Residential physician standardized training examination system based on knowledge graph and knowledge graph

Abstract

The invention discloses an in-patient doctor standardized training examination system based on a knowledge graph and the knowledge graph, and aims at simulating the flow and the examination range of the in-patient doctor standardized training examination to meet the relevant examination requirements, and the system comprises two subsystems of an intelligent medical question and answer system and an examination management system, the front end of the intelligent medical question-answering system uses a WeChat applet framework to realize intelligent medical question-answering and examination functions, and the rear end module comprises a knowledge graph module, an intelligent question-answering system module, an inquiry module responding to doctors, a medical science popularization question module responding to users and the like. The examination management system has the functions of user answering, answer submitting and the like; the test result analysis comprises answer result display and question analysis. According to the invention, a knowledge basis is provided for the standardized training examination, and the actual requirements of the standardized training examination of the resident physician can be well met.

Description

Technical Field

[0001] The present invention relates to the field of Internet diagnosis and treatment, and in particular to a standardized physician training and examination system based on a medical knowledge graph and a corresponding knowledge graph. Background Art

[0002] Question answering system is an important application in the field of natural language processing. Its core task is to find and provide corresponding answers from structured or semi-structured data based on the questions raised by users. In recent years, question answering systems have been widely used in many fields, including intelligent customer service, knowledge base question answering, and search engines. In the medical field, the research and application of question answering systems are also increasing. Existing question answering systems can be divided into rule-based, statistical-based, knowledge graph-based, and deep learning-based. In the medical field, knowledge graph-based question answering systems have attracted much attention. These systems partially solve the problem of inaccurate answers by building a medical knowledge base to obtain entity and relationship information. In the field of Chinese medicine, the combination of BiLSTM-CRF, BERT, and IDCNN technologies has achieved remarkable results in better capturing global context information. The introduction of these innovative methods has greatly improved the accuracy and reliability of question answering systems in the medical field.

[0003] However, the existing medical question-and-answer system does not achieve effective integration of relevant medical knowledge, but only provides answers to users in the form of questions and answers. The standardized training examination process for resident physicians is different from the question-and-answer system and requires multiple sub-questions. Ordinary question-and-answer systems cannot meet the actual needs of the standardized training examination for resident physicians.

[0004] Therefore, technical personnel in this field urgently need to develop a standardized training examination system for resident doctors to help doctors find gaps and improve their diagnostic level. Summary of the invention

[0005] The technical problem to be solved by the present invention is to develop a standardized training and examination system for resident physicians based on a knowledge graph, integrating medical Q&A, medical testing and resident physician training, providing users with an easy-to-operate standardized training and examination system for resident physicians, helping physicians to find gaps and improve their diagnostic level.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] A standardized training and examination system for resident doctors based on knowledge graph, including an intelligent medical question-and-answer system and an examination management system. The front end of the intelligent medical question-and-answer system uses the WeChat applet framework to implement the functions of intelligent medical question-and-answer and examination, including a user login module, an intelligent consultation and intelligent question-and-answer module, a medical examination module, a test result analysis module, a user personal information module, and a user message module;

[0008] The intelligent consultation and intelligent question-answering modules include knowledge graphs, intelligent question-answering systems, responses to doctors’ questions, and users’ medical science questions; the medical examination module includes functions such as user answering questions and answer submission; the test result analysis includes answer result display and question analysis;

[0009] The WeChat applet front-end mainly uses wx.request to communicate with the back-end interface and send HTTP requests to the back-end. These requests usually include user operations, such as clicking buttons, filling out forms, etc., and call the corresponding back-end interface to complete the communication task with the back-end.

[0010] The backend of the intelligent medical question-and-answer examination system is mainly responsible for processing front-end requests and implementing the question-and-answer system based on the knowledge graph, and uses the B / S architecture to complete front-end and back-end communication.

[0011] In one embodiment, the intelligent medical question-answering system specifically includes the following steps:

[0012] S1 user registration and login, used to detect user login information; for first-time users, user registration function is provided;

[0013] S2 intelligent consultation and intelligent question and answer are responsible for responding to doctors' consultation and users' medical science questions respectively; preferably, S2 specifically includes: building a knowledge graph, building an intelligent question and answer system, responding to doctors' consultation and users' medical science questions, etc.

