KR102688727B1 - Personalized smart blood sugar care monitoring system, method, and application using the same through biometric data measurement - Google Patents

Abstract

The present inventor's personalized smart blood sugar care monitoring system through biometric data measurement emits light in a preset wavelength band to the user's body and receives reflected or projected light reflected from the user's body to predict the user's blood sugar level. A prediction factor information measurement unit that acquires factor information, performs real-time model optimization using the prediction factor information and multi-modal data based on artificial neural network technology, and performs inference of blood sugar and blood sugar change values using the optimized model. A data inference unit that acquires the resulting inferred blood sugar information, and a data learning unit that performs retraining of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information. It can be included.

Description

Personalized smart blood sugar care monitoring system, method, and application using the same through biometric data measurement}

The present invention relates to a personalized smart blood sugar care monitoring system and method through biometric data measurement, and an application using the same. In particular, it analyzes and measures various biometric data such as body blood sugar, body temperature, movement, blood alcohol, and lactic acid to provide data values and It relates to a personalized smart blood sugar care monitoring system, method, and application using the same through biometric data measurement that can monitor relative changes in real time or intermittently.

For diabetic patients, continuous blood sugar monitoring is essential to prevent problems caused by sudden rises or falls in blood sugar levels and manage blood sugar levels.

In particular, type 1 diabetic patients are unable to produce insulin on their own, so there is a high possibility of hyperglycemia or hypoglycemia occurring and an emergency situation. To prevent this, continuously monitor the measured blood sugar to see if it is hyperglycemic or hypoglycemic. And, if necessary, insulin should be administered immediately or foods high in sugar content should be consumed.

In addition, type 2 diabetes patients must continuously monitor their blood sugar level and manage their blood sugar level by improving eating habits and lifestyle habits, so blood sugar monitoring must be performed.

Recently, various continuous blood sugar monitors have been introduced in Korea, and these diabetic patients are using the monitors to monitor their blood sugar levels.

However, since most continuous blood glucose monitors are invasive, a needle-shaped sensor must be inserted into the body and connected to the inside of the body. Not only must it be attached to the body for at least 10 days, but the needle marks can cause scarring even after use. There is a problem that side effects such as skin eczema occur at the attachment site.

To overcome the shortcomings of these invasive continuous blood glucose monitors, various non-invasive continuous blood glucose measurement technologies are being developed.

Non-invasive continuous blood sugar measurement technology includes the Reverse Iontophoresis method, which measures electrochemically using sweat generated from the body, the photoacoustic method, the RF high-frequency sensing method in the gigahertz (GHz) band, Raman spectroscopy, and the megahertz (MHz) method. Band-aided body impedance measurement methods and infrared spectroscopy have been developed.

Most of these conventionally developed methods have difficulties in commercialization due to problems such as low blood sugar measurement accuracy or difficulty in continuously measuring blood sugar.

In particular, there are differences in physical and electrical characteristics for each user or for each area where the device is worn, but existing methods apply the same blood sugar prediction model based on algorithm-based parameters to all devices, reflecting differences in characteristics by individual/location. There is a problem of low blood sugar measurement accuracy.

Korean Patent Publication No. 10-0871074 (Publication date: August 6, 2008)

In order to solve the problems of the prior art as described above, an embodiment of the present invention not only enables more accurate blood sugar inference by reflecting various biometric data by inferring blood sugar using optical spectroscopy, but also enables efficient and efficient use of invasive devices. We aim to provide a personalized smart blood sugar care monitoring system, method, and application using the same through biometric data measurement with an effective calibration method.

According to one aspect of the present invention to solve the above problems, a personalized smart blood sugar care monitoring system through biometric data measurement is provided. The personalized smart blood sugar care monitoring system through biometric data measurement emits light in a preset wavelength band to the user's body and receives reflected or projected light reflected from the user's body to provide a predictive factor for predicting the user's blood sugar level. Predictive factor information measurement unit to obtain information; Based on artificial neural network technology, real-time model optimization is performed using the prediction factor information and multi-modal data, and data inference is performed to infer blood sugar and blood sugar change values using the optimized model and obtain the resulting inferred blood sugar information. wealth; and the data learning unit that performs retraining of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information.

The prediction factor information measuring unit includes a light emitting module that emits light in the preset wavelength band to the user's body; a light receiving module that receives the reflected or projected light using at least one photodiode; and a wavelength band calculation module that adjusts the preset wavelength band to calculate a customized wavelength band for each user.

At least two light emitting modules are formed, each including an LED or a laser diode, and the preset wavelength band is a short wavelength infrared (SWIR) band and may include a first wavelength band and a second wavelength band. .

The data inference unit may include a multi-modal data acquisition module that acquires the multi-modal data; a model optimization module that adopts an optimized model structure among pre-learned models, performs model labeling, performs the real-time model optimization, and obtains the optimized model; and an inferred blood sugar information acquisition module that acquires the inferred blood sugar information using the optimized model.

It further includes a correction information processor configured to generate or process correction information necessary to obtain the inferred blood sugar information, wherein the correction information processor is connected to a controllable IR signal output device to output an IR signal different from the SWIR signal. An anti-interference module that separates time sections to prevent interference; a determination module for determining whether to obtain inferred blood sugar information by acquiring at least one of the user's body temperature, movement information, blood sugar, and ambient noise; and determining whether to obtain the inferred blood sugar information. and a blood sugar calibration module that receives blood sugar measurement information from the user and performs blood sugar calibration using the input blood sugar measurement information.

According to one aspect of the present invention, a personalized smart blood sugar care monitoring method is provided through biometric data measurement. The personalized smart blood sugar care monitoring method through biometric data measurement uses a prediction factor information measuring unit to emit light in a preset wavelength band to the user's body, and receives reflected or projected light reflected from the user's body to allow the user A prediction factor information measurement step of acquiring prediction factor information for predicting blood sugar level; Through the data inference unit, real-time model optimization is performed using the prediction factor information and multi-modal data based on artificial neural network technology, and inference of blood sugar and blood sugar change values is performed using the optimized model, and the resulting inferred blood sugar information is provided. Obtaining data inference step; and a data learning step in which a data learning unit performs re-training of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information.

According to one aspect of the present invention, an application using a personalized smart blood sugar care monitoring method through biometric data measurement is provided. The application using the personalized smart blood sugar care monitoring method through biometric data measurement emits light in a preset wavelength band to the user's body and receives reflected or projected light reflected from the user's body to predict the user's blood sugar level. a predictive factor information measurement step of acquiring predictive factor information for; Based on artificial neural network technology, real-time model optimization is performed using the prediction factor information and multi-modal data, and data inference is performed to infer blood sugar and blood sugar change values using the optimized model and obtain the resulting inferred blood sugar information. step; and the data learning step of performing re-training of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information. Stored in a storage medium of a digital device to perform. do.

A personalized smart blood sugar care monitoring system, method, and application using the same through measuring biometric data according to an embodiment of the present invention can infer blood sugar with greater accuracy by reflecting various biometric data by inferring blood sugar using optical spectroscopy. In addition, it has the effect of having an efficient and effective calibration method with invasive devices.

