is not late for the appointment time on the day of the visit. It can provide non-storm service that maximizes patient convenience.

The method for actively reserving a patient's visit schedule according to the second embodiment of the present invention described above can be implemented with a computer program stored in a medium according to the third embodiment of the present invention that includes all the same technical features. Although it will not be described in detail to prevent description, all the technical features applied to the method for actively reserving a patient's visit schedule according to the second embodiment of the present invention described above are stored in the medium according to the third embodiment of the present invention. Of course, the same can be applied to computer programs.

This time, the hospital recommendation device 100 according to the real-time monitoring of patient's personal health data according to the fourth embodiment of the present invention will be described. Unlike the device 100 for recommending a patient's personal health data real-time monitoring of the patient according to the first embodiment or the computer program stored in the media according to the third embodiment of the present invention, in a state in which the relationship between the patient and the attending physician is not established To this end, it relates to an apparatus and method for recommending a hospital.

To this end, reference is made to FIG. 12 again. FIG. 12 is a schematic diagram of the configuration as well as the overall configuration included in the apparatus 100 for inviting the attending physician to lead a patient's personal health data real-time monitoring according to the first embodiment of the present invention. Therefore, since it corresponds to the drawing showing the entire configuration of the hospital recommendation device 100 according to the real-time monitoring of the patient's personal health data according to the fourth embodiment of the present invention, refer to FIG. 12 again without attaching the same drawing again. so that the explanation continues.

However, this is only a preferred embodiment for achieving the object of the present invention, and some components may be added or deleted as necessary, and other components may also perform the role played by one component.

Hospital recommendation device 100 according to real-time monitoring of patient's personal health data according to a fourth embodiment of the present invention includes a processor 10, a network interface 20, a memory 30, a storage 40, and data connecting them. The bus 50 may be included, and may be implemented as an independent device by itself, or may be implemented in the form of a tangible physical service server such as an in-house system and a space rental system, or an intangible cloud service server. will be.

The processor 10 controls the overall operation of each component. The processor 10 may be any one of a central processing unit (CPU), a micro processor unit (MPU), a micro controller unit (MCU), or a type of processor widely known in the art to which the present invention belongs, and a machine learning model processor. Alternatively, it can be implemented as an artificial intelligence model processor such as a deep learning model processor. In addition, the processor 10 may perform an operation for at least one application or program for performing the method of recommending a hospital to establish a relationship between a patient and a doctor according to the fifth embodiment of the present invention.

The network interface 20 supports wired/wireless Internet communication of the hospital recommendation device 100 according to the real-time monitoring of patient's personal health data according to the fourth embodiment of the present invention, and may support other known communication methods. Accordingly, the network interface 20 may include a communication module according to it.

The memory 30 stores various types of information, commands, and/or information, and one or more computer programs from the storage 40 to perform a method of recommending a hospital to establish a relationship between a patient and a doctor according to the second embodiment of the present invention. (41) can be loaded. In FIG. 12 , RAM is shown as one of the memories 30 , but various storage media can be used as the memory 30 , of course.

Storage 40 may non-temporarily store one or more computer programs 41 and mass network information 42 . The storage 40 may be a non-volatile memory such as a read only memory (ROM), an erasable programmable ROM (EPROM), an electrically erasable programmable ROM (EEPROM), a flash memory, a hard disk, a removable disk, or a It may be any one of widely known computer-readable recording media in any form.

The computer program 41 is loaded into the memory 30 so that one or more processors 10 (A) receive the patient's personal health record (PHR) measurement value from the patient terminal based on the patient's PHR. An operation for calculating a PHR evaluation value including at least one of an abnormality score and a PHR abnormality grade in real time, (B) an operation for determining whether the patient's PHR evaluation value calculated in real time exceeds a predetermined threshold value, (C) As a result of the determination of the operation (B), if the location information of the patient terminal is exceeded, an operation of calculating and outputting a plurality of recommended hospitals capable of treating items related to the PHR evaluation value from the hospital information database using the location information of the patient terminal, (D) the above An operation of completing a visit reservation by receiving selection information on any one of the plurality of recommended hospitals output from the patient terminal, and (E) selection information on any one of the plurality of recommended hospitals output from the patient terminal. It is possible to execute an operation of receiving and registering the patient with the hospital of the attending physician that has established a relationship between the patient and the attending physician.

