CN111221947B - Multi-round dialogue realization method and device of ophthalmologic pre-consultation device - Google Patents
Multi-round dialogue realization method and device of ophthalmologic pre-consultation device Download PDFInfo
- Publication number
- CN111221947B CN111221947B CN201911365885.4A CN201911365885A CN111221947B CN 111221947 B CN111221947 B CN 111221947B CN 201911365885 A CN201911365885 A CN 201911365885A CN 111221947 B CN111221947 B CN 111221947B
- Authority
- CN
- China
- Prior art keywords
- information
- patient
- model
- dialogue
- history information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 claims abstract description 56
- 201000010099 disease Diseases 0.000 claims abstract description 55
- 208000024891 symptom Diseases 0.000 claims description 46
- 238000012549 training Methods 0.000 claims description 24
- 238000013507 mapping Methods 0.000 claims description 5
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000013461 design Methods 0.000 abstract description 25
- 238000002372 labelling Methods 0.000 abstract description 11
- 230000008569 process Effects 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000008961 swelling Effects 0.000 description 4
- 210000004556 brain Anatomy 0.000 description 3
- 230000004456 color vision Effects 0.000 description 3
- 206010061818 Disease progression Diseases 0.000 description 2
- 206010061218 Inflammation Diseases 0.000 description 2
- 206010027146 Melanoderma Diseases 0.000 description 2
- 230000005856 abnormality Effects 0.000 description 2
- 210000004087 cornea Anatomy 0.000 description 2
- 230000005750 disease progression Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 230000004054 inflammatory process Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 201000004614 iritis Diseases 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 208000035824 paresthesia Diseases 0.000 description 2
- 230000037380 skin damage Effects 0.000 description 2
- 208000029257 vision disease Diseases 0.000 description 2
- 230000000007 visual effect Effects 0.000 description 2
- 208000035473 Communicable disease Diseases 0.000 description 1
- 206010011033 Corneal oedema Diseases 0.000 description 1
- 206010027145 Melanocytic naevus Diseases 0.000 description 1
- 208000007256 Nevus Diseases 0.000 description 1
- 206010030113 Oedema Diseases 0.000 description 1
- 206010047642 Vitiligo Diseases 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 201000004778 corneal edema Diseases 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000037213 diet Effects 0.000 description 1
- 235000005911 diet Nutrition 0.000 description 1
- 208000035475 disorder Diseases 0.000 description 1
- 210000003128 head Anatomy 0.000 description 1
- 230000001788 irregular Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000474 nursing effect Effects 0.000 description 1
- 230000008506 pathogenesis Effects 0.000 description 1
- 230000007918 pathogenicity Effects 0.000 description 1
- 230000035790 physiological processes and functions Effects 0.000 description 1
- 239000000049 pigment Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000003813 thin hair Effects 0.000 description 1
- 230000002087 whitening effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/332—Query formulation
- G06F16/3329—Natural language query formulation or dialogue systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Artificial Intelligence (AREA)
- Human Computer Interaction (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Eye Examination Apparatus (AREA)
Abstract
The application discloses a multi-round dialogue realization method and device of an ophthalmic pre-consultation device, wherein the method comprises the following steps: basic information, current medical history information and past history information of a patient are sequentially collected; predicting disease information of the patient based on the base information, the current medical history information, the past history information, and the ophthalmic medical knowledge base; collecting the inquiry information of the patient, generating answering information according to the inquiry information and the user spoken language expression table, and generating an information list. The method adopts the whole serial and partial parallel dialogue design flow, thereby better meeting the scene requirement, greatly improving the flow realization speed and reducing the labeling cost.
Description
Technical Field
The application relates to the technical field of voice recognition, in particular to a multi-round dialogue realization method and device of an ophthalmic pre-consultation device.