[0014] S3 medical examination, which is used to provide physicians with an examination interface to assess their thyroid disease diagnosis;

[0015] S4 test result analysis, used to display the test score after submitting the test paper, total score, user answer, standard answer and analysis, etc.;

[0016] S5 User personal information is used to display the personal information of the current user.

[0017] In one embodiment, the examination management system comprises the following steps:

[0018] S1 manages system login and checks user login information.

[0019] S2 user information management, manage the user's grade, subject, and grade information;

[0020] S3 test paper and question management, edit and manage test questions and test papers;

[0021] S4 specifically includes the test paper and question management module, which consists of two parts: test question management and test paper management;

[0022] S5 user log management, recording and displaying user events;

[0023] S6 task management, assigning tasks that users need to complete;

[0024] S7 user message management manages and records messages sent between users, and can also send messages to users.

[0025] In one embodiment, in the S2 user information management module, the administrator can use the search function to find user information by user name. After the administrator enters the search conditions, clicks the query button, and the system obtains user information that meets the conditions from the database and displays it in a list on the interface.

[0026] In one embodiment, the task management module of S6 is divided into task creation and task list. In task creation, the administrator can set the grade and title of the task and add test papers. After clicking the Add Test Paper button, a test paper selection dialog box will pop up. After clicking Add Test Paper, the test papers that have not been assigned tasks can be selected in the pop-up dialog box. All candidates will receive test paper tasks and they will be displayed in the task list.

[0027] In one embodiment, the knowledge graph uses a graphical structure to organize and represent knowledge, which describes the connections and attributes between entities through nodes and relationship edges. The system organizes and manages medical knowledge information through the knowledge graph, uses Py2neo and Neo4j as knowledge graph construction tools, and constructs a medical-based knowledge graph.

[0028] This application also provides a knowledge graph based on the questions and answers of the standardized training examination for resident physicians. The steps for constructing the data of the knowledge graph are as follows:

[0029] First, read the original data from the open source data set, analyze the relationship between the data, and build a knowledge graph based on the association between the data;

[0030] Secondly, import the data into the Neo4j database one by one and create the relationships between nodes and edges;

[0031] Finally, build a flexible and scalable knowledge graph to continuously optimize and update the knowledge base based on user needs and feedback.

[0032] In one embodiment, it also includes a question-answering system, which is composed of a question classification module, a question analysis module, and an answer search module;

[0033] The question classification module is responsible for classifying the questions input by the user to determine the type of question;

[0034] Question Parsing Module and Answer Search Module After the questions are classified, the question parsing module will be responsible for converting natural language questions into executable query statements, extracting entity information in the questions, and building corresponding Cypher query statements according to the question type to retrieve relevant information in the knowledge graph.

[0035] Compared with the prior art, the technical solution of the present invention has the following beneficial effects:

[0036] This application integrates medical Q&A, medical testing and resident physician training. The system adopts a B / S architecture, the front end uses WeChat applet as the examination interface, the back end uses spring+MySQL architecture to build a management platform, and a medical knowledge graph is constructed.

[0037] This application realizes the simulation of the standardized training examination for resident doctors through module setting. In the simulation examination system, the robot acts as a patient, and the doctor diagnoses the robot by asking questions. The robot can answer the doctor's questions. After the questions are finished, the doctor diagnoses the patient through the examination paper, and the doctor is scored by the result analysis. It satisfies the real-life simulation of doctors taking the exam. In the simulation examination system, doctors diagnose patients through examination papers, and score doctors through result analysis. The whole process fully meets the requirements of the standardized training examination for doctors.

[0038] The knowledge graph based on the questions and answers of the standardized training examination for resident physicians in this application facilitates doctors or other users to quickly query medical-related knowledge or conduct medical popularization. The system built using the knowledge graph uses natural language processing technology to achieve question classification, question analysis and answer search, forming a complete question-and-answer system, and finally presenting concise and clear answers to users. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 A system framework diagram of an embodiment of the present invention.