The operation performed by the computer program 41 briefly mentioned above can be regarded as one function of the computer program 41, and a more detailed description is to recommend a hospital to establish a relationship between a patient and a doctor according to the fifth embodiment of the present invention. It will be described later in the description of how to do it.

The data bus 50 serves as a transfer path for commands and/or information between the processor 10 , the network interface 20 , the memory 30 and the storage 40 described above.

If the hospital recommendation device 100 according to the fourth embodiment of the present invention described above is implemented in the form of an independent device, for example, a server, according to the real-time monitoring of the patient's personal health data, it is connected and the personal health data is transmitted. A patient terminal (not shown) may be a terminal equipped with various sensors capable of measuring a patient's bio-signal, and any specific type of terminal may be used.

On the other hand, the hospital terminal (not shown) does not need to measure the doctor's vital signals, so it does not have to be a terminal equipped with a sensor for this purpose. (not shown) will be a representative figure, but since it can also happen that a doctor is in the position of a patient, the case of a terminal of the same type as a patient terminal (not shown) will not be excluded.

On the other hand, both the patient terminal (not shown) and the hospital terminal (not shown) access the hospital recommendation device 100 according to the real-time monitoring of the patient's personal health data according to the fourth embodiment of the present invention through a network, A method of actively reserving a hospital visit schedule may be provided, and for this purpose, dedicated applications may be installed in both the patient terminal (not shown) and the attending physician terminal (not shown). That is, the dedicated application itself provides a specific function provided by the hospital recommendation device 100 according to real-time monitoring of the patient's personal health data according to the fourth embodiment of the present invention to a patient terminal (not shown) and a hospital terminal (not shown). ) or a configuration that serves as a medium for receiving various information from them.

To put it a little easier, the dedicated application here is the device 100 for recommending visits led by the attending physician according to real-time monitoring of the patient's personal health data according to the first embodiment of the present invention described above or the media according to the third embodiment of the present invention. It can be seen as the same application as the dedicated application mentioned in the computer program stored in, and the patient terminal (not shown) is the same, and in the case of a hospital terminal (not shown), the present embodiment is viewed in the same configuration as the doctor terminal (not shown). It is believed that there will be no problem in understanding.

Hereinafter, a method of recommending a hospital to establish a relationship between a patient and a doctor according to a fifth embodiment of the present invention will be described with reference to FIGS. 17 to XX.

FIG. 17 is a flowchart illustrating representative steps of a method of recommending a hospital for a patient-supervising doctor relationship according to a fifth embodiment of the present invention.

However, this is only a preferred embodiment in achieving the object of the present invention, and some steps may be added or deleted as necessary, and any one step may be included in another step and performed.

In addition, each step is performed on the premise that the hospital recommendation device (100, hereinafter referred to as “device”) according to the real-time monitoring of the patient's personal health data according to the fourth embodiment of the present invention is performed in the form of a server. In the case of a patient terminal (not shown), the explanation will be continued on the premise that it is a smart watch equipped with various sensors capable of measuring biosignals while having a network function, and a desktop PC in the case of a hospital terminal (not shown). The intelligent healthcare device 3 shown in FIG. 1 may be used as the hospital recommendation device 100 .

First, based on the patient's personal health record (PHR) measurement value received by the device 100 from the patient terminal, a PHR evaluation including at least one of the patient's PHR abnormality score and PHR abnormality grade is obtained in real time. Calculate (S1710).

Here, the patient's personal health data received from the patient terminal (not shown) may be different depending on the sensor mounted in the patient terminal (not shown), so it is possible to receive all of the various types of personal health data from as many sensors as possible. However, receiving only the personal health data necessary for analyzing the patient's condition according to the patient's settings on the dedicated application or the doctor's setting is a shortcut to accurately analyze the patient's condition, prevent problems with storage space, and prevent adverse effects of data waste. will be.

For example, if personal health data that can be measured by a patient terminal (not shown) includes blood pressure, body temperature, heart rate, blood sugar level, and electrocardiogram, and a patient belonging to a diabetes risk group, only blood pressure and blood sugar level can be set to be received as personal health data. There will be.

The patient's personal health data received from the patient terminal (not shown) is the personal health data measured by the patient terminal (not shown) in real time immediately before sending, and the patient's personal health data is measured according to the patient settings on the dedicated application or the doctor's settings to be described later. Personal health data can be set to be measured at regular intervals, and furthermore, it can be set to be measured based on certain conditions.