Background
In the related art, there are two main ways to implement multi-round dialogue at present:
1. task type multi-round dialogue based on NLU DM NLG
The task type robot core module mainly comprises three modules: natural language understanding module (NLU) dialogue management module natural language generation module (NLG). Where natural language understanding and dialog management are the core. The natural language understanding functions to understand the user utterance, and the meaning of the understanding is to parse the user utterance according to different semantic representations. When the user language passes through the natural language understanding module, the user language is required to pass through the field recognition, the user intention recognition and the slot position extraction. Domain identification, namely, whether the sentence belongs to the task scene or not is identified, and when a plurality of robots are integrated, such as a boring robot, a question-answering robot and the like, the domain identification should be judged and distributed before entering the task robot; the intention recognition, namely, the user intention recognition, subdivides the sub-scene under the task scene; entity identification and slot filling for input to the dialog management module. The dialogue management module is equivalent to the brain of the task robot, and is divided into a dialogue state maintenance DST+ action candidate ranking Policy. These two parts form a multi-round conversational experience between humans. DM is largely divided into two functions, one part is recording user history utterances and the other part is generating system decisions. The triplet output of the natural language understanding module will be the input to the dialog management system. The state tracking module includes various information for continuing the session, and updates the current session state based on the old state, the user state (i.e., the triples described above), and the system state (i.e., by querying the database). The dialogue strategy is closely related to the task scene, and is usually used as an output of a dialogue management module, such as a back-question strategy for the missing slots in the scene.
2. End-to-end method for constructing dialogue model
Another technical solution for implementing a medical consultation session is to use a model method of end-to-end generation (Seq 2 Seq). Such as: an end-to-end task type dialogue system comprises a data preprocessing module, a named body extraction module, a compiling module, a dialogue history encoder module and a decoding output module. While such end-to-end dialog systems may reduce cumbersome manual rules and reduce the amount of data for training models, it is difficult to deal with the dialog logic of medical triage. The adoption of the end-to-end system can introduce certain uncertainty and unexplainability, so that the dialogue of each round cannot strictly follow the previously designed flow, and the problems of logic disorder or repeated inquiry and the like are caused.
Disclosure of Invention
The present application aims to solve at least one of the technical problems in the related art to some extent.
Therefore, the application aims to provide a multi-round dialogue implementation method of an ophthalmic pre-consultation device, which adopts a whole serial and partial parallel dialogue design flow, thereby better meeting scene requirements, greatly improving the flow implementation speed and reducing the labeling cost.
Another object of the present application is to propose a multi-turn dialogue implementation device of an ophthalmic pre-consultation device.
In order to achieve the above objective, an embodiment of an aspect of the present application provides a method for implementing multiple rounds of dialogue in an ophthalmic pre-consultation device, including the following steps: basic information, current medical history information and past history information of a patient are sequentially collected; predicting disease information of the patient based on the base information, the current medical history information, the past history information, and an ophthalmic medical knowledge base; collecting the inquiring information of the patient, generating answering information according to the inquiring information and a user spoken language expression table, and generating an information list.
According to the multi-round dialogue implementation method of the ophthalmic pre-consultation device, a set of dialogue design flow aiming at medical pre-consultation is provided by fusing a model and a template decision scheme, and a set of model data design specifications of medical pre-consultation are designed from the dialogue flow to training data of each sub-model module, so that the ophthalmic pre-consultation device adopts an integral serial and partial parallel dialogue design flow, and the system is enabled to realize more efficient and better suitability for medical pre-consultation tasks; the data format of each part is designed aiming at the whole set of inquiry flow, and the complete system presents various designs required by the whole set of ophthalmic inquiry dialogue; the training data generation method is innovatively introduced, so that training of the algorithm module can be completed under limited labeling data, and applicability and high efficiency are achieved.
In addition, the multi-round dialogue implementing method of the ophthalmic pre-consultation device according to the above embodiment of the present application may further have the following additional technical features:
further, in an embodiment of the present application, the generating answering information according to the overt information and the spoken utterance of the user includes: obtaining a query corpus from the query information; inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences; and determining the answering information according to the SQL query statement.
Further, in one embodiment of the present application, the method further includes: matching an ending keyword from the overtaking information; if the matching is successful, the dialogue flow is ended.
Further, in one embodiment of the present application, the predicting the disease information of the patient based on the basic information, the current medical history information, the past history information, and the ophthalmic medical knowledge base includes: all intents are acquired, and for each intention, the intent is selected from the pre-written slot values and filled into the data template.
Further, in one embodiment of the present application, the predicting the disease information of the patient based on the basic information, the current medical history information, the past history information, and the ophthalmic medical knowledge base includes: and adopting a Rank model to match the information with the content in the ophthalmic medical knowledge base.
To achieve the above object, another embodiment of the present application provides a multi-round dialogue implementation device of an ophthalmic pre-inquiry device, including: the acquisition module is used for sequentially acquiring basic information, current medical history information and past history information of a patient; a prediction module for predicting disease information of the patient based on the base information, the current medical history information, the past history information, and an ophthalmic medical knowledge base; the generating module is used for collecting the inquiring information of the patient, generating answering information according to the inquiring information and the user spoken language expression table, and generating an information list.