[0040] Figure 2 FIG. 4 is a system module diagram of an embodiment of the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical scheme and advantages of the embodiment of the present application clearer, the technical scheme of the embodiment of the present application will be clearly and completely described in conjunction with the drawings of the embodiment of the present application. Obviously, the described embodiment is a part of the embodiment of the present application, not all of the embodiments. Based on the described embodiment of the present application, all other embodiments obtained by ordinary technicians in this field without creative work belong to the scope of protection of this application.

[0042] like Figure 1 and Figure 2 As shown in the figure, the standardized training examination system for resident doctors is mainly divided into two parts: intelligent medical question-and-answer system and examination management system. The two systems are divided into three layers: front-end, back-end and database. Each layer has specific ideas and functions, and adopts a front-end and back-end separation architecture design.

[0043] The front end of the intelligent medical question-and-answer system uses the WeChat applet framework to implement the functions of intelligent medical question-and-answer and examination, including user login module, intelligent diagnosis and intelligent question-and-answer module, medical examination module, test result analysis module, user personal information module, and user message module. Among them, the login module of the intelligent medical question-and-answer examination system includes functions such as user login and permission verification.

[0044] The intelligent consultation and intelligent question-and-answer modules include functions such as building knowledge graphs, building intelligent question-and-answer systems, responding to doctors' questions and users' medical science questions; the medical examination module includes functions such as user answering questions and answer submission; and test result analysis includes answer result display and question analysis.

[0045] The WeChat applet front end mainly uses wx.request to communicate with the back-end interface and send HTTP requests to the back end. These requests usually include user operations, such as clicking buttons, filling out forms, etc., and call the corresponding back-end interfaces to complete the communication tasks with the back end.

[0046] The backend of the intelligent medical question-answering examination system is mainly responsible for processing front-end requests and implementing the question-answering system based on the knowledge graph, using the B / S architecture to complete the front-end and back-end communication. The intelligent question-answering part of the system uses Flask as the back-end framework and designs related interfaces to complete the front-end and back-end communication and data processing. When the back-end receives an HTTP request from the front-end, Flask returns the answer to the corresponding question to the front-end by querying the knowledge graph.

[0047] The intelligent medical question-answering system specifically includes the following steps:

[0048] S1 User registration and login. Used to detect user login information; for first-time users, user registration function is provided.

[0049] S2 Intelligent consultation and intelligent question and answer. Responsible for responding to doctors' consultations and users' medical science questions respectively. Preferably, the S2 specifically includes: building a knowledge graph, building an intelligent question and answer system, responding to doctors' consultations and users' medical science questions, etc.

[0050] S3 Medical Examination: Used to provide physicians with an examination interface to assess their thyroid disease diagnosis.

[0051] S4 test result analysis. Used to display the test score after submitting the test paper, total score, user answer, standard answer and analysis, etc.

[0052] S5 User personal information. Used to display the personal information of the current user.

[0053] The examination management system includes the following steps:

[0054] S1 manages system login and checks user login information.

[0055] S2 User Information Management. Manage the user's grade, subject, and grade information. Preferably, S2 specifically includes: In the user information management module, the administrator can use the search function to find user information by user name. After the administrator enters the search conditions and clicks the query button, the system obtains the user information that meets the conditions from the database and displays it in a list on the interface.

[0056] S3 test paper and question management. Edit and manage test questions and test papers. Preferably, S3 specifically includes a test paper and question management module consisting of two parts: test question management and test paper management.

[0057] S4 User log management. Record and display user events.

[0058] S5 Task Management: Assign tasks to users.

[0059] S6 User Message Management: Manage and record messages sent between users, and also send messages to users.

[0060] Preferably, the task management module of S5 is divided into task creation and task list. In task creation, the administrator can set the grade and title of the task and add test papers. After clicking the Add Test Paper button, a test paper selection dialog box will pop up. After clicking Add Test Paper, a test paper that has not been assigned a task can be selected in the pop-up dialog box. All students in the grade will receive the test paper task and display it in the task list.

[0061] The front-end layer of the examination management system mainly presents the visual interface of the system to users. The front-end of the examination management system uses the Vue framework to implement the interface display of the examination management function, including the management system login module, user information management module, test paper management module, user log management module and task management module.

[0062] The management system login module includes functions such as user login and authority verification; the user information management module includes functions such as user paging, user addition, deletion, modification and query; the test paper and question management module includes functions such as test paper paging, test paper query, test paper editing, question editing, question query, question paging, etc.; the user log management module includes user logs, log paging, etc.

