Pre-inquiry method and device

Abstract

The invention relates to the technical field of large models, and particularly provides a pre-inquiry method and device, and the method comprises the following steps: S1, constructing a medical knowledge graph; s2, model training; s3, patient information collection; and S4, generating a pre-inquiry sheet. Compared with the prior art, the pre-inquiry efficiency and accuracy can be remarkably improved, the workload of doctors is relieved, and the medical service quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of large models, and specifically provides a pre-diagnosis method and device. Background Art

[0002] With the continuous advancement of medical technology and the development of medical information technology, pre-consultation is an important part of the medical process, and its efficiency and accuracy are of great significance to improving the quality of medical services. The traditional pre-consultation method mainly relies on manual questioning and recording, which is not only time-consuming and labor-intensive, but also prone to information omissions or misunderstandings due to human factors.

[0003] In recent years, with the rapid development of technologies such as big data and artificial intelligence, especially the application of deep learning technology and knowledge graphs in the medical field, new ideas have been provided for the innovation of pre-diagnosis methods.

[0004] However, most of the pre-consultation systems on the market currently have the following problems:

[0005] 1. The medical knowledge base is incomplete, which makes it impossible for the system to accurately understand the patient's symptoms and give reasonable suggestions;

[0006] Second, it lacks intelligent conversation capabilities and cannot dynamically adjust the content of inquiries based on the patient's answers;

[0007] 3. Insufficient information integration capabilities, unable to effectively integrate the collected information and generate a pre-consultation form with guiding significance.

[0008] Therefore, developing a pre-diagnosis method based on large models and knowledge graphs to solve the above problems has important practical significance and application value. Summary of the invention

[0009] The present invention aims at solving the above-mentioned deficiencies of the prior art and provides a highly practical pre-diagnosis method.

[0010] A further technical task of the present invention is to provide a pre-diagnosis device with reasonable design, safety and applicability.

[0011] The technical solution adopted by the present invention to solve its technical problem is:

[0012] A pre-diagnosis method comprises the following steps:

[0013] S1. Construction of medical knowledge graph;

[0014] S2, model training;

[0015] S3, patient information collection;

[0016] S4. Generate a pre-consultation form.

[0017] Furthermore, in step S1, it further includes:

[0018] S1-1, data cleaning, collecting and organizing medical knowledge, cleaning the collected data, and removing redundant and contradictory information;

[0019] S1-2, knowledge ablation and verification, using machine learning technology to integrate and verify medical knowledge, eliminate duplicate and contradictory information, and provide a medical basis for subsequent steps;

[0020] S1-3. Construction of knowledge graph.

[0021] Furthermore, in step S1-3, a knowledge graph construction tool suitable for the medical field is selected or developed, and the knowledge graph construction tool is used to construct a structured medical knowledge graph. The nodes in the graph represent medical entities, and the edges represent the relationships between entities, providing a medical basis for subsequent steps.

[0022] Furthermore, in step S2, it further includes:

[0023] S2-1. Data preprocessing: preparing training data, including a large number of real doctor-patient conversation records and corresponding medical diagnosis results, and preprocessing the doctor-patient conversation records;

[0024] S2-2, model training and optimization, based on deep learning technology, train a large model that can understand and process natural language;

[0025] S2-3. The training model has the ability to conduct multiple rounds of conversations, dynamically adjust subsequent inquiries based on the patient's answers, and introduce semantic understanding technology.

[0026] Furthermore, in step S3, it further includes:

[0027] S3-1, pre-extraction of chief complaint inquiry;

[0028] S3-2, key information retrieval;

[0029] S3-3, large model intelligent inquiry of current medical history;

[0030] S3-4, large model intelligent inquiry of past history;

[0031] S3-5, large model intelligent inquiry of personal history;

[0032] S3-6. Large model intelligent inquiry of family history.