According to the multi-round dialogue implementation device of the ophthalmic pre-consultation device, a set of dialogue design flow aiming at medical pre-consultation is provided by fusing a model and a template decision scheme, and a set of model data design specifications of medical pre-consultation are designed from the dialogue flow to training data of each sub-model module, so that the ophthalmic pre-consultation device adopts an integral serial and partial parallel dialogue design flow, and the system is enabled to realize more efficient and better suitability for medical pre-consultation tasks; the data format of each part is designed aiming at the whole set of inquiry flow, and the complete system presents various designs required by the whole set of ophthalmic inquiry dialogue; the training data generation method is innovatively introduced, so that training of the algorithm module can be completed under limited labeling data, and applicability and high efficiency are achieved.
In addition, the multi-turn dialogue realizing device of the ophthalmic pre-consultation device according to the above embodiment of the present application may further have the following additional technical features:
further, in an embodiment of the present application, the generating module is further configured to obtain a query corpus from the query information; inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences; and determining the answering information according to the SQL query statement.
Further, in one embodiment of the present application, the method further includes: and the matching module is used for matching the ending keywords from the inquiry information and ending the dialogue flow when the matching is successful.
Further, in an embodiment of the present application, the prediction module is further configured to obtain all intents, and select and fill in the data template from the pre-written slot values for each intention.
Further, in an embodiment of the present application, the prediction module is further configured to match information with content in the ophthalmic medical knowledge base using a Rank model.
Additional aspects and advantages of the application will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the application.
Drawings
The foregoing and/or additional aspects and advantages of the application will become apparent and readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings, in which:
FIG. 1 is a flow chart of a method for implementing a multi-round dialogue of an ophthalmic pre-consultation device according to an embodiment of the present application;
FIG. 2 is a flow chart of a consultation according to an embodiment of the present application;
FIG. 3 is a flow chart of disease prediction according to an embodiment of the present application;
fig. 4 is a schematic structural diagram of a multi-turn dialogue implementing device of an ophthalmic pre-consultation device according to an embodiment of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative and intended to explain the present application and should not be construed as limiting the application.
The present application has been made based on the knowledge and findings of the inventors of the following problems:
the ophthalmologic pre-consultation device achieves the aim of collecting the data before the patient consultation through multi-round dialogue interaction with the user. The device needs to design a set of integral multi-round inquiry modes, including basic information inquiry, main complaint inquiry, disease history inquiry, disease development inquiry, user answer and the like. These processes place high demands on model design and training data construction.
In order to solve the problems that a single dialogue model is difficult to process complex interaction in a pre-consultation process and the like, the embodiment of the application provides a dialogue design flow which is wholly serial and partially parallel. The whole dialogue logic is used for processing basic information, main complaint information, physiological state, medical history information, disease development, disease prediction and related inquiry, user answering and information list generation. The method can effectively collect the information of the complete user and control the conversation process. Meanwhile, in order to ensure the conversation diversity, a parallel module is used in each independent part for process management, for example, in a disease development module, the method can determine the aspect to query according to the context, so that the conversation process diversity is improved.
Meanwhile, the labeling data of the training model is a great cost in the actual landing project. In order to more effectively utilize the existing labeling data, the embodiment of the application designs a set of data formats aiming at the ophthalmic pre-consultation problem, thereby completing model training with minimum labeling cost. The method is divided into the following three parts: 1) Sequence annotation data; 2) Short text similarity data; 3) NL2SQL data. The implementation will be described in detail below.
Aiming at two problems of difficult flow design and difficult label data acquisition in an ophthalmic pre-consultation device, the embodiment of the application provides a method and a device for realizing multi-round dialogue of the ophthalmic pre-consultation device, which greatly improve the flow realization speed and reduce the labeling cost, and can be used for ophthalmic consultation and easily migrate to other medical consultation scenes after some simple deletion.
The following describes a multi-round dialogue implementation method and apparatus for an ophthalmic pre-consultation apparatus according to an embodiment of the present application with reference to the accompanying drawings, and first describes a multi-round dialogue implementation method for an ophthalmic pre-consultation apparatus according to an embodiment of the present application with reference to the accompanying drawings.
Fig. 1 is a flowchart of a multi-round dialogue implementation method of an ophthalmic pre-consultation device according to an embodiment of the present application.