[0063] The task management module includes tasks paging, task query, task editing and other functions. In addition, the front end of the project also uses Echarts for chart drawing, Ueditor for topic editing, and Element UI and Vue-element-admin frameworks, which provide important support for the front-end implementation. At the same time, Axios is used as a communication tool to send HTTP requests to the back end.

[0064] The backend of the examination management system is mainly responsible for processing front-end requests, supporting front-end function implementation, processing data, and implementing the addition, deletion, modification, and query of the database. The system as a whole uses the B / S (Browser / Server) architecture to complete the front-end and back-end communication. The system uses MyBat is to perform SQL queries, insertions, updates, and other operations. By defining the Mapper interface and SQL mapping file, Java objects are mapped to database tables to implement data addition, deletion, modification, and query operations. Subsequently, the backend returns the response data. After processing the business logic, the backend encapsulates the processing results into JSON format data and returns it to the front end through an HTTP response. After receiving the response from the backend, the front end performs corresponding processing based on the response data, such as updating the data on the interface, displaying prompt information, etc.

[0065] We will update and iterate the product and continuously optimize the system functions. Based on a certain amount of existing data, we will adjust the backend database, optimize the database structure, improve the accuracy of the scoring and the stability of the model, and ultimately achieve the goal of a better comprehensive scoring system, providing doctors and examinees with better exam simulation services.

[0066] The database layer is mainly responsible for data storage and management. The Neo4j database implements the storage and management of medical knowledge graphs. The system's database implements the storage and management of test questions and paper data, user data, logs and other data, and responds to the management call of the database by the backend method through the Mybatis middleware, ensuring that the data can be effectively managed.

[0067] The present invention also proposes a knowledge graph based on the standardized training examination for resident physicians. The knowledge graph is a method of organizing and representing knowledge using a graph structure, which describes the connections and attributes between entities through nodes and relationship edges. The system organizes and manages knowledge information in the medical field through the knowledge graph, including the relationships between multiple entities such as diseases, drugs, food, departments, and examinations. Py2neo and Neo4j are selected as tools for building the knowledge graph to build a medical-based knowledge graph.

[0068] In terms of data model design, the system has designed multiple entity nodes, including diseases, drugs, precautions, departments, etc., as well as the relationships between them, such as the relationship between diseases and drugs, the relationship between diseases and precautions, etc. Each node has some attributes to describe its characteristics. For example, the disease node includes attributes such as description, prevention method, and cause. The steps for constructing the knowledge graph data are as follows:

[0069] First, read the original data from the open source data set, analyze the relationship between the data, and build a knowledge graph based on the association between the data.

[0070] The data sources of the knowledge graph are divided into two parts. The first part is the medical data related to thyroid cancer provided by doctors of Shanghai Sixth People's Hospital, and the second part is the medical open source data set obtained from medical websites. The data set can be divided into information such as diseases, drugs, food, examinations, departments, etc., and stored in data files in JSON format. The data in JSON format is converted into a Python dictionary through Python library functions; then the information of each disease is traversed to extract the name, description, prevention method, cause, susceptible population, cure time, treatment department, treatment route, cure probability and other information of the disease; finally, the relevant entities of each disease are extracted, such as symptoms, complications, recommended foods, taboo foods, commonly used drugs, recommended drugs, examination items, etc., to facilitate the subsequent construction of knowledge graphs by constructing entity, relationship, and attribute triples.

[0071] Secondly, import the data into the Neo4j database one by one and create the nodes and edge relationships.

[0072] The method of constructing a knowledge graph by constructing entity, relationship, and attribute triples is as follows:

[0073] Entity construction. First, different types of entities are read from the preprocessed medical data file, such as drugs, foods, examinations, departments, pharmaceutical factories, symptoms, and diseases, and are represented by lists, such as drugs, foods, checks, etc. In the process of reading the data file, the attribute information of each entity is also extracted and mapped to the corresponding entity object. For example, information such as the description of the disease and prevention methods will be stored in a dictionary and then added to the disease entity list as a whole.