[0033] Furthermore, in step S3-1, the patient describes his / her chief complaint through the interface input or voice, and the trained large model is used to analyze the chief complaint and extract key symptoms and information;

[0034] In step S3-2, based on the main complaint, relevant information is retrieved from the knowledge graph;

[0035] In step S3-3, based on the information extracted from the chief complaint, the large model intelligently generates a series of medical history inquiry questions related to the current symptoms.

[0036] Furthermore, in step S3-4, based on the understanding of the current medical history, the large model further inquires about the patient's past medical history, including past illnesses, surgeries, hospitalizations, and allergies;

[0037] In step S3-5, personal history inquiries include lifestyle habits, dietary habits, occupational exposures, and family medical history;

[0038] In step S3-6, the large model asks whether there are people with similar symptoms in the patient's family and whether there is specific information about the family's history of infectious diseases and genetic diseases;

[0039] The big model also intelligently recommends genetic counseling or screening that patients may need based on the information in the medical vector knowledge base.

[0040] Furthermore, in step S4, it further includes:

[0041] S4-1, information integration, based on the natural language understanding ability of the big model, summarize the collected information and generate a pre-diagnosis form;

[0042] S4-2, based on the collected information and knowledge graph, intelligently recommend disease lists and examination items;

[0043] S4-3. Dynamic updating and sharing allows doctors to dynamically update and supplement the pre-consultation form during the subsequent consultation process.

[0044] A pre-diagnosis device, comprising: at least one memory and at least one processor;

[0045] The at least one memory is used to store a machine-readable program;

[0046] The at least one processor is used to call the machine-readable program to execute a pre-diagnosis method.

[0047] Compared with the prior art, the pre-diagnosis method and device of the present invention have the following outstanding beneficial effects:

[0048] The present invention improves the efficiency and accuracy of medical services: by intelligently analyzing patient complaints, automatically retrieving key medical information, intelligently generating inquiry questions and recommending possible diseases and examination items, the method significantly shortens the time of the traditional consultation process and improves the accuracy of diagnosis.

[0049] Optimize the patient medical experience: Patients can describe their symptoms through a simple interface or voice input, and the system can respond quickly and provide personalized inquiries and suggestions, making the medical process more convenient and humane.

[0050] Enhance the work efficiency of doctors: Doctors can quickly understand the basic situation of patients based on the pre-consultation form generated by the system, so as to formulate diagnosis and treatment plans more efficiently, reduce unnecessary repeated inquiries, and improve diagnosis and treatment efficiency.

[0051] Promote the rational allocation of medical resources: Through intelligent recommendations of possible diseases and examination items, it helps hospitals to rationally arrange medical resources, reduce unnecessary examinations and drug use, and reduce medical costs.

[0052] Support cross-platform collaboration and information sharing: The system supports the sharing and transmission of pre-consultation forms between different medical platforms, which promotes collaboration and communication between doctors and helps to form a more comprehensive diagnosis and treatment plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0054] Attached Figure 1 It is a flowchart of the pre-diagnosis method. DETAILED DESCRIPTION

[0055] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0056] A best embodiment is given below:

[0057] like Figure 1 As shown, a pre-diagnosis method in this embodiment has the following steps:

[0058] S1. Construction of medical knowledge graph;

[0059] Features include:

[0060] S1-1. Data cleaning: Collect and organize comprehensive medical knowledge, including but not limited to disease symptoms, diagnostic criteria, treatment methods, drug information, medical examinations, etc. Then, clean the collected data to remove redundant and contradictory information.

[0061] S1-2, knowledge ablation and verification, use machine learning technology to integrate and verify medical knowledge, eliminate duplicate and contradictory information, ensure the accuracy and completeness of the knowledge graph, and provide a solid medical foundation for subsequent steps.

[0062] S1-3. Knowledge graph construction: select or develop knowledge graph construction tools suitable for the medical field (such as Neo4j, Dgraph, etc.), and use the knowledge graph construction tools to construct a structured medical knowledge graph based on this knowledge. The nodes in the graph represent medical entities (such as diseases, symptoms, drugs, etc.), and the edges represent the relationships between entities (such as causal relationships, therapeutic relationships, etc.).