As shown in fig. 1, the method for implementing multiple rounds of dialogue of the ophthalmic pre-consultation device includes the following steps:
in step S101, basic information, current medical history information, and past history information of the patient are sequentially acquired.
It can be appreciated that based on the collection and arrangement of the common ophthalmic inquiry records, the inquiry dialogue is found to have the characteristic of fixed topic flow. As shown in FIG. 2, the inquiry flow is mainly divided into five modules, namely basic information, current medical history, past history, disease prediction and patient inquiry, in which the inquiry is ordered, and the sub-modules of specific inquiry are often irregular in order. Therefore, the embodiment of the application adopts a mode of parallel-in-serial and parallel-in of the outer string to carry out the man-machine inquiry dialogue, namely the whole flow is orderly, and the inquiry sequence is determined in the module according to the information loss condition provided by the user.
In step S102, disease information of the patient is predicted based on the basic information, the current medical history information, the past medical history information, and the ophthalmic medical knowledge base.
It will be appreciated that for the expertise and complexity of ophthalmic medical data, embodiments of the present application construct an ophthalmic medical knowledge base divided into 11 fields and corresponding values, as shown in table 1, where table 1 is an ophthalmic disease knowledge base.
TABLE 1
In addition, the embodiment of the application performs entity matching data of related symptoms, wherein the related symptoms are important points for predicting diseases. To extract relevant symptoms from the complaints and stages of development entered by the user, embodiments of the present application label the common user spoken language expression for each symptom against 509 eye condition symptoms in the knowledge base. As shown in table 2, wherein table 2 is a spoken language expression of the relevant symptoms.
TABLE 2
Further, in one embodiment of the present application, predicting disease information for a patient based on the base information, the present medical history information, the past history information, and the ophthalmic medical knowledge base includes: acquiring all intents, selecting from the pre-written slot values for each intention, and filling the slot values into a data template; and (3) adopting a Rank model to match the information with the content in the ophthalmic medical knowledge base.
It can be appreciated that in order to meet the requirements of the dialogue system on the algorithm function, the named entity recognition (NLU), the phrase matching model and the NL2SQL model are designed as main functions of the algorithm engine according to the embodiment of the present application. Because the obtaining cost of the marking data is too high, the training data is generated in a mode based on an external knowledge template, and the training cost is greatly reduced. The NLU model training data and the phrase matching model will be described in detail below, and the NL2SQL model will be described later.
Specifically, (1) NLU model training data
The NLU model is responsible for extracting the required slot values from the user's input, while training an entity recognition model requires a large amount of sequence annotation data. In order to reduce the cost of manually marking data, the embodiment of the application adopts a data generation mode based on a data template to train data, as shown in a table 3, and the table 3 is an NLU model training data table. Wherein for each intention, the padding into the data template (here denoted by% for the padding position) is selected from the pre-written slot values.
TABLE 3 Table 3
(2) Phrase matching model
Because the user does not express the name of the disease or the related symptoms very accurately, a phrase matching model is required to match the user expression with the content in the content library. The embodiment of the application adopts a Rank model, and uses a Pairwise mode to make positive and negative example samples for similarity calculation. And taking each user expression as a keyword, taking the corresponding related symptom as a positive example, randomly selecting one related symptom as a negative example, and making a triplet to train a phrase matching model, wherein as shown in a table 4, the table 4 is a phrase matching model table.
TABLE 4 Table 4
| Symptoms associated with | Expression 1 | Expression 2 | Expression 3 | Expression 4 | Expression 5 |
| Corneal edema | Swelling of cornea | Swelling of eyes | Corneal swelling | Inflammation of the cornea | Oedema of eyes |
| Pigment skin damage | Vitiligo | Bai Seban | White spot | White skin damage | Skin whitening |
| Blind spot | Invisible point | Object missing points | Dead zone | Point of visual field loss | Blind spot of visual field |
| Black mole | Black nevus | Black dot | Black plaque | Black spot | Black spot |
| Iris inflammation | Inflammation of the eyes | Iris inflammation | Iris redness and swelling | Iris pain | Iris pain |
| Pillow for baby | Less hair | Less hair on the pillow | Thin hair | Bald of brain | Less head of brain |
| Loss of colour vision | Invisible color | No color can be seen | Loss of colour vision | No color | Without colour perception |
Furthermore, based on mapping data of a large number of diseases and related symptoms, the embodiments of the present application can predict by adopting the following two strategies, specifically as follows:
strategy 1: selecting the most specific symptoms
Assuming that the risk of developing each symptom of the disease is significantly different and marked with a distinguishing signature, for example, one symptom may set three scores, 3, 2, 1, from high to low.