[0074] Relationship construction. The processed medical data also contains association information between entities. For example, a disease may recommend or avoid certain foods, require specific examinations, and may be associated with certain drugs. Based on this association information, the relationship between different entities is constructed. Relationships include but are not limited to the relationship between disease and food, the relationship between disease and medicine, the relationship between disease and examination, etc. These relationships are saved in the form of a list, and each relationship is represented in a specific format, such as <disease, food> represents the relationship between disease and food.

[0075] Create nodes. After completing the construction of entities, the system creates corresponding nodes based on these entities and stores them in the graph database. Using the Neo4j graph database, each entity will become a node in the database. When creating a node, a label will be assigned to each node, and the label corresponds to the type of entity, such as medicine, food, inspection, etc.

[0076] Create relationship edges. After completing the establishment of nodes, create relationship edges between different entities based on the previously constructed relationship information and store them in the graph database. These relationship edges represent the association between entities. For example, the relationship edge between a disease and a drug indicates that the drug is a drug for treating the disease. The creation of relationship edges is achieved by executing the corresponding Cypher query statements, which include information such as the starting node, the end node, and the relationship type. These query statements will be sent to the Neo4j database to establish relationships between entities.

[0077] In terms of data query and data analysis, the system uses the Cypher query language to flexibly query and analyze the data in the knowledge graph, including querying relevant information of a certain disease, commonly used drugs, precautions, etc.

[0078] The knowledge graph is one of the core elements of the question-answering system. The system builds and manages the medical knowledge graph, uses a graph structure to represent knowledge, achieves efficient data storage and query, and improves the query speed and accuracy of the system.

[0079] Finally, the constructed knowledge graph is also flexible and scalable. It can continuously optimize and update the knowledge base according to user needs and feedback, continuously improve the intelligence level of the system, and provide users with more personalized and accurate services.

[0080] The intelligent question-answering system built on the knowledge graph includes question classification, question analysis and answer search, and finally presents a concise and clear answer to the user. The question-answering system consists of a question classifier, a question analysis module and an answer search module.

[0081] The system built by the knowledge graph uses natural language processing technology to classify exam questions, analyze questions, and search for answers, ultimately presenting users with concise and clear answers. The question-answering system consists of a question classification module, a question analysis module, and an answer search module.

[0082] Question classification module: This module is responsible for classifying the questions input by the user to determine the type of question. First, in the initialization stage, various feature dictionaries are loaded, including diseases, departments, examinations, drugs, foods, major categories of medicines, and symptoms; then, by constructing a list of keywords for various question types (such as symptoms, causes, complications, etc.), combined with the entity types involved in the user's question, the type of question can be accurately determined. For example, if the user's question contains keywords about symptoms and involves disease entity types, it can be determined that the question belongs to the disease symptom query type. Finally, a dictionary is assembled based on the classification results and returned, including information such as the type of question and the entity types involved.

[0083] Question parsing module and answer search module After the questions are classified, the question parsing module will be responsible for converting natural language questions into executable query statements, extracting entity information in the questions, and constructing corresponding Cypher query statements according to the question type to retrieve relevant information in the knowledge graph. The main function of question type parsing supported by the system receives the result of question classification and the constructed entity dictionary as input. In this function, according to the question type, the question parsing module implements a series of Cypher query generation strategies. These strategies generate corresponding Cypher query statements based on the question type and related entities, which are used to retrieve relevant information from the medical knowledge graph.

[0084] The acquisition module, together with the knowledge graph, classification module, parsing module and answer search module, builds a complete medical intelligent system. The system first classifies the questions raised by users through the classifier, then the parsing module constructs the corresponding query statement according to the question type, and finally the answer search module executes the query in the graph database and presents the results in a beautiful way, realizing the automatic processing and accurate answering of medical questions raised by users, providing users with convenient and efficient services.

[0085] A complete medical knowledge graph was built using Python and Neo4j graph database. The optimization module is used to continuously optimize the model to improve accuracy and model stability. This module optimizes the model based on a certain scale of data, adjusts hyperparameters, etc., and ultimately achieves the accuracy target.

[0086] The system built using knowledge graphs uses natural language processing technology to implement question classification, question analysis, and answer search to form a complete question-and-answer system that ultimately presents concise and clear answers to users.