[0063] The key to this step is to ensure the accuracy and completeness of the knowledge graph and provide a solid medical foundation for subsequent steps.

[0064] S2, model training;

[0065] Further including:

[0066] S2-1, data preprocessing;

[0067] Prepare training data, including a large number of real doctor-patient conversation records and corresponding medical diagnosis results, and pre-process the doctor-patient conversation records, including noise removal, word segmentation, part-of-speech tagging, etc.

[0068] S2-2, model training and optimization;

[0069] Based on deep learning technology, train a large model that can understand and process natural language (such as BERT, GPT, etc.). The model must be able to extract key information from the patient's complaints and generate targeted inquiries based on the knowledge in the knowledge graph. The training data should include a large number of real doctor-patient conversation records and corresponding medical diagnosis results to ensure that the model can accurately understand the patient's intentions and give reasonable inquiries.

[0070] S2-3, multi-round dialogue capability;

[0071] The trained model has the ability to conduct multiple rounds of conversations, can dynamically adjust subsequent inquiries based on the patient's answers, and introduce semantic understanding technology to improve the model's ability to analyze key symptoms and information.

[0072] S3, patient information collection;

[0073] Specifically include:

[0074] S3-1. Chief complaint inquiry and extraction;

[0075] The patient describes his / her main complaint through the interface input or voice, and the system uses the trained large model to analyze the main complaint and extract key symptoms and information. The purpose of this step is to quickly identify the patient's main health problem and provide direction for subsequent inquiries.

[0076] S3-2, key information retrieval;

[0077] Based on the chief complaint, relevant information is retrieved from the knowledge graph, including common symptoms, necessary examinations, medicines, and other information.

[0078] S3-3, large model intelligent inquiry of current medical history;

[0079] Based on the information extracted from the chief complaint, the big model intelligently generates a series of questions about the current medical history related to the current symptoms, such as the duration of symptoms, frequency of attacks, aggravating or alleviating factors, etc. These questions are designed to gain a deeper understanding of the patient's current health status and provide more clues for diagnosis.

[0080] S3-4, large model intelligent inquiry of past history;

[0081] Based on the current medical history, the big model further asks the patient about his past medical history, including past illnesses, surgeries, hospitalizations, allergies, etc. This information helps identify potential health risks and may affect the treatment of current diseases.

[0082] S3-5, large model intelligent inquiry of personal history;

[0083] Personal history inquiries include lifestyle habits (such as smoking, drinking), eating habits, occupational exposure, family medical history, etc. These factors may be directly related to the patient's health status and are crucial for assessing disease risks and formulating preventive measures.

[0084] S3-6, large model intelligent inquiry of family history;

[0085] The large model asks whether there are people with similar symptoms in the patient's family, and whether there is specific information about the family's history of infectious diseases and genetic diseases, such as the severity of the illness and treatment status of the sick relatives.

[0086] In addition, the big model will intelligently recommend genetic counseling or screening that patients may need based on the information in the medical vector knowledge base. Such in-depth inquiries help to gain a more comprehensive understanding of the patient's family health status and provide patients with more accurate diagnosis and treatment recommendations.

[0087] S4, generating a pre-inquiry form;

[0088] Specifically include:

[0089] S4-1. Information integration: Based on the natural language comprehension ability of the large model, summarize the information collected in the above steps to generate a pre-interview form, including the patient's basic information, chief complaint, current medical history, past medical history, personal history and family history.

[0090] S4-2. Based on the collected information and knowledge graph, intelligently recommend possible disease lists and examination items.

[0091] S4-3, dynamic update and sharing, allows doctors to dynamically update and supplement the pre-consultation form during the subsequent consultation process. In addition, it supports the sharing and transmission of pre-consultation forms between different medical platforms to facilitate collaboration and communication between doctors.