More specifically, in the former case, for example, the blood glucose level is measured and sent at intervals of 1 hour, 6 hours, 1 day, etc., or in the latter case, the blood glucose level is measured only when the blood pressure is above a certain value, for example The latter can be used more usefully when personal health data that can be used for analyzing the condition of a patient with a certain disease or disease is personal health data that is correlated with each other.

The apparatus 100 calculates a PHR evaluation value including at least one of a PHR abnormal score and a PHR grade of a patient in real time based on a personal health data measurement value received from a patient terminal (not shown).

Here, the patient's PHR abnormality score and PHR grade are the same as the previous description, so detailed explanations are omitted to prevent redundant description. It will be possible to calculate only points, only PHR ratings, or both.

On the other hand, when calculating both the PHR abnormality score and the PHR grade in calculating the PHR evaluation value, the processor 10 included in the device 100 is implemented as a processor capable of parallel processing or implemented as a multi-processor to improve the calculation speed. Might do.

If the PHR evaluation value is calculated in real time, the apparatus 100 determines whether the patient's PHR evaluation value calculated in real time exceeds a predetermined threshold value (S1720).

Here, the predetermined threshold value may be different depending on the type of personal health data received in the previous step S1710. When the PHR evaluation value includes only the PHR or higher score, the threshold value is the score and the PHR evaluation value includes only the PHR or higher score. In this case, the threshold value may be set as a grade, and in the case of including both of them, each threshold value may be set as a score and a grade.

On the other hand, the threshold value may be individually set according to the patient on a dedicated application even if the attending physician is the same type of PHR evaluation value. For example, patient A is a hypertensive patient and patient B is a diabetic patient. Even if the PHR evaluation value for blood pressure was calculated from these two, since the blood pressure value is more important in hypertensive patients than the blood pressure value in diabetic patients (in the case of diabetic patients, blood sugar level corresponds to the most important personal health data), hypertension It means that the blood pressure threshold for the patient can be set lower than the blood pressure threshold for the diabetic patient. It should be set as a standard value to determine the degree to which the level is reached.

Steps S1710 and S1720 described above are the same as the description of steps S1310 and S1320 described above, and from the description of step S1730 below, the hospital recommendation process due to the state in which the relationship between the patient and the attending physician is not established is started in earnest. would be said to correspond to

It is determined whether the PHR evaluation value exceeds a preset threshold value, and if the determination result is exceeded, the device 100 provides recommended hospitals that can treat items related to the PHR evaluation value by using the location information of the patient terminal (not shown). Hospital information It is calculated in the database and output in plurality (S1730).

Here, the recommended hospital that can treat items related to the PHR evaluation value refers to the personal health data measurement value that is the basis for calculating the PHR evaluation value and the hospital that provides treatment for diseases or diseases related to it. When the health data measurement value is the blood glucose level, internal medicine that can provide treatment related to hypertension can be output as a recommended hospital.

On the other hand, in calculating and outputting a plurality of recommended hospitals, among a plurality of hospitals located within a predetermined distance from the patient terminal (not shown), member hospitals that have completed membership in a dedicated application can be output in preference to non-member hospitals Bar, this is to induce hospitals to sign up for membership, and in some cases, such as in the United States, if the hospital to which the medical insurance is applicable is different depending on the type of medical insurance held by the individual patient, the recommended hospital may be output in consideration of this In this case, member hospitals that have completed membership registration in the dedicated application will be able to be output in priority over non-member hospitals.

Here, member hospitals that have completed membership registration in the dedicated application have nothing to do with whether or not they paid to become a member, but if they did, member hospitals that did not pay fees and non-member hospitals that did not pay fees It can be said that outputting with more priority than in the case of is corresponding to the operation method suitable for the purpose of the service.

Apart from the description of step S1730 being included because the method for recommending a hospital to establish a relationship between a patient and a doctor according to the fifth embodiment of the present invention is that the device 100 recommends a hospital, the patient terminal (not shown) Of course, it is possible to search for a hospital optimized for the patient by running an installed dedicated application and accessing the occupational device 100, which is also matched with the search sentence input by the patient and at a predetermined distance from the patient terminal (not shown). Among a plurality of hospitals located within the hospital, the member hospitals that have completed membership registration in the dedicated application will be output prior to non-member hospitals.