Diseases containing symptom(s) are selected as candidates by extracting symptom descriptions in complaints, disease progression, vision disorders, paresthesias, and appearance abnormalities and mapping their corresponding related symptoms.
And selecting the related symptoms with highest pathogenicity probability from the related symptoms of the candidate diseases for inquiry. If the user answers, adding a score corresponding to the disease with the symptom and the disease with the symptom, and removing the disease without the symptom from the candidate set; if not, the disease with the symptom is not filtered out, so that the patient can make wrong judgment if some symptoms are not obvious.
Finally, three diseases are obtained through filtering to be used as prediction results, wherein the higher the score is, the higher the disease probability is.
Strategy 2: selecting the symptom with the highest frequency of occurrence
The probability of pathogenesis of each relevant symptom of the disease is assumed to be approximate.
Diseases containing symptom(s) are selected as candidates by extracting symptom descriptions in complaints, disease progression, vision disorders, paresthesias, and appearance abnormalities and mapping their corresponding related symptoms, including one related symptom plus 1 score.
And selecting and inquiring the symptoms with the largest number of the related symptoms of the candidate diseases. If the user answers, adding 1 point to the disease with the symptom, and removing the disease without the symptom from the candidate set; if not, the disease with the symptom is not filtered out, so that the patient can make wrong judgment if some symptoms are not obvious.
Finally, three diseases are obtained through filtering to be used as prediction results, wherein the higher the score is, the higher the disease probability is.
The first is based on differentiated disease symptom data and the second is applicable to non-differentiated disease symptom data. Therefore, the embodiment of the application can tentatively adopt the second strategy to predict the diseases.
In addition, the flow of disease prediction is shown in fig. 3, wherein the specific pseudo code of the disease prediction strategy algorithm main program is as follows:
traversing the candidate additional symptom dictionary S to find the symptom pmax which has the maximum key value array length and does not belong to userSym for additional
In step S103, the inquiry information of the patient is collected, and answering information and an information list are generated according to the inquiry information and the spoken language expression table of the user.
Further, in one embodiment of the present application, generating answer information according to the challenge information and the spoken language expression of the user includes: obtaining a query corpus from the query information; inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences; and determining answering information according to the SQL query statement.
It will be appreciated that following disease prediction, the user will make relevant queries regarding the predicted disease. The user's query corpus is used to train the NL2SQL model, i.e., the questions are directly converted into SQL query statements, as shown in table 5, table 5 being a patient query procedure table.
TABLE 5
| User challenge field | Corpus sample |
| Disease prediction | Please ask me what is this a disease? |
| Introduction to diseases | If i get the disease, explain this? |
| Nursing method | How do care for? |
| Diet contraindication | What is there to be attention on eating? |
| Pre-examination matters | What is there a need for a prior check? |
| Infectious diseases | Is this disease not transmitted? |
| Etiology/cause | How about it would be sick? |
Furthermore, the last user inquiry link in the system adopts NL2SQL to select the answers of the questions from the database. The embodiment of the application adopts a model based on the Seq2Tree, converts each intention shown in the table 5 into SQL sentences, and trains the model in an end-to-end mode, such as 7 SQL sentence major classes shown in the table 6. After receiving the user input, the model will make a preamble traversal decision from the large class of SQL syntax tree consisting of 7 SQL statements below until going to the root node in turn. Wherein, table 6 is an SQL statement class table.
TABLE 6
Further, in one embodiment of the present application, the method further includes: matching the ending keywords from the inquiry information; if the matching is successful, the dialogue flow is ended.
It will be appreciated that the ending phrase is divided into the end of one query module and the end of the entire dialog. Corpus samples are shown in table 7, for example. Wherein, table 7 is an end keyword matching table.
TABLE 7
| End phrase | Corpus sample |
| Stage end term | Kappy-end-of-a-person; i know |
| Dialogue end term | Without any provision for |
In summary, the multi-round dialogue implementation method of the ophthalmic pre-consultation device provided by the embodiment of the application combines a model and a template decision scheme to provide a set of dialogue design flow aiming at medical pre-consultation, and designs a whole set of model data design specification of medical pre-consultation from the dialogue flow to training data of each sub-model module, so that the ophthalmic pre-consultation device adopts an integral serial and partially parallel dialogue design flow, and the system is enabled to realize more efficient and better suitability for medical consultation tasks; the data format of each part is designed aiming at the whole set of inquiry flow, and the complete system presents various designs required by the whole set of ophthalmic inquiry dialogue; the training data generation method is innovatively introduced, so that training of the algorithm module can be completed under limited labeling data, and applicability and high efficiency are achieved.