[0092] Based on the above method, a pre-diagnosis device in this embodiment includes: at least one memory and at least one processor;

[0093] The at least one memory is used to store a machine-readable program;

[0094] The at least one processor is used to call the machine-readable program to execute a pre-diagnosis method.

[0095] The above-mentioned specific implementations are only specific cases of the present invention. The patent protection scope of the present invention includes but is not limited to the above-mentioned specific implementations. Any technical solutions that conform to the above-mentioned specific implementations of the present invention and any appropriate changes or substitutions made by ordinary technicians in the relevant technical field shall fall within the patent protection scope of the present invention.

[0096] 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 pre-diagnosis method, characterized in that: The steps are as follows: S1. Construction of medical knowledge graph; S2, model training; S3, patient information collection; S4. Generate a pre-consultation form.

2. A pre-diagnosis method according to claim 1, characterized in that: In step S1, further comprising: S1-1, data cleaning, collecting and organizing medical knowledge, cleaning the collected data, and removing redundant and contradictory information; S1-2, knowledge ablation and verification, using machine learning technology to integrate and verify medical knowledge, eliminate duplicate and contradictory information, and provide a medical basis for subsequent steps; S1-3. Construction of knowledge graph.

3. A pre-diagnosis method according to claim 2, characterized in that: In step S1-3, a knowledge graph construction tool suitable for the medical field is selected or developed, and the knowledge graph construction tool is used to construct a structured medical knowledge graph. The nodes in the graph represent medical entities, and the edges represent the relationships between entities, providing a medical basis for subsequent steps.

4. A pre-diagnosis method according to claim 3, characterized in that: In step S2, further comprising: S2-1. Data preprocessing: preparing training data, including a large number of real doctor-patient conversation records and corresponding medical diagnosis results, and preprocessing the doctor-patient conversation records; S2-2, model training and optimization, based on deep learning technology, train a large model that can understand and process natural language; S2-3. The training model has the ability to conduct multiple rounds of conversations, dynamically adjust subsequent inquiries based on the patient's answers, and introduce semantic understanding technology.

5. A pre-diagnosis method according to claim 4, characterized in that: In step S3, further comprising: S3-1, pre-extraction of chief complaint inquiry; S3-2, key information retrieval; S3-3, large model intelligent inquiry of current medical history; S3-4, large model intelligent inquiry of past history; S3-5, large model intelligent inquiry of personal history; S3-6. Large model intelligent inquiry of family history.

6. A pre-diagnosis method according to claim 5, characterized in that: In step S3-1, the patient describes his / her chief complaint through the interface input or voice, and the trained large model is used to analyze the chief complaint and extract key symptoms and information; In step S3-2, based on the main complaint, relevant information is retrieved from the knowledge graph; In step S3-3, based on the information extracted from the chief complaint, the large model intelligently generates a series of medical history inquiry questions related to the current symptoms.

7. A pre-diagnosis method according to claim 6, characterized in that: In step S3-4, based on the current medical history, the large model further inquires about the patient's past medical history, including past illnesses, surgeries, hospitalizations, and allergies; In step S3-5, personal history inquiries include lifestyle habits, dietary habits, occupational exposures, and family medical history; In step S3-6, the large model asks whether there are people with similar symptoms in the patient's family and whether there is specific information about the family's history of infectious diseases and genetic diseases; The big model also intelligently recommends genetic counseling or screening that patients may need based on the information in the medical vector knowledge base.

8. A pre-diagnosis method according to claim 7, characterized in that: In step S4, further comprising: S4-1, information integration, based on the natural language understanding ability of the big model, summarize the collected information and generate a pre-diagnosis form; S4-2, based on the collected information and knowledge graph, intelligently recommend disease lists and examination items; S4-3. Dynamic updating and sharing allows doctors to dynamically update and supplement the pre-consultation form during the subsequent consultation process.

9. A pre-diagnosis device, characterized in that: include: at least one memory and at least one processor; The at least one memory is used to store a machine-readable program; The at least one processor is configured to call the machine-readable program to execute the method according to any one of claims 1 to 8.

Images