[0087] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A standardized training examination system for resident doctors based on knowledge graph, characterized in that: It includes an intelligent medical question-and-answer system and an examination management system. The front-end of the intelligent medical question-and-answer system uses the WeChat applet framework to implement the functions of intelligent medical question-and-answer and examination, including a user login module, intelligent consultation and intelligent question-and-answer module, medical examination module, test result analysis module, user personal information module, and user message module; The intelligent consultation and intelligent question-answering modules include functions such as building knowledge graphs, intelligent question-answering, responding to doctors' questions and users' medical science questions; the medical examination module includes functions such as users answering questions and submitting answers; the test result analysis includes the display of answer results and question analysis; The backend of the intelligent medical question-and-answer examination system is mainly responsible for processing front-end requests and implementing the question-and-answer system based on the knowledge graph, and uses the B / S architecture to complete front-end and back-end communication.

2. The standardized training examination system for resident doctors based on knowledge graph according to claim 1 is characterized in that: The intelligent medical question-answering system specifically includes the following steps: S1 user registration and login, used to detect user login information; for first-time users, user registration function is provided; S2 intelligent consultation and intelligent question and answer are responsible for responding to doctors' consultation and users' medical science questions respectively; preferably, S2 specifically includes: building a knowledge graph, building an intelligent question and answer system, responding to doctors' consultation and users' medical science questions, etc. S3 medical examination, which is used to provide physicians with an easy-to-understand examination interface to assess their thyroid disease diagnosis; S4 test result analysis, used to display the test score after submitting the test paper, total score, user answer, standard answer and analysis, etc.; S5 User personal information, used to display the current user's personal information.

3. The standardized training examination system for resident doctors based on knowledge graph according to claim 1 is characterized in that: The examination management system comprises the following steps: S1 manages system login and checks user login information. S2 user information management, manage the user's grade, subject, and grade information; S3 test paper and question management, edit and manage test questions and test papers; S4 specifically includes the test paper and question management module, which consists of two parts: test question management and test paper management; S5 user log management, recording and displaying user events; S6 task management, assigning tasks that users need to complete; S7 user message management manages and records messages sent between users, and can also send messages to users.

4. The standardized training examination system for resident doctors based on knowledge graph according to claim 3 is characterized in that: In the S2 user information management module, administrators can use the search function to find user information by user name. After the administrator enters the search criteria and clicks the query button, the system obtains user information that meets the criteria from the database and displays it in a list on the interface.

5. The standardized training examination system for resident doctors based on knowledge graph according to claim 3 is characterized in that: The task management module of S6 is divided into task creation and task list. In task creation, the administrator can set the grade and title of the task and add test papers. After clicking the Add Test Paper button, the test paper selection dialog box will pop up. After clicking Add Test Paper, the test papers that have not been assigned to tasks can be selected in the pop-up dialog box. All students in this grade will receive the test paper tasks and will be displayed in the task list.

6. A knowledge graph based on a standardized resident physician training examination system, characterized in that: Knowledge graphs use graph structures to organize and represent knowledge. They describe the connections and attributes between entities through nodes and edges. The system organizes and manages medical knowledge information through knowledge graphs, and uses Py2neo and Neo4j as knowledge graph construction tools to build a medical-based knowledge graph.

7. The knowledge graph based on the standardized training and examination system for resident physicians according to claim 6 is characterized in that: The steps for constructing the knowledge graph data are as follows: First, read the original data from the open source data set, analyze the relationship between the data, and build a knowledge graph based on the association between the data; Secondly, import the data into the Neo4j database one by one and create the relationships between nodes and edges; Finally, build a flexible and scalable knowledge graph to continuously optimize and update the knowledge base based on user needs and feedback.

8. The knowledge graph based on the standardized training and examination system for resident physicians according to claim 7 is characterized in that: It also includes a question-answering system, which is composed of a question classification module, a question analysis module, and an answer search module; The question classification module is responsible for classifying the questions input by the user to determine the type of question; Question Parsing Module and Answer Search Module After the questions are classified, the question parsing module will be responsible for converting natural language questions into executable query statements, extracting entity information in the questions, and building corresponding Cypher query statements according to the question type to retrieve relevant information in the knowledge graph.

Images