Next, a multi-turn dialogue-realization device of an ophthalmic pre-inquiry apparatus according to an embodiment of the present application will be described with reference to the accompanying drawings.
Fig. 4 is a schematic structural view of a multi-turn dialogue implementing device of an ophthalmic pre-consultation device according to an embodiment of the present application.
As shown in fig. 4, the multi-turn dialogue-realization device 10 of the ophthalmic pre-inquiry apparatus includes: the system comprises an acquisition module 100, a prediction module 200 and a generation module 300.
The acquisition module 100 is used for sequentially acquiring basic information, current medical history information and past history information of a patient; the prediction module 200 is used for predicting disease information of a patient according to basic information, current medical history information, past history information and an ophthalmic medical knowledge base; the generating module 300 is used for collecting the inquiry information of the patient, generating answering information according to the inquiry information and the spoken language expression table of the user, and generating an information list. The device 10 of the embodiment of the application adopts the whole serial and partial parallel dialogue design flow, thereby better meeting the scene requirement, greatly improving the flow realization speed and reducing the labeling cost.
Further, in an embodiment of the present application, the generating module is further configured to obtain an additional corpus from the additional information; inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences; and determining answering information according to the SQL query statement.
Further, in one embodiment of the present application, the method further includes: and the matching module is used for matching the ending keywords from the inquiry information and ending the dialogue flow when the matching is successful.
Further, in one embodiment of the present application, the prediction module 200 is further configured to obtain all intents, and select and fill in the data template from the pre-written slot values for each intention.
Further, in one embodiment of the present application, the prediction module 200 is further configured to match the information to the content in the ophthalmic medical knowledge base using a Rank model.
It should be noted that the explanation of the embodiment of the method for implementing multiple rounds of dialogue in the ophthalmic pre-inquiry device is also applicable to the multiple rounds of dialogue implementing device of the ophthalmic pre-inquiry device in this embodiment, and will not be repeated here.
According to the multi-round dialogue implementation device of the ophthalmic pre-consultation device provided by the embodiment of the application, a set of dialogue design flow aiming at medical pre-consultation is provided by fusing a model and a template decision scheme, and a whole set of model data design specification of medical pre-consultation is designed from the dialogue flow to training data of each sub-model module, so that the ophthalmic pre-consultation device adopts an integral serial and partial parallel dialogue design flow, and the system is enabled to realize higher efficiency and better suitability for medical consultation tasks; the data format of each part is designed aiming at the whole set of inquiry flow, and the complete system presents various designs required by the whole set of ophthalmic inquiry dialogue; the training data generation method is innovatively introduced, so that training of the algorithm module can be completed under limited labeling data, and applicability and high efficiency are achieved.
Furthermore, the terms "first," "second," and the like, are used for descriptive purposes only and are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include at least one such feature. In the description of the present application, the meaning of "plurality" means at least two, for example, two, three, etc., unless specifically defined otherwise.
In the present application, unless expressly stated or limited otherwise, a first feature "up" or "down" a second feature may be the first and second features in direct contact, or the first and second features in indirect contact via an intervening medium. Moreover, a first feature being "above," "over" and "on" a second feature may be a first feature being directly above or obliquely above the second feature, or simply indicating that the first feature is level higher than the second feature. The first feature being "under", "below" and "beneath" the second feature may be the first feature being directly under or obliquely below the second feature, or simply indicating that the first feature is less level than the second feature.
In the description of the present specification, a description referring to terms "one embodiment," "some embodiments," "examples," "specific examples," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms are not necessarily directed to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, the different embodiments or examples described in this specification and the features of the different embodiments or examples may be combined and combined by those skilled in the art without contradiction.
While embodiments of the present application have been shown and described above, it will be understood that the above embodiments are illustrative and not to be construed as limiting the application, and that variations, modifications, alternatives and variations may be made to the above embodiments by one of ordinary skill in the art within the scope of the application.