Figure 1 is a block diagram of a personalized smart blood sugar care monitoring system through biometric data measurement according to an embodiment of the present invention.
Figure 2 is a block diagram of the prediction factor information measurement unit of Figure 1.
Figure 3 is a block diagram of the data inference unit of Figure 1.
FIG. 4 is a block diagram of the correction information processing unit of FIG. 1.
FIG. 5 is a flowchart showing a customized wavelength band calculation method performed in the wavelength band calculation module of FIG. 2.
Figure 6 is a flowchart of a personalized smart blood sugar care monitoring method through biometric data measurement according to an embodiment of the present invention.
Figure 7 is a flow chart of step S11 in Figure 6.
Figure 8 is a flowchart of step S13 in Figure 6.
Figure 9 is a flow chart of step S17 in Figure 6.
FIG. 10 is a flowchart showing how step S115 of FIG. 7 is performed.
Figure 11 is an exemplary diagram of the system of the present invention.
Figure 12 is an example diagram of a first wavelength band and a second wavelength band formed to determine the wavelength of SWIR in the present invention.
FIG. 13 is a flowchart explaining FIGS. 5 and 10 according to a specific embodiment.
Figure 14 is an example diagram visualizing the contents performed in the data inference unit (data inference step) and the data learning unit (data learning step) of the present invention.
Figure 15 is a diagram showing a) an example flowchart and b) an example data input screen of the calibration performed in the blood sugar calibration module (blood sugar calibration step) of the present invention.

Hereinafter, some embodiments of the present disclosure will be described in detail with reference to illustrative drawings. In adding reference numerals to components in each drawing, the same components may have the same reference numerals as much as possible even if they are shown in different drawings. Additionally, in describing the present embodiments, if it is determined that a detailed description of a related known configuration or function may obscure the gist of the present technical idea, the detailed description may be omitted. When “comprises,” “has,” “consists of,” etc. mentioned in the specification are used, other parts may be added unless “only” is used. When a component is expressed in the singular, it can also include the plural, unless specifically stated otherwise.

Additionally, in describing the components of the present disclosure, terms such as first, second, A, B, (a), and (b) may be used. These terms are only used to distinguish the component from other components, and the nature, sequence, order, or number of the components are not limited by the term.

In the description of the positional relationship of components, when two or more components are described as being “connected,” “coupled,” or “connected,” the two or more components are directly “connected,” “coupled,” or “connected.” ", but it should be understood that two or more components and other components may be further "interposed" and "connected," "combined," or "connected." Here, other components may be included in one or more of two or more components that are “connected,” “coupled,” or “connected” to each other.

In the explanation of temporal flow relationships related to components, operation methods, production methods, etc., for example, temporal precedence relationships such as “after”, “after”, “after”, “before”, etc. Or, when a sequential relationship is described, non-continuous cases may be included unless “immediately” or “directly” is used.

On the other hand, when a numerical value or corresponding information (e.g. level, etc.) for a component is mentioned, even if there is no separate explicit description, the numerical value or corresponding information is related to various factors (e.g. process factors, internal or external shocks, It can be interpreted as including the error range that may occur due to noise, etc.).

1 to 5 show a personalized smart blood sugar care monitoring system through biometric data measurement according to an embodiment of the present invention. FIG. 1 is a block diagram of a personalized smart blood sugar care monitoring system through biometric data measurement according to an embodiment of the present invention, FIG. 2 is a block diagram of the prediction factor information measurement unit of FIG. 1, and FIG. 3 is a data inference of FIG. 1. This is a block diagram of the unit, FIG. 4 is a block diagram of the correction information processing unit of FIG. 1, and FIG. 5 is a flowchart showing a customized wavelength band calculation method performed in the wavelength band calculation module of FIG. 2. Hereinafter, a personalized smart blood sugar care monitoring system (1, hereinafter referred to as system for convenience) through biometric data measurement according to an embodiment of the present invention will be described in detail using FIGS. 1 to 5.

The system 1 according to an embodiment of the present invention infers blood sugar using optical spectroscopy, and for this purpose, not only uses a model in which machine learning has been performed, but also performs calibration by obtaining the user's actual blood sugar measurement value. By applying conditions, it can be configured to not only output a calibration request signal to the user when calibration is necessary, but also provide customized monitoring information for blood sugar management, such as adjusting the monitoring cycle using the user's current biometric data.

To this end, the system 1 according to an embodiment of the present invention may be formed to include a prediction factor information measurement unit 11, a data inference unit 13, and a data learning unit 15, as shown in FIG. 1. there is.

The prediction factor information measuring unit 11 emits light in a preset wavelength band to the user's body and receives reflected or projected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level. To this end, the prediction factor information measuring unit 11 of the present invention is formed to include a light emitting module 111, a light receiving module 113, and a wavelength band calculation module 115, as shown in FIG. 2.

The light emitting module 111 is formed to emit light in a preset wavelength band to the user's body. Here, at least two light emitting modules 111 are formed, each including an LED or a laser diode, and the preset wavelength band is a short wavelength infrared (SWIR) band to include a first wavelength band and a second wavelength band. can be formed.

At least one of the two light emitting modules 111 is configured to emit a first wavelength, which is a wavelength included in the first wavelength band, and at least the other one is configured to emit a second wavelength, which is a wavelength included in the second wavelength band. The first wavelength band may be 1600 nm to 1850 nm, and the second wavelength band may be 2200 nm to 2400 nm.

In summary, in one embodiment of the present invention, at least two light emitting modules 111 are formed to emit light in different wavelength bands to the body, and the different wavelength bands may be a first wavelength band and a second wavelength band. . That is, the light emitting module 111 of the present invention can be formed to emit light in both the first and second wavelength bands.

In one embodiment of the present invention, the light emitting module 111 is formed to output an output wavelength band of a preset output range with the first wavelength or the second wavelength as the peak wavelength, and the default setting of the first wavelength is 1650 nm and the second wavelength is 1650 nm. The default setting of the wavelength may be 2300nm.

Additionally, the light emitting module 111 may be formed to change the output peak wavelength or output wavelength band by receiving light emitting module control information. For example, in the case of the second wavelength band, it can be configured to be adjustable in 10 steps in 20nm units, and this adjustable range and unit can be changed according to the administrator's settings.

The light receiving module 113 is formed to receive reflected or projected light using at least one photodiode. The light receiving module 113 may receive reflected light or projected light included in at least one of the first wavelength band or the second wavelength band, and receives light receiving module control information to change the wavelength band of the received reflected light or projected light. It can be formed as follows.

Figure 12 is an illustration of the first and second wavelength bands formed to determine the wavelength of SWIR in the present invention, and displays the above-described first and second wavelength bands, and the table on the left shows the glucose This is a diagram summarizing the absorption wavelength band of .

The output wavelength band determined by the light emitting module 111 and the light receiving module 113 of the present invention may be part of the entire first wavelength band as shown in FIG. 12.

The wavelength band calculation module 115 is configured to calculate a customized wavelength band for each user by adjusting a preset wavelength band. To this end, the wavelength band calculation module 115 according to an embodiment of the present invention can calculate a customized wavelength band according to the flowchart shown in FIGS. 5 and 13.

FIG. 13 is a flowchart explaining FIGS. 5 and 10 according to a specific embodiment. Referring to FIG. 13, when using the system of the present invention for the first time, the wavelength band calculation module 115 of the present invention performs multi-sensing through a SWIR signal and an invasive blood glucose meter (I-CGM), and the first and It is configured to receive a SWIR signal for the second wavelength band and perform calibration (tuning) to determine a user-customized optimal wavelength value. Afterwards, error monitoring between predicted values and I-CGM actual values is performed. Here, it is possible to check whether the error consistently appears at 20% or more, and if it appears continuously, re-determine the customized optimal wavelength value. If it appears at less than 20%, error monitoring can be performed continuously.

In more detail, referring to FIG. 5, the wavelength band calculation module 115 according to an embodiment of the present invention may first obtain the regional blood sugar prediction value (step A). Step A is a step of obtaining a regional blood sugar prediction value by selecting at least two wavelengths within each of the first and second wavelength bands and obtaining an input value for blood sugar prediction through emission or reception, respectively.

Once the regional blood glucose predictions are obtained, a customized region is next selected (step B). In step B, the actual blood sugar measurement value is acquired through the user terminal, and the obtained blood sugar measurement value is compared with the area blood sugar prediction value, and the blood sugar prediction value with the smallest error among the area blood sugar prediction values and the area containing the corresponding blood sugar prediction value are selected as the custom area. You can. At this time, step B may be repeated at least twice or more depending on the administrator's settings, and may be configured to select the area with more selections after repetition as the customized area.

Next, obtain the optimal wavelength value (step C). In step C, if a customized area is selected, the optimal wavelength value can be obtained by performing tuning a preset number of times within the customized area. Here, tuning means obtaining the blood sugar prediction value for each number of times, obtaining the actual blood sugar prediction value corresponding to the preset number of times, and determining the optimal wavelength value by obtaining the actual measurement value from the user terminal.

For example, in step C, if tuning is performed a total of 10 times, the optimal wavelength value can be determined by obtaining 10 blood sugar prediction values and then obtaining actual measured values obtained through actual body blood from the user or through I-CGM. there is. Step C can also be repeated multiple times depending on the administrator's settings.

In step C, the optimal predicted value, which is the predicted value at the optimal wavelength value, is compared with the actual measured value to obtain an error value, and if the error value is more than a preset ratio and is repeated more than a preset number of times, tuning in the corresponding area is performed as preset. This is repeated the number of times, and if the error value appears at a preset rate or more than the preset number of times, information indicating that the current error range is out of the range is output to the user terminal, and the step of selecting the custom area can be performed again.

For example, when data is repeated several times where the predicted value of the optimal wavelength value has an error of more than 20% compared to the actual measured value, the wavelength band calculation module 115 of the present invention determines the optimal wavelength through tuning within the corresponding wavelength region. The wavelength value calculation process can be repeated again (repeat step C). If the error continues to be more than 20% even if repeated again, the wavelength band calculation module 115 notifies the user that the current error range is out of the range, returns to the process of selecting the wavelength band, and performs comparison and wavelength band selection again through body blood. You may proceed (re-do Step A).

The data inference unit 13 performs real-time model optimization using predictive factor information and multi-modal data based on artificial neural network technology, performs inference of blood sugar and blood sugar change values using the optimized model, and infers the resulting blood sugar level. Formed to obtain information.

The data inference unit 13 of the present invention is formed to obtain inferred blood sugar information by processing data acquired through the light receiving module 113, and may use a data processing unit (ASIC or FPGA). The data inference unit 13 according to an embodiment of the present invention is based on artificial neural network technology and uses real-time data acquired through the light receiving module 113 and additionally usable multi-modal data to create an optimized model among the previously learned models. It is formed to proceed with real-time model optimization by adopting the model structure and performing model labeling. Here, model optimization means selecting one of the ensemble / expert model sets contained in ASIC or FPGA, and the re-learning process can be performed through an external terminal using wireless communication technology rather than on-chip. Ensemble / expert weak model set (model set) refers not only to a model set for multi-modal data application, but also to a model set that can be selectively utilized optimized for each signal situation. In addition, securing a large number of weak models used as ensemble models or expert models can have the effect of lightening models, estimating accelerator efficiency, and optimizing user data.

To this end, the data inference unit 13 according to an embodiment of the present invention includes a multi-modal data acquisition module 131, a model optimization module 133, and an inferred blood sugar information acquisition module 135, as shown in FIG. 3. It can be formed.

The multi-modal data acquisition module 131 is configured to acquire multi-modal data. In one embodiment of the present invention, multi-modal data is acquired using pre-entered data such as the user's gender, age, height, weight, actual blood sugar value using an invasive blood glucose meter, body temperature, movement, and sound input, or using a peripheral device. It may include at least one of the surrounding data.

The model optimization module 133 is configured to obtain an optimized model by adopting an optimized model structure among pre-learned models, performing model labeling, and performing real-time model optimization. Model optimization may mean selecting one of a plurality of pre-stored model sets as described above.

The model optimization module 133 may be configured to improve accelerator efficiency by checking unused model sections within the ASIC or FPGA based on the optimized model and power/clock gating the corresponding model sections.

The inferred blood sugar information acquisition module 135 is configured to acquire inferred blood sugar information using an optimized model. The inferred blood sugar information acquisition module 135 according to an embodiment of the present invention stops acquiring inferred blood sugar information or generates a signal to re-perform real-time model optimization if there is data that changes among the multi-modal data and changes more than a preset size. This can be formed to re-obtain the optimized model.

Through this, the inferred blood sugar information acquisition module 135 of the present invention can have the effect of responding to real-time model optimization and data variation without performing a re-learning process.

The data learning unit 15 is configured to retrain a basic model that can be used for real-time model optimization using prediction factor information, multi-modal data, and inferred blood sugar information. The data learning unit 15 according to an embodiment of the present invention performs retraining for optimization and error correction on predictive factor information in order to retrain the basic model, and learns peculiarities about the surrounding variation environment. In order to do so, it can be configured to perform retraining on the model using the corresponding variant environment information and then retraining on the entire base model.

As an example, the data learning unit 15 of the present invention may be configured to relearn and strengthen the required model structure and some weak models based on metadata based on accumulated data received through wireless communication. In this embodiment, the data learning unit 15 may receive inference data synchronized with the invasive blood glucose meter measurement from the ASIC or FPGA. The retraining process of the artificial neural network model based on the received and stored data performed in the data learning unit 15 of the present invention is as follows.

First, re-learning is performed for optimization and error correction of prediction factor information, which is data obtained from the light receiving module 133. This involves model calibration based on measurements from an invasive blood glucose meter. Next, retraining of the corresponding weak model is performed to learn the peculiarities of the variant environment among the surrounding data, and finally, retraining of the entire ensemble / expert set is performed to reorganize the weak model set, thereby retraining the entire data. It can be done. The retrained and reorganized weak model and weak model set can be defined as the retrained basic model and transmitted to the data inference unit 13.