To this end, the processor 10 of the device 100 may include a deep learning natural language processing model that analyzes the search sentence input by the patient, for example, "score of 70 or higher", "specialized in hypertension", " If you enter something like "hospital with short waiting time" or "hospital with many reviews", the deep learning model analyzes this to create search conditions and then performs a search using those conditions, rather than the patient entering a search sentence. It will work in the same way even if you enter the word you want to search for.

Meanwhile, the hospital search is performed in the hospital information database (not shown), and returns to the description of step S1730 to describe the hospital information database (not shown).

The hospital information database (not shown) may include both information about member hospitals (including both paid and free) that have completed membership registration in a dedicated application and information about non-member hospitals, and information about member hospitals and members Hospital information other than hospital name, address, latitude/longitude of hospital, hospital email address, hospital website, hospital specialty, hospital specialty, information on doctors working in hospital, hospital hours, It may include any one or more of information about the hospital's treatment hours by day and time slot, the type of medical insurance applied, hospital reviews, hospital ratings, and hospital hashtags.

Information on member hospitals and information on non-member hospitals may be collected through periodic crawling, or, of course, may be directly input by patients corresponding to users.

Among information on member hospitals and information on non-member hospitals, hospital name, address, latitude/longitude of hospital, hospital email address, hospital website, hospital department, hospital specialty, and doctors working at the hospital As it is common for hospitals to disclose information on hospital visits, treatment hours of hospitals, treatment hours by day and time of day of the hospital, and types of applicable medical insurance on the portal site, they can be collected relatively easily through crawling. , hospital ratings and information about hospital hashtags need to be treated a little differently.

More specifically, in the case of hospital reviews, patients corresponding to the user can directly enter or collect them through crawling, and in the case of hospital ratings, reviews directly entered by patients corresponding to the user or not given ratings can be collected through deep learning. Using a natural language processing model, the device 100 may analyze reviews on its own and automatically assign ratings. In the case of a hashtag for a hospital, the device 100 can automatically extract and generate it from hospital reviews and hospital information using a keyword extraction deep learning natural language processing model, and when a patient searches for a hospital or filters on whether conditions are matched, It can be used and automatically updated according to collected reviews.

On the other hand, the device 100 may periodically update a hospital information database (not shown) by bringing authorized hospital information through association with an external public hospital information system, for example, the Health Insurance Review and Assessment Service in Korea, In countries where linkage is not possible, hospital information can be automatically collected through crawling, or information can be manually collected by administrators. In general, since the hospital information database (not shown) will often contain only basic information about the hospital, such as hospital name, address, phone number, and treatment course, additional information such as hospital reviews are collected through crawling technology. It will be updated, and the information on the latitude/longitude of the hospital can be found by geocoding using the hospital address information.

Crawling, which is a method related to the construction of the hospital information database (not shown) described above, is used to build a hospital information database (not shown) when it is impossible to link with the public hospital information system in addition to collecting and updating hospital reviews or other additional information. It is possible, and accordingly, it is called smart crawling.

If it is impossible to establish a hospital information database (not shown) because it is impossible to link with a public hospital information system, information on hospitals in the country can be collected through smart crawling, more specifically, "hospital" After collecting only hospital-related data using hospital-related keywords such as , "clinic" and "medical center", it is analyzed through a deep learning natural language processing model to confirm whether the hospital information is correct. If the deep learning model determines that it is not hospital information, the data is discarded. If it is determined that the hospital information is correct, it extracts basic information such as hospital name, hospital address, phone number, email, etc. A hospital information database (not shown) is searched using and stored in the hospital information database as a new hospital if it is not an existing hospital in the database.

Such smart crawling is also used to update the latest information about the hospital. In the case of a hospital that has already been stored in the hospital information database (not shown), the hospital website is periodically crawled using the homepage URL recorded in the database. Basic information such as name, e-mail, phone number, address, and medical department is automatically updated, and if the hospital address is changed, the latitude/longitude is also updated to match the new address.

If there is no response from the homepage for a certain period of time due to a change in the homepage address during crawling, the device 100 sends a homepage address change confirmation request mail to the e-mail address recorded in the hospital information database (not shown) to update the changed address. If the corresponding e-mail address is also changed (including no e-mail address) and the e-mail cannot be transmitted, the information is changed to Unconfirmed, and the manager of the device 100 manually updates it later.

Furthermore, smart crawling is also used to collect reviews about hospitals. In the case of reviews, reviews on the hospital homepage are less reliable, so after crawling from sites other than the hospital's official homepage, such as map services, other blogs, and communities, , Reviews can be analyzed with a deep learning natural language processing model (sentiment analysis model) to give ratings.

Here, the rating is calculated and given on a scale of 1 to 100, and the higher the positive degree, the closer to 100 points are assigned. It will be possible to calculate the average of the ratings automatically assigned by analysis. In addition, in order to prevent duplicate reviews, a separate deep learning natural language processing model is used to calculate the similarity of previously collected reviews and newly collected reviews, and if the newly collected reviews are close to 1 in similarity with the existing reviews, Newly collected reviews may be discarded without being reflected.

On the other hand, among hospitals that have not signed up for membership in the dedicated application through smart crawling, in the case of a newly opened hospital, a membership registration guide mail may be sent to an email address containing information crawled by the device 100, in this case Membership registration can be induced by sending together information such as benefits and promotions when registering as a member.

The hospital information database (not shown) described above may be a configuration included in the device 100, but it is preferable to implement it in an external server as much as possible because a problem in storage space may occur due to a large amount of data. Not limiting, but will.

Let's go back to the description of FIG. 17 again.

If a plurality of recommended hospitals are calculated from a hospital information database (not shown) and outputted, the device 100 receives selection information for any one of the plurality of recommended hospitals outputted from a patient terminal (not shown) to make a visit reservation. It is completed (S1740).

In order to prevent redundant description, it is described that the visit reservation is completed in step S1740, but this is the same as the description of steps S1340 to S1360 in the method for actively reserving a patient's visit schedule according to the second embodiment of the present invention. can see. That is, the process of sending the visit schedule and sending selection information between the patient terminal (not shown) and the hospital terminal (not shown) may be repeated, and accordingly, the finally reserved visit schedule is displayed in both the patient calendar and the hospital calendar. can be displayed and displayed.

On the other hand, in step S1740, although the visit reservation has been completed, the hospital will not be able to receive the patient's personal health data without the patient's consent because it is still in a state before the relationship between the patient and the attending physician is established. Accordingly, after step S1740, the device Consent information for sending any one or more of the Chinese character personal health data measurement value and the calculated PHR evaluation value received from the patient terminal (not shown) by (100) to the hospital terminal (not shown) of the hospital that completed the visit reservation. A step of receiving (S1745) may be performed from the patient terminal (not shown), and if the consent information is not received, only the information input by the patient on the dedicated application through the patient terminal (not shown) is the hospital terminal (not shown). ) will be able to receive.

Thereafter, the device 100 receives selection information for any one of a plurality of recommended hospitals output from the patient terminal (not shown) and registers the patient as a primary care physician hospital that has a relationship between the patient and the primary care physician (S1750), Selection information about one or more doctors working in the hospital registered as the primary care physician may be received from the patient terminal (not shown) and registered as the primary care physician with whom the patient has a relationship with the primary care physician (S1760).

Steps S1750 and S1760 can be regarded as steps for establishing a relationship between a patient and a doctor, and more specifically, step S1750 can be regarded as a step for registering a relationship between a doctor's hospital or a regular hospital, and step S1760 can be regarded as a step for registering a doctor relationship. Including everyone, it can be seen as a patient-presiding doctor relationship.

On the other hand, steps S1750 and S1760 for establishing a relationship between the patient and the attending physician have been described after the step S1740 in which the visit reservation is completed, but the temporal precedence does not necessarily happen. - Of course, it will be possible to register as a doctor, and as described above, it will be said that the patient can freely establish a relationship between the patient and the doctor even for the hospital that the patient himself searched for. This will be said to be the same explanation for the cancellation of the relationship between the patient and the attending physician.

The relationship between the patient and the attending physician can be established through steps S1750 and S1760 described above, and after that, the hospital terminal (not shown) selects any one of the measured value of the personal health data of Chinese characters and the calculated PHR evaluation value from the device 100. The abnormality can be received, and the method of actively reserving a patient's visit schedule according to the second embodiment of the present invention described above will be performed with respect to the subsequent appointment.

So far, a method for recommending a hospital to establish a relationship between a patient and a doctor according to a fifth embodiment of the present invention has been described. According to the present invention, if there is no hospital that has accumulated medical trust through long-term use, the device 100 easily recommends a hospital optimized for the patient by simply sending the patient's personal health data measurement value, so the regular hospital Alternatively, the setting of the attending physician can be made easily. In addition, considering the difficulty of retrieving information about an appropriate hospital, the device 100

It is performed instead of a hospital search and a hospital information database (not shown) is periodically updated to recommend a hospital, but the patient's convenience can be remarkably improved by making it easy to make a reservation for a visit to the hospital. Furthermore, when an abnormality occurs according to the monitoring result of the patient's personal health data, by recommending a hospital that can perform treatment for this, it is possible to alert the patient to health management and establish a relationship between the patient and the attending physician. By creating a hospital, you can build an environment where systematic health management is possible.

The method for recommending a hospital for a patient-supervising doctor relationship according to the fifth embodiment of the present invention described above can be implemented with a computer program stored in a medium according to the sixth embodiment of the present invention that includes all the same technical features, Although it will not be described in detail to prevent redundant description, all technical features applied to the method for recommending a hospital for a patient-presiding physician relationship according to the fifth embodiment of the present invention described above, media according to the sixth embodiment of the present invention Of course, the same can be applied to a computer program stored in .

All embodiments of the present invention have been described so far. To this end, terms such as an intelligent health care device (3), a device for recommending visits led by the attending physician (100) based on real-time monitoring of the patient's personal health data, and a device for recommending hospitals (100) based on real-time monitoring of the patient's personal health data are used, respectively. Although the description corresponding to the embodiment has been carried out, all of these are implemented through a dedicated application installed on a patient terminal (not shown) or a hospital (primary doctor) terminal (not shown), or a source of service provision, that is, a kind of server. The server can be implemented as a platform, and the dedicated application can be implemented as an Intelligent Proactive Conversational Agent (IPCA), a chatbot using deep learning technology.

Such an IPCA can lead various processes described above, perform various functions, and contribute to enabling users (including both patients and doctors) to use the present invention more easily and conveniently. Unlike general chatbots that use functions, it monitors the user's current condition, including PHR, and the external environment around the user, such as weather and temperature. can initiate conversation. Through this, when an abnormal PHR signal is detected, a change in the surrounding environment such as changing seasons, or other important events occur, it is possible to actively start a conversation with the user to collect information about health information from the user (questionnaire) and to deliver information. All conversations with the user are made through the IPCA client (function included in the dedicated application) installed on the user terminal, and not only text, but also voice communication using TTS (Text-to-speech) and STT (Speech-to-Text) technologies. conversation is possible with In addition, the IPCA client can give psychological stability and trust to the user by communicating with the user through a virtual agent (avatar) in the form of a doctor.

In addition to this, the following techniques can be used to allow the user to continue the conversation more conveniently. First, when IPCA needs to obtain information or receive answers from users, it provides multiple-choice options so that users can easily make a choice. In addition, even when short-form answers are needed, examples can be provided so that users can answer more easily, and then, during conversation with the user, when the user cannot answer the IPCA question (breakdown occurs), etc. If it is determined that the user is hesitant to answer by using a method such as measuring the time taken from asking a question to entering an answer, an additional detailed explanation of the answer or an example of the answer may be automatically provided. This is a must-have function, especially given the nature of healthcare with many medical terms.

The IPCA server, which is an engine for implementing IPCA, not only includes an interface function with the monitoring engine of the platform to monitor the user's status and external environment, but also analyzes the user's question, command, or answer. It may have embedded deep learning natural language processing models. This engine can perform the IPCA's first understanding of what the user intends to do with the user's input, where the user's intention is 1) question/information inquiry, 2) PCA's previous (short-answer) question. response, 3) function execution order, and 4) others (if a clear intention cannot be grasped, talking to oneself, complaining, etc.). Conduct detailed content analysis according to intent.

More specifically, when the user's intention is 1) question/information search, it can be divided into information/data within the platform (patient's PHR information, patient's reservation information, platform function inquiry, etc.) inquiry and general question/information search. In the case of information/data search within the platform, after making the user's input into a search condition for searching on the platform, the query can be executed and the result can be notified to the user. Making the user's input into a search condition uses a search condition generating deep learning model provided by the platform, and this model is automatically trained according to user feedback. If the intention is to search for general questions/information that cannot be answered in the information/data within the platform, inform the user that they are searching the web because they cannot be answered on the platform, and search the entered content using an external search engine such as Google to find it. You can display an external page for results.

On the other hand, if the user's intention is 2) an answer to the previous (short-answer) question of the PCA, it is judged whether it is appropriate as an answer to the question, and this also uses a deep learning model, and the question-answer relevance (0 to 1) is If it is less than 0.5, it is possible to induce the user to provide more accurate answers/information by giving hints or examples of answers (for example, IPCA asks about the amount of time exercised for a week, and if the user answers that he or she is type A, 3 hours or Please answer in a format such as 2 hours and 30 minutes).

On the other hand, if the user's intention is 4) others, a message encouraging the user to input more specific information may be displayed.

In addition, for the IPCA open-ended question, in the case of 1) question/information search or 3) function execution command, where the intention of the content entered immediately after the user is not 2) an answer, an accurate answer is required to continue the conversation with the user. After performing the operation corresponding to 1) question/information search or 3) function execution command, the original question can be asked again to the user, and for this purpose, when IPCA asks the user a short-answer question, the answer whether or not it can be tracked.

Through such IPCA, the present invention belonging to the field of healthcare service can provide significant convenience to users, and the known papers Cartreine et al., 2010; As disclosed in Dayan et al., 2007, Locke et al., 1992, and Richman et al., 1999, when people have to disclose sensitive information to others, they are more concerned about privacy concerns and fear of negative reviews than real people. Since these fewer computer agents tend to be more comfortable, there is no doubt that IPCA can produce optimal effects when applied to the present invention.

Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art to which the present invention pertains can be implemented in other specific forms without changing the technical spirit or essential features of the present invention. you will be able to understand Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting.

10: Remote healthcare system
1: User device
2: medical personnel device
3: AI-based digital healthcare device
11: storage unit
12: Communication department
13: control unit

Claims (11)

A computer-implemented method executed by an intelligent healthcare device,
registering device information of a first device for measuring a bio-signal of the user by using user authentication information of the user;
determining whether biometric data, wherein the biometric data includes a measured value of the biosignal, is collected from the first device within a measurement period set for the biosignal; and
When it is determined that the biometric data has not been received from the first device within the measurement period, the first device and the second device - the second device being the user's communication terminal capable of communicating with the intelligent healthcare device - transmitting a measurement request for encouraging the user to measure the physiological signal using the first device to at least one of the following;
including,
Registering the device information,
issuing, to the first device, an access token in which the user authentication information is combined with device information of the first device;
Including,
The access token is
It is used for encryption of the measured value of the biosignal by the first device,
determining validity of the access token when receiving the biometric data received from the first device in a state encrypted by the access token;
decrypting the biometric data received from the first device using a decryption key corresponding to the access token when it is determined that the access token is valid; and
updating a first personal health time-series data set of the user by using the decoded biometric data, wherein the first personal health time-series data set includes a change history of the measured value over time;
Including more,
generating an artificial intelligence model for processing the first personal health time series data set; and
calculating an abnormality score related to the health condition of the user based on the first personal health time-series data set using the artificial intelligence model;
Including more,
The step of generating the artificial intelligence model,
Generating a first model outputting a first score related to the biosignal by performing comparison between at least one normal range determined for the biosignal and at least one measured value included in the first personal health time series data set doing;
Learning from a plurality of reference health time-series data sets for the bio-signals previously acquired from a plurality of individuals other than the user, processing the first personal health time-series data set to output a second score related to the bio-signals 2 Creating a model;
generating a third model learned from the first personal health time-series data set and outputting a third score related to the bio-signal by processing the first personal health time-series data set; and
combining the first model, the second model, and the third model to output a weighted average of the first score, the second score, and the third score as the ideal score;
including,
Computer-implemented method.
delete delete According to claim 1,
modifying the measurement period set for the bio-signal based on at least one of the first personal health time-series data set and external environment information;
Further comprising, a computer-implemented method.
According to claim 1,
If it is determined that a medical examination of the user is necessary, generating a digital medical examination table to be presented to the user, wherein the digital medical examination table includes a plurality of medical examination items, each of which is a multiple choice type or a subjective type;
sending a medical examination request for the digital medical questionnaire to the second device to call an intelligent active conversation agent installed in the second device;
receiving a medical examination result for the digital medical examination form from the second device through interaction with the intelligent active conversation agent; and
Updating a second personal health time-series data set of the user by processing the medical examination result, the second personal health time-series data set including a history of change over time of at least one health-related item different from the biosignal step;
Further comprising, a computer-implemented method.
delete delete According to claim 1,
The step of generating the second model,
A first candidate model is generated through supervised learning using the plurality of reference health time series data sets, wherein each measurement value included in the plurality of reference health time series data sets is classified as a normal measurement value or an abnormal measurement value. doing;
When the ratio of abnormal measurement values of the plurality of reference health time series data sets is greater than or equal to the first reference ratio and less than the second reference ratio, downsampling of normal measurement values and abnormal measurement values of the plurality of reference health time series data sets processing the plurality of reference health time-series data sets through at least one execution of up-sampling, and generating a second candidate model through supervised learning using the processed plurality of reference health time-series data sets;
When the ratio of abnormal measurement values of the plurality of reference health time series data sets is less than the first reference ratio, a third candidate model through unsupervised learning using only normal measurement values of the plurality of reference health time series data sets generating;
evaluating performance of each of the first candidate model, the second candidate model, and the third candidate model using a first test data set;
selecting one of the first candidate model, the second candidate model, and the third candidate model evaluated to have the highest performance as the second model;
A computer-implemented method comprising:
According to claim 1,
The step of generating the third model,
generating a fourth candidate model through supervised learning using the first personal health time series data set;
When the ratio of abnormal measurement values of the first personal health time series data set is greater than or equal to the third reference ratio and less than the fourth reference ratio, downsampling for normal measurement values and abnormal measurement values of the first personal health time series data set processing the first personal health time-series data set through at least one execution of up-sampling, and generating a fifth candidate model through supervised learning using the processed first personal health time-series data set;
generating a sixth candidate model through unsupervised learning using only normal measurement values of the first personal health time-series data set when the ratio of abnormal measurement values of the first personal health time-series data set is less than the third reference ratio; ;
evaluating performance of each of the fourth candidate model, the fifth candidate model, and the sixth candidate model using a second test data set;
selecting one of the fourth candidate model, the fifth candidate model, and the sixth candidate model evaluated to have the highest performance as the third model;
A computer-implemented method comprising:
According to claim 1,
Combining the first model, the second model and the third model,
adjusting first weights, second weights, and third weights according to the number of measured values recorded in the first personal health time-series data set;
Including,
The sum of the first weight, the second weight, and the third weight is 1,
Whenever the number of measured values recorded in the first personal health time series data set increases by the reference number, the first weight is decreased by a first adjustment value within a first range, and the second weight is within a second range decreases by a second adjustment value within a range, and the third weight increases by a third adjustment value within a third range.
a storage unit including a computer-readable recording medium in which a computer program is stored; and
A control unit operably coupled to the storage unit;
The control unit, when the computer program is executed,
a process of registering device information of a first device for measuring a bio-signal of the user by using user authentication information of the user;
a process of determining whether biometric data, wherein the biometric data includes a measurement value of the biosignal, is collected from the first device within a measurement period set for the biosignal;
When it is determined that the biometric data has not been received from the first device within the measurement period, the first device and the second device, wherein the second device is the user's communication terminal capable of communicating with the intelligent healthcare device, are between Execute a process of transmitting, to at least one, a measurement request for encouraging the user to measure the biosignal using the first device;
The process of registering the device information,
a process of issuing, to the first device, an access token in which the user authentication information is combined with the device information of the first device;
Including,
The access token is
It is used for encryption of the measured value of the biosignal by the first device,
a process of determining validity of the access token upon receiving the biometric data received from the first device in a state encrypted by the access token;
a process of decrypting the biometric data received from the first device using a decryption key corresponding to the access token when it is determined that the access token is valid; and
a process of updating a first personal health time-series data set of the user using the decoded biometric data, the first personal health time-series data set including a change history of the measurement values over time;
Including more,
a process of generating an artificial intelligence model for processing the first personal health time series data set; and
A process of calculating an abnormality score related to the health state of the user based on the first personal health time-series data set using the artificial intelligence model;
Including more,
The process of generating the artificial intelligence model,
Generating a first model that outputs a first score related to the biosignal by performing comparison between at least one normal range determined for the biosignal and at least one measured value included in the first personal health time series data set the process of doing;
Learning from a plurality of reference health time-series data sets for the bio-signals previously acquired from a plurality of individuals other than the user, processing the first personal health time-series data set to output a second score related to the bio-signals 2 The process of creating a model;
a process of generating a third model learned from the first personal health time-series data set and outputting a third score related to the bio-signal by processing the first personal health time-series data set; and
a process of combining the first model, the second model, and the third model to output a weighted average of the first score, the second score, and the third score as the ideal score;
, intelligent healthcare device.

Images