Claims (4)
1. The multi-round dialogue implementation method of the ophthalmic pre-consultation device is characterized by comprising the following steps of:
basic information, current medical history information and past history information of a patient are sequentially collected;
predicting disease information of the patient based on the base information, the current medical history information, the past history information, and an ophthalmic medical knowledge base; and
collecting the inquiring information of the patient, generating answering information according to the inquiring information and a user spoken language expression table, and generating an information list;
and generating answering information according to the overtaking information and the user spoken language expression table, wherein the method comprises the following steps:
obtaining a query corpus from the query information;
inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences;
determining the answering information according to the SQL query statement;
the predicting disease information of the patient based on the base information, the current medical history information, the past history information, and an ophthalmic medical knowledge base includes:
acquiring all intents, selecting from the pre-written slot values for each intention, and filling the slot values into a data template;
adopting a Rank model, matching the information with the content in the ophthalmic medical knowledge base to form short sentences, and performing similarity calculation by using a Pairwise mode to form positive and negative example samples when the short sentences are matched;
the method further comprises the steps of:
extracting a required slot value from the input of a user by using an NLU model, wherein training data of the NLU model is generated by adopting a data generation mode based on a data template;
each user expression is used as a keyword, the corresponding related symptom is used as a positive example, one related symptom is randomly selected as a negative example, and a triplet is made to train a phrase matching model;
disease information for the patient is predicted based on mapping data for a number of diseases and associated symptoms.
2. The method as recited in claim 1, further comprising:
matching an ending keyword from the overtaking information;
if the matching is successful, the dialogue flow is ended.
3. A multi-round dialogue realizing device of an ophthalmologic pre-consultation device, comprising:
the acquisition module is used for sequentially acquiring basic information, current medical history information and past history information of a patient;
a prediction module for predicting disease information of the patient based on the base information, the current medical history information, the past history information, and an ophthalmic medical knowledge base; and
the generating module is used for collecting the inquiring information of the patient, generating answering information according to the inquiring information and the user spoken language expression table, and generating an information list;
the generation module is further used for obtaining an additional corpus from the additional information; inputting the additional corpus into a trained NL2SQL model to directly convert the questions into SQL query sentences; determining the answering information according to the SQL query statement;
the prediction module is further used for acquiring all intents, selecting from the pre-written slot values for each intention and filling the slot values into the data template;
the prediction module is further used for matching the information with the content in the ophthalmic medical knowledge base by adopting a Rank model;
the prediction module is further used for extracting a required slot value from the input of a user by using an NLU model, wherein training data of the NLU model is generated by adopting a data generation mode based on a data template;
the prediction module is further used for randomly selecting one relevant symptom as a negative example by taking each user expression as a keyword and the corresponding relevant symptom as a positive example, and forming a triplet to train a phrase matching model;
the prediction module is further to predict disease information for the patient based on mapping data for a plurality of diseases and associated symptoms.
4. A device according to claim 3, further comprising:
and the matching module is used for matching the ending keywords from the inquiry information and ending the dialogue flow when the matching is successful.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911365885.4A CN111221947B (en) | 2019-12-26 | 2019-12-26 | Multi-round dialogue realization method and device of ophthalmologic pre-consultation device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911365885.4A CN111221947B (en) | 2019-12-26 | 2019-12-26 | Multi-round dialogue realization method and device of ophthalmologic pre-consultation device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111221947A CN111221947A (en) | 2020-06-02 |
| CN111221947B true CN111221947B (en) | 2023-11-24 |
Family
ID=70825974
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911365885.4A Active CN111221947B (en) | 2019-12-26 | 2019-12-26 | Multi-round dialogue realization method and device of ophthalmologic pre-consultation device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111221947B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230169074A1 (en) * | 2021-11-29 | 2023-06-01 | 42Maru Inc. | Device and method for converting natural language query into sql query |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112948599A (en) * | 2021-05-13 | 2021-06-11 | 北京富通东方科技有限公司 | Disease diagnosis method based on path reasoning of medical knowledge graph |