Meanwhile, Figure 14 shows an example diagram visualizing the contents performed in the data inference unit (data inference step) and the data learning unit (data learning step) of the present invention. Referring to FIG. 14, the present invention includes SWIR first and second wavelength band input data (prediction factor information acquired through the light receiving module), gender, age, height, weight, (continuous) measured blood sugar value, body temperature, movement, and Information such as sound can be obtained. Here, gender, age, height, weight, (continuous) actual blood sugar values measured using an invasive blood glucose meter, etc. can be entered directly by the user through the user terminal.

In addition, the data inference unit 13 and the data learning unit 15 perform re-learning of each section using only selected elements, and may be configured to perform re-learning of the entire model when re-learning of each section is completed. .

Meanwhile, the system 1 according to an embodiment of the present invention may further include a correction information processing unit 17 as shown in FIG. 1. The correction information processing unit 17 may be configured to generate or process correction information necessary to obtain inferred blood sugar information. The correction information processing unit 17 may be configured to generate various types of correction information or to generate and process specific information to prevent errors from occurring. To this end, the correction information processing unit 17 according to an embodiment of the present invention includes an interference prevention module 171, a determination module 173 for obtaining inferred blood sugar information, and a blood sugar calibration module 175, as shown in FIG. 4. It can be formed to do so.

The interference prevention module 171 is connected to a controllable IR signal output device and is formed to prevent interference by separating the output time section of the SWIR signal and other IR signals. Here, the time section is divided into high/low form to output SWIR in the high time section and other IR signals in the low time section. In one embodiment, the signal is formed to output in the form of a group pulse, and the SWIR output pulse The group section and IR output pulse group section can be output separately. Through this, the interference prevention module 171 of the present invention can prevent aliasing.

The determining module 173 for determining whether to obtain inferred blood sugar information is configured to obtain at least one of the user's body temperature, movement information, blood sugar, and ambient noise to determine whether to measure or obtain inferred blood sugar information.

When blood sugar levels in the body increase significantly, the body reacts similarly to being infected with a pathogen, causing an increase in body temperature. Reflecting this, the module 173 for determining whether to obtain inferred blood sugar information of the present invention may perform sophisticated monitoring by reducing the cycle of temperature and blood sugar measurement to less than 1 minute when the blood sugar level rises.

Additionally, when the human body moves or actions such as massaging or pressing the body around the area where the blood sugar device is worn are performed, the accuracy of blood sugar sensing is significantly reduced. Therefore, the determination module 173 of the present invention is configured to stop acquiring inferred blood sugar information for a preset time when the user's movement is detected to be greater than a preset threshold using an IMU (Inertial Measurement Unit). You can.

In addition, the determination module 173 of the present invention determines whether to obtain the inferred blood sugar information by changing the body temperature acquisition cycle and the inferred blood sugar information acquisition cycle to 1 minute when the user's inferred blood sugar information increases by a preset rate compared to the average blood sugar value for a preset time. It can also be reduced to less than.

Meanwhile, the module 173 for determining whether to obtain inferred blood sugar information of the present invention may be formed to increase accuracy and usability by utilizing a plurality of non-invasive and invasive measurements. In the present invention, a user wears a wearable terminal capable of recognizing food and sound, and the worn wearable terminal recognizes the user's voice, food or drink intake, etc., or uses information directly input by the user to provide information on the user's current status. It can be obtained. The determining module 173 for determining whether to obtain inferred blood sugar information according to an embodiment of the present invention analyzes ambient noise and determines inferred blood sugar information when the sound of food/drink consumption continues for more than a preset time or when eating status information is input from the user. The acquisition cycle can also be reduced to less than 1 minute.

The blood sugar calibration module 175 according to an embodiment of the present invention may be configured to receive blood sugar measurement information from the user and perform blood sugar calibration using the input blood sugar measurement information.

The blood sugar calibration module 175 of the present invention receives blood sugar measurement information, and the blood sugar measurement information measured by the invasive continuous blood glucose meter is obtained in the form of a graph including time information, and a plurality of blood sugar information is obtained through analysis of the graph. When receiving input, the error rate can be obtained by comparing previously stored blood sugar values.

In addition, the blood sugar calibration module 175 of the present invention is used before the user's meal when the change range of the inferred blood sugar information is maintained below a preset error rate for a preset period and when the user's wakefulness is detected through constant motion detection. It may be configured to output calibration request information for performing blood sugar calibration to the user in at least one of the following situations: before/after the user consumes a beverage, before/after the user exercises, and in the user's resting state.

Figure 15 is a diagram showing a) an example flowchart and b) an example data input screen of the calibration performed in the blood sugar calibration module (blood sugar calibration step) of the present invention. Referring to FIG. 15, the blood sugar calibration module 175 according to an embodiment of the present invention uses SWIR first and second wavelength band input data (prediction factor information acquired through a light receiving module) and an invasive blood glucose meter (I-CGM). ), and when I-CGM data is input from the user, it can be configured to check body temperature and movement, and exclude the corresponding predicted value if there is an abnormality in the temperature or movement data.

In addition, the blood sugar calibration module 175 according to an embodiment of the present invention uses SWIR first and second wavelength band input data (predictive factor information acquired through a light receiving module and multi-processing data through an invasive blood glucose meter (I-CGM)). When performing sensing and receiving an I-CGM graph from the user, the temperature and movement data of the time section is analyzed through graph analysis, and if an abnormality occurs in the temperature or movement data section, calibration is performed excluding the corresponding predicted value. It may be formed to do so.

The screen on which the above-described FIG. 15A is displayed is shown in FIG. 15B. FIG. 15B is a calibration screen that allows the user to directly input blood sugar or a graph, and the current state can be configured to select one of the examples shown in FIG. 15B when directly inputting blood sugar.

Meanwhile, the system 1 of the present invention may be represented by an exemplary diagram as shown in FIG. 11. Figure 11 is an exemplary diagram of the system of the present invention. The system 1 according to an embodiment of the present invention acquires measured values obtained from an invasive finger stick or continuous blood glucose meter (I-CGM) through a user terminal, and the user terminal is equipped with a data learning unit 15. It can be formed as much as possible. In addition, prediction factor information obtained from the multi SWIR sensors 111 and 113 is acquired by the prediction factor information measurement unit 11, and the obtained prediction factor information is transmitted to the data inference unit 13 and inferred through the central controller. Blood sugar information can be output. As described above, the central controller may be connected to a motion sensor, a temperature sensor, a microphone, etc. to obtain calibration or external environment information.

Meanwhile, Figures 6 to 10 show a flowchart of a personalized smart blood sugar care monitoring method (hereinafter referred to as the method for convenience) through biometric data measurement according to an embodiment of the present invention. Figure 6 is a flowchart of a personalized smart blood sugar care monitoring method through biometric data measurement according to an embodiment of the present invention, Figure 7 is a flowchart of step S11 of Figure 6, Figure 8 is a flowchart of step S13 of Figure 6, FIG. 9 is a flowchart of step S17 of FIG. 6, and FIG. 10 is a flowchart showing how step S115 of FIG. 7 is performed. Hereinafter, the method 10 of the present invention will be described in detail using FIGS. 6 to 10. For convenience of explanation, the system of FIG. 1 is used, but the present invention is not limited thereto.

The method 10 according to an embodiment of the present invention infers blood sugar using optical spectroscopy, not only uses a model in which machine learning has been performed for this purpose, but also performs calibration by obtaining the user's actual blood sugar measurement value. By applying conditions, it can be configured to not only output a calibration request signal to the user when calibration is necessary, but also provide customized monitoring information for blood sugar management, such as adjusting the monitoring cycle using the user's current biometric data.

To this end, the method 10 according to an embodiment of the present invention can be formed to include a prediction factor information measurement step (S11), a data inference step (S13), and a data learning step (S15) as shown in FIG. 6. there is.

The prediction factor information measurement step (S11) emits light in a preset wavelength band to the user's body through the prediction factor information measurement unit and receives reflected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level. do. To this end, the prediction factor information measurement step (S11) of the present invention is formed to include a light emitting step (S111), a light receiving step (S113), and a wavelength band calculation step (S115) as shown in FIG. 7.

The light emission step (S111) is formed so that the user's body emits light in a preset wavelength band. Here, the light-emitting step (S111) is performed using at least two light-emitting modules, each including an LED or a laser diode, and the preset wavelength band is a short wavelength infrared (SWIR) band, which includes the first wavelength band and the second wavelength band. It can be formed to include a wavelength band.

At least one of the at least two light emitting modules is configured to emit a first wavelength, which is a wavelength included in the first wavelength band, and the other at least one is configured to emit a second wavelength, which is a wavelength included in the second wavelength band, The first wavelength band may be 1600 nm to 1850 nm, and the second wavelength band may be 2200 nm to 2400 nm.

In summary, in one embodiment of the present invention, the light emitting step (S111) includes at least two light emitting modules, and is formed to emit light in different wavelength bands to the body, and the different wavelength bands are a first wavelength band and a second wavelength band. It can be. That is, the light emitting module of the present invention can be formed to emit light in both the first and second wavelength bands.

In one embodiment of the present invention, the light emitting module is formed to output an output wavelength band of a preset output range with the first wavelength or the second wavelength as the peak wavelength, the default setting of the first wavelength is 1650nm, and the default setting of the second wavelength is 1650nm. The setting may be 2300nm.

Additionally, the light emitting step (S111) may be configured to receive light emitting module control information and change the peak wavelength or output wavelength band output from the light emitting module. For example, in the case of the second wavelength band, it can be configured to be adjustable in 10 steps in 20nm units, and this adjustable range and unit can be changed according to the administrator's settings.

The light receiving step (S113) is configured to receive reflected or projected light using at least one photodiode. In the light receiving step (S113), reflected light included in at least one of the first wavelength band and the second wavelength band can be received using a light receiving module, and the wavelength band of the received reflected light can be changed by receiving light receiving module control information. It can be formed as follows.

Figure 12 is an illustration of the first and second wavelength bands formed to determine the wavelength of SWIR in the present invention, and displays the above-described first and second wavelength bands, and the table on the left shows the glucose This is a diagram summarizing the absorption wavelength band of .

The output wavelength band determined in the light emitting step (S111) and the light receiving step (S113) of the present invention may be part of the entire first wavelength band as shown in FIG. 12.

The wavelength band calculation step (S115) is formed to calculate a customized wavelength band for each user by adjusting the preset wavelength band. To this end, the wavelength band calculation step (S115) according to an embodiment of the present invention can calculate a customized wavelength band according to the flowchart shown in FIGS. 10 and 13.

FIG. 13 is a flowchart explaining FIGS. 5 and 10 according to a specific embodiment. Referring to Figure 13, in the wavelength band calculation step (S115) of the present invention, when using the system of the present invention for the first time, multi-sensing is performed through a SWIR signal and an invasive blood glucose meter (I-CGM), and the first and It is configured to receive a SWIR signal for the second wavelength band and perform calibration (tuning) to determine a user-customized optimal wavelength value. Afterwards, error monitoring between predicted values and I-CGM actual values is performed. Here, it is possible to check whether the error consistently appears at 20% or more, and if it appears continuously, re-determine the customized optimal wavelength value. If it appears at less than 20%, error monitoring can be performed continuously.

In more detail, referring to FIG. 10, the wavelength band calculation step (S115) according to an embodiment of the present invention may first obtain a regional blood sugar prediction value (step A). Step A is a step of obtaining a regional blood sugar prediction value by selecting at least two wavelengths within each of the first and second wavelength bands and obtaining an input value for blood sugar prediction through emission or reception, respectively.

Once the regional blood glucose predictions are obtained, a customized region is next selected (step B). In step B, the actual blood sugar measurement value is acquired through the user terminal, and the obtained blood sugar measurement value is compared with the area blood sugar prediction value, and the blood sugar prediction value with the smallest error among the area blood sugar prediction values and the area containing the corresponding blood sugar prediction value are selected as the custom area. You can. At this time, step B may be repeated at least twice or more depending on the administrator's settings, and may be configured to select the area with more selections after repetition as the customized area.

Next, obtain the optimal wavelength value (step C). In step C, if a customized area is selected, the optimal wavelength value can be obtained by performing tuning a preset number of times within the customized area. Here, tuning means obtaining the blood sugar prediction value for each number of times, obtaining the actual blood sugar prediction value corresponding to the preset number of times, and determining the optimal wavelength value by obtaining the actual measurement value from the user terminal.

For example, in step C, if tuning is performed a total of 10 times, the optimal wavelength value can be determined by obtaining 10 blood sugar prediction values and then obtaining actual measured values obtained through actual body blood from the user or through I-CGM. there is. Step C can also be repeated multiple times depending on the administrator's settings.

In step C, the optimal predicted value, which is the predicted value at the optimal wavelength value, is compared with the actual measured value to obtain an error value, and if the error value is more than a preset ratio and is repeated more than a preset number of times, tuning in the corresponding area is performed as preset. This is repeated the number of times, and if the error value appears at a preset rate or more than the preset number of times, information indicating that the current error range is out of the range is output to the user terminal, and the step of selecting the custom area can be performed again.

For example, when data is repeated several times where the predicted value of the optimal wavelength value has an error of more than 20% compared to the actual measured value, the wavelength band calculation step (S115) of the present invention determines the optimal wavelength through tuning within the corresponding wavelength region. The wavelength value calculation process can be repeated again (repeat step C). If the error continues to be more than 20% even if repeated again, the wavelength band calculation step (S115) notifies the user that the current error range is out of the range, and returns to the process of selecting the wavelength band, again comparing through body blood and selecting the wavelength band. You may proceed (re-do Step A).

In the data inference step (S13), real-time model optimization is performed using predictive factor information and multi-modal data based on artificial neural network technology using the data inference unit, and inference of blood sugar and blood sugar change values is performed using the optimized model. It is configured to obtain the resulting inferred blood sugar information.

The data inference step (S13) of the present invention is formed to obtain inferred blood sugar information by processing the data obtained through the light reception step (S113), and can use a data processing unit (ASIC or FPGA). The data inference step (S13) according to an embodiment of the present invention is based on artificial neural network technology and uses real-time data acquired through the light reception step (S113) and additionally available multi-modal data to create an optimized model among the previously learned models. It is formed to proceed with real-time model optimization by adopting the model structure and performing model labeling. Here, model optimization means selecting one of the ensemble / expert model sets contained in ASIC or FPGA, and the re-learning process can be performed through an external terminal using wireless communication technology rather than on-chip. Ensemble / expert weak model set (model set) refers not only to a model set for multi-modal data application, but also to a model set that can be selectively utilized optimized for each signal situation. In addition, securing a large number of weak models used as ensemble models or expert models can have the effect of lightening models, estimating accelerator efficiency, and optimizing user data.

To this end, the data inference step (S13) according to an embodiment of the present invention includes a multi-modal data acquisition step (S131), a model optimization step (S133), and an inferred blood sugar information acquisition step (S135) as shown in FIG. It can be formed.

The multi-modal data acquisition step (S131) is configured to acquire multi-modal data. In one embodiment of the present invention, multi-modal data is acquired using pre-entered data such as the user's gender, age, height, weight, actual blood sugar value using an invasive blood glucose meter, body temperature, movement, and sound input, or using a peripheral device. It may include at least one of the surrounding data.

The model optimization step (S133) is formed to obtain an optimized model by adopting an optimized model structure among the already learned models, performing model labeling, and performing real-time model optimization. Model optimization may mean selecting one of a plurality of pre-stored model sets as described above.

The model optimization step (S133) may be configured to check unused model sections within the ASIC or FPGA based on the optimized model and improve accelerator efficiency by power/clock gating the corresponding model sections.

The inferred blood sugar information acquisition step (S135) is formed to obtain inferred blood sugar information using an optimized model. In the inferred blood sugar information acquisition step (S135) according to an embodiment of the present invention, if there is data that changes among the multi-modal data and changes more than a preset size, acquisition of inferred blood sugar information is stopped or a real-time model optimization re-performance signal is generated. This can be formed to re-obtain the optimized model.

Through this, the inferred blood sugar information acquisition step (S135) of the present invention can have the effect of responding to real-time model optimization and data variation without a re-learning process.

The data learning step (S15) is formed to perform retraining of a basic model that can be used for real-time model optimization using prediction factor information, multi-modal data, and inferred blood sugar information through the data learning unit. The data learning step (S15) according to an embodiment of the present invention performs relearning for optimization and error correction on predictive factor information in order to retrain the basic model, and learns peculiarities about the surrounding variation environment. In order to do so, it can be configured to perform retraining on the model using the corresponding variant environment information and then retraining on the entire base model.

As an example, the data learning step (S15) of the present invention may be formed to relearn and strengthen the required model structure and some weak models based on metadata based on accumulated data received through wireless communication. In this embodiment, the data learning step (S15) may receive inference data synchronized with the invasive blood glucose meter measurement from the ASIC or FPGA. The retraining process of the artificial neural network model based on the received and stored data performed in the data learning step (S15) of the present invention is as follows.

First, re-learning is performed for optimization and error correction of prediction factor information, which is data obtained from the light reception step (S133). This involves model calibration based on measurements from an invasive blood glucose meter. Next, retraining of the corresponding weak model is performed to learn the peculiarities of the variant environment among the surrounding data, and finally, retraining of the entire ensemble / expert set is performed to reorganize the weak model set, thereby retraining the entire data. It can be done. The retrained and reorganized weak model and weak model set can be defined as the retrained basic model and passed to the data inference step (S13).

Meanwhile, Figure 14 shows an example diagram visualizing the contents performed in the data inference unit (data inference step) and the data learning unit (data learning step) of the present invention. Referring to FIG. 14, the present invention includes SWIR first and second wavelength band input data (prediction factor information acquired through the light receiving module), gender, age, height, weight, (continuous) measured blood sugar value, body temperature, movement, and Information such as sound can be obtained. Here, gender, age, height, weight, (continuous) actual blood sugar values measured using an invasive blood glucose meter, etc. can be entered directly by the user through the user terminal.

In addition, the data inference step (S13) and the data learning step (S15) perform re-learning of each section using only the selected elements, and can be configured to perform re-learning of the entire model when re-learning of each section is completed. .

Meanwhile, the method 10 according to an embodiment of the present invention may further include a correction information processing step (S17) as shown in FIG. 6. The correction information processing step (S17) may be configured to generate or process correction information necessary to obtain inferred blood sugar information using a correction information processor. The correction information processing step (S17) may be configured to generate various types of correction information or to generate and process specific information to prevent errors from occurring. To this end, the correction information processing step (S17) according to an embodiment of the present invention includes an interference prevention step (S171), a determining step (S173) whether to obtain inferred blood sugar information, and a blood sugar calibration step (S175) as shown in FIG. 9. It can be formed to include.

The interference prevention step (S171) is connected to a controllable IR signal output device to separate the output time section of the SWIR signal and other IR signals to prevent interference. Here, the time section is divided into high/low form to output SWIR in the high time section and other IR signals in the low time section. In one embodiment, the signal is formed to output in the form of a group pulse, and the SWIR output pulse The group section and IR output pulse group section can be output separately. Through this, the interference prevention step (S171) of the present invention can prevent the aliasing phenomenon.

The determining step (S173) of whether to obtain inferred blood sugar information is formed to obtain at least one of the user's body temperature, movement information, blood sugar, and ambient noise to determine whether to measure or obtain inferred blood sugar information.

When blood sugar levels in the body increase significantly, the body reacts similarly to being infected with a pathogen, causing an increase in body temperature. Reflecting this, the step of determining whether to obtain inferred blood sugar information (S173) of the present invention may enable sophisticated monitoring to be performed by reducing the cycle of temperature and blood sugar measurement to less than 1 minute when the blood sugar level rises.

Additionally, when the human body moves or actions such as massaging or pressing the body around the area where the blood sugar device is worn are performed, the accuracy of blood sugar sensing is significantly reduced. Therefore, the determination step (S173) of the present invention for obtaining inferred blood sugar information is configured to stop the acquisition of inferred blood sugar information for a preset time when the user's movement is detected to be greater than a preset threshold using an IMU (Inertial Measurement Unit). You can.

In addition, in the step of determining whether to obtain inferred blood sugar information of the present invention (S173), when the user's inferred blood sugar information increases by a preset rate compared to the average blood sugar value for a preset time, the body temperature acquisition cycle and the inferred blood sugar information acquisition cycle are changed to 1 minute. It can also be reduced to less than.

Meanwhile, the determining step (S173) of whether to obtain inferred blood sugar information of the present invention may be formed to increase accuracy and usability by utilizing a plurality of non-invasive and invasive measurements. In the present invention, a user wears a wearable terminal capable of recognizing food and sound, and the worn wearable terminal recognizes the user's voice, food or drink intake, etc., or uses information directly input by the user to provide information on the user's current status. It can be obtained. The step (S173) of determining whether to obtain inferred blood sugar information according to an embodiment of the present invention analyzes ambient noise and determines inferred blood sugar information if the sound of food/drink consumption continues for more than a preset time or if eating status information is input from the user. The acquisition cycle can also be reduced to less than 1 minute.

The blood sugar calibration step (S175) according to an embodiment of the present invention may be configured to receive blood sugar measurement information from the user and perform blood sugar calibration using the input blood sugar measurement information.

The blood sugar calibration step (S175) of the present invention receives blood sugar measurement information, and the blood sugar measurement information measured by the invasive continuous blood glucose meter is obtained in the form of a graph including time information, and a plurality of blood sugar information is obtained through analysis of the graph. When receiving input, the error rate can be obtained by comparing previously stored blood sugar values.

In addition, the blood sugar calibration step (S175) of the present invention is performed when the change range of the inferred blood sugar information is maintained below a preset error rate for a preset period, when the user's wakefulness is detected through constant motion detection, and before the user's meal. It may be configured to output calibration request information for performing blood sugar calibration to the user in at least one of the following situations: before/after the user consumes a beverage, before/after the user exercises, and in the user's resting state.

Figure 15 is a diagram showing a) an example flowchart and b) an example data input screen of the calibration performed in the blood sugar calibration module (blood sugar calibration step) of the present invention. Referring to FIG. 15, the blood sugar calibration step (S175) according to an embodiment of the present invention uses SWIR first and second wavelength band input data (prediction factor information obtained through a light receiving module and an invasive blood glucose meter (I-CGM) ), and when I-CGM data is input from the user, it can be configured to check body temperature and movement, and exclude the corresponding predicted value if there is an abnormality in the temperature or movement data.

In addition, the blood sugar calibration step (S175) according to an embodiment of the present invention is the SWIR first and second wavelength band input data (prediction factor information acquired through the light receiving module and multi-processing through an invasive blood glucose meter (I-CGM)). When performing sensing and receiving an I-CGM graph from the user, the temperature and movement data of the time section is analyzed through graph analysis, and if an abnormality occurs in the temperature or movement data section, calibration is performed excluding the corresponding predicted value. It may be formed to do so.

The screen on which the above-described FIG. 15A is displayed is shown in FIG. 15B. FIG. 15B is a calibration screen that allows the user to directly input blood sugar or a graph, and the current state can be configured to select one of the examples shown in FIG. 15B when directly inputting blood sugar.

The personalized smart blood sugar care monitoring method through biometric data measurement according to the embodiments of the present invention described above can be implemented as an application (computer program) stored in a computer's storage medium.

Here, the computer may include a personalized smart blood sugar care monitoring system through biometric data measurement.

The operating system of the computer is either Windows or Macintosh, which are installed on general PCs such as desktops and laptops, or iOS, Android, etc., which are installed on mobile devices such as smartphones and tablet PCs. It may be a mobile-specific operating system, etc.

The personalized smart blood sugar care monitoring method through biometric data measurement according to the embodiments of the present invention described above is implemented as an application (i.e., computer program) installed by default on a computer or installed by the user, and is stored in computer-readable storage. Can be stored (recorded) on media

An application (computer program) that implements a personalized smart blood sugar care monitoring method through biometric data measurement according to embodiments of the present invention and is stored and executed in a computer's storage medium emits light in a preset wavelength band to the user's body and , a prediction factor information measuring step of receiving reflected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level; Based on artificial neural network technology, real-time model optimization is performed using the prediction factor information and multi-modal data, and data inference is performed to infer blood sugar and blood sugar change values using the optimized model and obtain the resulting inferred blood sugar information. step; and the data learning step of performing re-training of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information; etc. can be performed.

In this way, in order for the computer to read the program recorded on the storage medium and execute the personalized smart blood sugar care monitoring method through biometric data measurement according to the present embodiments implemented as a program, the above-described application (application program) is installed on the computer. It may include code coded in a computer language such as C, C++, JAVA, or machine language that can be read by a processor (CPU).

These codes may include functional codes related to functions that define the above-described functions, and may also include control codes related to execution procedures necessary for the computer's processor to execute the above-mentioned functions according to predetermined procedures.

In addition, these codes may further include memory reference-related codes that determine which location (address address) in the computer's internal or external memory the additional information or media required for the computer's processor to execute the above-mentioned functions should be referenced. .

In addition, if the computer's processor needs to communicate with any other remote computer or server in order to execute the above-mentioned functions, the code must be used to enable the computer's processor to communicate with the computer's communication module (e.g., wired and/or wireless communication module). ) may be used to further include communication-related codes on how to communicate with any other computer or server located remotely, and what information or media should be transmitted and received during communication.

In addition, functional programs and related codes and code segments for implementing the present embodiments are designed by programmers in the technical field to which the present invention belongs, taking into account the system environment of a computer that reads a storage medium and executes the program. It can also be easily inferred or changed by others.

In addition, a computer-readable storage medium recording the above-described program can be distributed to a computer system connected to a network, so that computer-readable code can be stored and executed in a distributed manner. In this case, any one or more of the multiple distributed computers may execute some of the functions presented above, transmit the execution results to one or more of the other distributed computers, and receive the results. A computer can also execute some of the functions presented above and provide the results to other distributed computers as well.

As described above, computer-readable storage media recording an application for executing the personalized smart blood sugar care monitoring method through biometric data measurement according to embodiments of the present invention include, for example, ROM, RAM, It may include CD-ROM, magnetic tape, floppy disk, optical media storage, etc.

In addition, a computer-readable storage medium recording an application, which is a program for executing a personalized smart blood sugar care monitoring method through biometric data measurement according to embodiments of the present invention, is an application store server, an application Or, it may be a storage medium (e.g. hard disk, etc.) included in an application provider server, including a web server related to the service, or it may be the application provider server itself, or another device that records the program. It may be a computer or its storage medium.

A computer that can read a storage medium recording an application, a program for executing a personalized smart blood sugar care monitoring method through biometric data measurement according to embodiments of the present invention, is not only a general PC such as a general desktop or laptop, It may include mobile terminals such as smart phones, tablet PCs, PDAs (Personal Digital Assistants), and mobile communication terminals, and should be interpreted as all devices capable of computing.

The above description is merely an illustrative explanation of the technical idea of the present invention, and various modifications and variations will be possible to those skilled in the art without departing from the essential characteristics of the present invention. Accordingly, the embodiments disclosed in the present invention are not intended to limit the technical idea of the present invention, but are for illustrative purposes, and the scope of the technical idea of the present invention is not limited by these embodiments. The scope of protection of the present invention should be interpreted in accordance with the claims below, and all technical ideas within the equivalent scope should be construed as being included in the scope of rights of the present invention.

The present invention's personalized smart blood sugar care monitoring system and method through measuring biometric data, and applications using the same can be used in various personalized smart blood sugar care monitoring systems, methods, and applications using the same.

1: Personalized smart blood sugar care monitoring system through biometric data measurement
11: Prediction factor information measurement unit
13: Data inference unit
15: Data learning unit
17: Correction information processing unit
111: light emitting module
113: Light receiving module
115: Wavelength band calculation module
131: Multi-modal data acquisition module
133: Model optimization module
135: Inferred blood sugar information acquisition module
171: Anti-interference module
173: Module for determining whether to obtain inferred blood sugar information
175: Blood sugar calibration module

Claims (7)

A prediction factor information measuring unit that emits light in a preset wavelength band to the user's body and receives reflected or projected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level;
Based on artificial neural network technology, real-time model optimization is performed using the prediction factor information and multi-modal data, and data inference is performed to infer blood sugar and blood sugar change values using the optimized model and obtain the resulting inferred blood sugar information. wealth; and
A data learning unit that performs retraining of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information,
The predictive factor information measuring unit,
a light emitting module that emits light in the preset wavelength band to the user's body;
a light receiving module that receives the reflected light using at least one photodiode; and
It includes a wavelength band calculation module that adjusts the preset wavelength band to calculate a customized wavelength band for each user,
The wavelength band calculation module,
A regional blood sugar prediction value acquisition step of selecting at least two wavelengths within each of the first and second wavelength bands as input values and obtaining a regional blood sugar prediction value using the input values;
Obtain the actual blood sugar measurement value using the user terminal, compare the actual blood sugar measurement value with the region blood sugar prediction value, and select the blood sugar prediction value with the smallest error among the plurality of region blood sugar prediction values and the area containing the corresponding blood sugar prediction value as a customized area. selecting a custom area of your choice; and
Tuning is performed a preset number of times within the customized area to obtain an optimal wavelength value, and the tuning obtains a blood sugar prediction value for each number of times to obtain a number of actual blood sugar prediction values corresponding to the preset number of times, and the actual blood sugar measurement is performed. Obtaining an optimal wavelength value by determining the optimal wavelength value using a value; calculating a customized wavelength band for each user,
Personalized smart blood sugar care monitoring system through biometric data measurement. delete According to paragraph 1,
At least two light emitting modules are formed, each including an LED or a laser diode, and the preset wavelength band is a short wavelength infrared (SWIR) band, including a first wavelength band and a second wavelength band, Personalized smart blood sugar care monitoring system through data measurement. According to paragraph 3,
The data inference unit,
a multi-modal data acquisition module that acquires the multi-modal data;
a model optimization module that adopts an optimized model structure among pre-learned models, performs model labeling, performs the real-time model optimization, and obtains the optimized model; and
an inferred blood sugar information acquisition module that acquires the inferred blood sugar information using the optimized model; Personalized smart blood sugar care monitoring system through biometric data measurement, including. According to clause 4,
It further includes a correction information processing unit configured to generate or process correction information necessary to obtain the inferred blood sugar information,
The correction information processing unit,
An anti-interference module that is connected to a controllable IR signal output device and separates the output time section of the SWIR signal from other IR signals to prevent interference;
a determination module for determining whether to obtain inferred blood sugar information by acquiring at least one of the user's body temperature, movement information, blood sugar, and ambient noise; and determining whether to obtain the inferred blood sugar information. and
A blood sugar calibration module that receives blood sugar measurement information from the user and performs blood sugar calibration using the blood sugar measurement information. A personalized smart blood sugar care monitoring system through biometric data measurement. Prediction factor information measurement that uses a prediction factor information measurement unit to emit light in a preset wavelength band to the user's body and receive reflected or projected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level. step;
Through the data inference unit, real-time model optimization is performed using the prediction factor information and multi-modal data based on artificial neural network technology, and inference of blood sugar and blood sugar change values is performed using the optimized model, and the resulting inferred blood sugar information is provided. Obtaining data inference step; and
A data learning step in which a data learning unit performs retraining of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information,
The prediction factor information measurement step is,
A light emitting step of emitting light in the preset wavelength band to the user's body;
A light receiving step of receiving the reflected light using at least one photodiode; and
It includes a wavelength band calculation step of calculating a customized wavelength band for each user by adjusting the preset wavelength band,
The wavelength band calculation step is,
A regional blood sugar prediction value acquisition step of selecting at least two wavelengths within each of the first and second wavelength bands as input values and obtaining a regional blood sugar prediction value using the input values;
Obtain the actual blood sugar measurement value using the user terminal, compare the actual blood sugar measurement value with the region blood sugar prediction value, and select the blood sugar prediction value with the smallest error among the plurality of region blood sugar prediction values and the area containing the corresponding blood sugar prediction value as a customized area. selecting a custom area of your choice; and
Tuning is performed a preset number of times within the customized area to obtain an optimal wavelength value, and the tuning obtains a blood sugar prediction value for each number of times to obtain a number of actual blood sugar prediction values corresponding to the preset number of times, and the actual blood sugar measurement is performed. Obtaining an optimal wavelength value by determining the optimal wavelength value using a value; calculating a customized wavelength band for each user,
Personalized smart blood sugar care monitoring method through biometric data measurement. A prediction factor information measurement step of emitting light in a preset wavelength band to the user's body and receiving reflected or projected light reflected from the user's body to obtain prediction factor information for predicting the user's blood sugar level;
Based on artificial neural network technology, real-time model optimization is performed using the prediction factor information and multi-modal data, and data inference is performed to infer blood sugar and blood sugar change values using the optimized model and obtain the resulting inferred blood sugar information. step; and
A data learning step of performing retraining of a basic model that can be used for real-time model optimization using the prediction factor information, the multi-modal data, and the inferred blood sugar information; stored in a storage medium of a digital device to perform;
The prediction factor information measurement step is,
A light emitting step of emitting light in the preset wavelength band to the user's body;
A light receiving step of receiving the reflected light using at least one photodiode; and
It includes a wavelength band calculation step of calculating a customized wavelength band for each user by adjusting the preset wavelength band,
The wavelength band calculation step is,
A regional blood sugar prediction value acquisition step of selecting at least two wavelengths within each of the first and second wavelength bands as input values and obtaining a regional blood sugar prediction value using the input values;
An actual blood sugar measurement value is obtained using a user terminal, and a comparison is performed between the actual blood sugar measurement value and the region blood sugar prediction value. The blood sugar prediction value with the smallest error among the plurality of region blood sugar prediction values and the area containing the corresponding blood sugar prediction value are selected as a customized area. selecting a custom area of your choice; and
Tuning is performed a preset number of times within the customized area to obtain an optimal wavelength value, and the tuning obtains a blood sugar prediction value for each number of times to obtain a number of actual blood sugar prediction values corresponding to the preset number of times, and the actual blood sugar measurement is performed. A step of obtaining an optimal wavelength value of determining the optimal wavelength value using a value; performing a step of calculating a customized wavelength band for each user,
An application stored on the storage medium of a digital device.