| CN114300160B (en) * | 2021-11-16 | 2022-10-18 | 北京左医科技有限公司 | Inquiry dialogue method and system |
| CN117093679B (en) * | 2023-06-19 | 2024-04-02 | 无码科技（杭州）有限公司 | Large language model intelligent inquiry dialogue method, system, equipment and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008293294A (en) * | 2007-05-24 | 2008-12-04 | Fuji Xerox Co Ltd | Medical information processing system, information processing method, and computer program |
| CN105912846A (en) * | 2016-04-07 | 2016-08-31 | 南京小网科技有限责任公司 | Intelligent medical aid decision making system on basis of cloud computing technique and medical knowledge base technique |
| CN107247868A (en) * | 2017-05-18 | 2017-10-13 | 深思考人工智能机器人科技（北京）有限公司 | A kind of artificial intelligence aids in interrogation system |
| CN109065172A (en) * | 2018-07-04 | 2018-12-21 | 平安科技（深圳）有限公司 | Illness information acquisition method, device, computer equipment and storage medium |
| CN109408526A (en) * | 2018-10-12 | 2019-03-01 | 平安科技（深圳）有限公司 | SQL statement generation method, device, computer equipment and storage medium |
| CN109522393A (en) * | 2018-10-11 | 2019-03-26 | 平安科技（深圳）有限公司 | Intelligent answer method, apparatus, computer equipment and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20060136142A1 (en) * | 2000-11-02 | 2006-06-22 | Kurt Berlin | Systems, methods and computer program products for guiding selection of a therapeutic treatment regimen based on the methylation status of the DNA |
-
2019
- 2019-12-26 CN CN201911365885.4A patent/CN111221947B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2008293294A (en) * | 2007-05-24 | 2008-12-04 | Fuji Xerox Co Ltd | Medical information processing system, information processing method, and computer program |
| CN105912846A (en) * | 2016-04-07 | 2016-08-31 | 南京小网科技有限责任公司 | Intelligent medical aid decision making system on basis of cloud computing technique and medical knowledge base technique |
| CN107247868A (en) * | 2017-05-18 | 2017-10-13 | 深思考人工智能机器人科技（北京）有限公司 | A kind of artificial intelligence aids in interrogation system |
| CN109065172A (en) * | 2018-07-04 | 2018-12-21 | 平安科技（深圳）有限公司 | Illness information acquisition method, device, computer equipment and storage medium |
| CN109522393A (en) * | 2018-10-11 | 2019-03-26 | 平安科技（深圳）有限公司 | Intelligent answer method, apparatus, computer equipment and storage medium |
| CN109408526A (en) * | 2018-10-12 | 2019-03-01 | 平安科技（深圳）有限公司 | SQL statement generation method, device, computer equipment and storage medium |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20230169074A1 (en) * | 2021-11-29 | 2023-06-01 | 42Maru Inc. | Device and method for converting natural language query into sql query |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111221947A (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111221947B (en) | Multi-round dialogue realization method and device of ophthalmologic pre-consultation device | |
| CN112559556B (en) | Language model pre-training method and system for table mode analysis and sequence mask | |
| CN111933127B (en) | Intention recognition method and intention recognition system with self-learning capability | |
| CN107577826A (en) | Classification of diseases coding method and system based on raw diagnostic data | |
| CN105589844A (en) | Missing semantic supplementing method for multi-round question-answering system | |
| CN110059169B (en) | Intelligent robot chat context implementation method and system based on corpus labeling | |
| CN109271506A (en) | A kind of construction method of the field of power communication knowledge mapping question answering system based on deep learning | |
| CN107705839A (en) | Disease automatic coding and system | |
| CN106357942A (en) | Intelligent response method and system based on context dialogue semantic recognition | |
| CN110175229B (en) | Method and system for on-line training based on natural language | |
| US20230394247A1 (en) | Human-machine collaborative conversation interaction system and method | |
| CN105631468A (en) | RNN-based automatic picture description generation method | |
| CN115269857A (en) | Knowledge graph construction method and device based on document relation extraction | |
| CN108345686A (en) | A kind of data analysing method and system based on search engine technique | |
| CN107038154A (en) | A kind of text emotion recognition methods and device | |
| CN111026884B (en) | Dialog corpus generation method for improving quality and diversity of man-machine interaction dialog corpus | |
| CN109460459A (en) | A kind of conversational system automatic optimization method based on log study | |
| CN109033073B (en) | Text inclusion recognition method and device based on vocabulary dependency triple | |
| CN110687999A (en) | Method and device for semantic processing of electroencephalogram signals | |
| CN111428104A (en) | Epilepsy auxiliary medical intelligent question-answering method based on viewpoint type reading understanding | |
| CN110347811A (en) | A kind of professional knowledge question and answer robot system based on artificial intelligence | |
| CN110895936B (en) | Voice processing method and device based on household appliance | |
| CN114155962A (en) | Data cleaning method and method for constructing disease diagnosis by using knowledge graph | |
| CN117370525B (en) | Intelligent diagnosis guiding method based on fine tuning large model | |
| CN115204187A (en) | Auxiliary method and system for excavation of conversational process based on real person conversational corpus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |