Traditional Chinese medicine standard diagnosis and treatment path and curative effect generation method and device
By constructing a knowledge graph and large language model to generate standard diagnosis and treatment paths and efficacy standards in traditional Chinese medicine, the problem of inefficiency of traditional methods is solved, efficient and accurate generation results are achieved, and the needs of traditional Chinese medicine medical insurance payment reform are met.
Patent Information
- Application Number
- CN202510297675.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-13
- Publication Date
- 2025-07-11
AI Technical Summary
Traditional manual formulation of standard diagnosis and treatment paths and efficacy standards for traditional Chinese medicine is inefficient, subjective, and difficult to cover a large number of case data, resulting in the lack of uniformity and objectivity of the standards formulated and cannot meet the precise payment requirements of traditional Chinese medicine medical insurance payment reform.
Natural language processing technology is used to obtain text data related to traditional Chinese medicine diagnosis and treatment, build a knowledge graph, generate standard diagnosis and treatment paths and efficacy standards through large language models, and accurately search and generate them in combination with disease course records.
Rapidly generate standard diagnosis and treatment paths and efficacy standards for traditional Chinese medicine, improve work efficiency, ensure the fit and credibility of the generated results with clinical data, and reduce errors caused by model limitations.
Smart Images
Figure CN120299650A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and particularly to a method and device for generating a standard diagnosis and treatment path and curative effect in traditional Chinese medicine. Background Art
[0002] Under the background of the reform of traditional Chinese medicine medical insurance payment, in order to implement a diversified compound medical insurance payment method mainly based on payment by disease type, it is necessary to formulate a standard diagnosis and treatment path and corresponding curative effect standards for each disease type, so that the medical insurance department can make payments according to the curative effect.
[0003] There are many limitations in the traditional method of manually formulating standard diagnosis and treatment paths and curative effect standards. First of all, the manual formulation process is inefficient, requiring a large number of professionals to invest a lot of time in analyzing and summarizing a large amount of traditional Chinese medicine clinical data, and it is difficult to complete the formulation of diagnosis and treatment paths and curative effect standards for many disease types in a short time. Secondly, the manual formulation process is highly subjective. Different experts may have different views on the diagnosis and treatment paths and curative effect standards based on their own experience and understanding, resulting in the lack of unity and objectivity of the formulated standards. In addition, it is difficult for manual formulation to cover a large amount of case data and make full use of rich clinical experience and data resources, so that the formulated standards may not be comprehensive and accurate enough to meet the requirements of accurate payment in the reform of traditional Chinese medicine medical insurance payment. Summary of the Invention
[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method and device for generating a standard diagnosis and treatment path and curative effect in traditional Chinese medicine, so as to solve the technical problem that the traditional method of manually formulating standard diagnosis and treatment paths and curative effect standards is difficult to meet the current needs.
[0005] To achieve the above object, the present invention is implemented by the following technical solutions: In a first aspect, the present invention provides a method for generating a standard diagnosis and treatment path and curative effect in traditional Chinese medicine, including: Obtaining text data related to traditional Chinese medicine diagnosis and treatment, and cleaning and screening the text data; Based on the cleaned and screened text data, constructing a knowledge graph by using natural language processing technology; Obtaining a medical record, and retrieving relevant entries from the knowledge graph for the entities in the medical record based on semantic similarity; Combining the relevant entries, the diagnosis and treatment records in the medical record, and a preset question model and inputting them into a large language model to obtain corresponding standard diagnosis and treatment paths and curative effect standards.
[0006] Optionally, the obtaining of the text data related to traditional Chinese medicine diagnosis and treatment includes: For paper documents related to traditional Chinese medicine (TCM) diagnosis and treatment, the paper content is scanned into electronic images through a scanning device, and the optical character recognition technology is applied to the electronic images to convert the text content in the electronic images into electronic text; For electronic documents related to TCM diagnosis and treatment, electronic text is downloaded by authorizing and accessing the corresponding websites and databases.
[0007] Optionally, the cleaning and screening of the text data include: For each document in the text data, use text editing tools and programming scripts to remove invalid information, convert it into the target encoding format, and eliminate documents that are irrelevant to the TCM diagnosis and treatment path and efficacy.
[0008] Optionally, the removal of invalid information includes removing headers, footers, page numbers, garbled characters, and special characters.
[0009] Optionally, the construction of the knowledge graph using natural language processing technology includes: Input the text data after cleaning and screening into a large language model to identify target entities and relationships; For each pair of the target entities, calculate the cosine similarity as the semantic association degree : ; In the formula, is the th target entity, is a pre-trained medical word embedding model, is the norm of the vector; Statistically count the frequency of occurrence of each relationship between each pair of the target entities, and calculate the weight of each relationship in combination with the semantic association degree : ; In the formula, is the th relationship of the th target entity, is the frequency of occurrence, is the weight of; Construct a knowledge graph based on the entities, relationships, and weights of the relationships.
[0010] Optionally, the target entities include diseases, drugs, and symptoms; the relationships include causal relationships, parallel relationships, and interaction relationships.
[0011] Optionally, the retrieval of relevant entries from the knowledge graph for the entities in the case record based on semantic similarity includes: Form entity pairs by combining all entities in the case record and all entities in the knowledge graph, and calculate the cosine similarity of the entity pairs as the semantic association degree; For each entity in the case record, based on the semantic association degree of the entity pairs, retrieve multiple entries most relevant to it from the knowledge graph as relevant entries.
[0012] In a second aspect, the present invention provides a traditional Chinese medicine standard diagnosis and treatment path and efficacy generation device, including: A data processing module, configured to obtain text data related to traditional Chinese medicine diagnosis and treatment, and clean and screen the text data; A knowledge graph module, configured to construct a knowledge graph based on the cleaned and screened text data by using natural language processing technology; An entry retrieval module, configured to obtain a case record, and retrieve relevant entries from the knowledge graph for entities in the case record based on semantic similarity; A question and answer execution module, configured to merge the relevant entries, the diagnosis and treatment records in the case record, and a preset question model and input them into a large language model to obtain corresponding standard diagnosis and treatment paths and efficacy criteria.
[0013] In a third aspect, the present invention provides an electronic device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the above method.
[0014] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented.
[0015] Compared with the prior art, the beneficial effects achieved by the present invention: The present invention provides a traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method and device. 1) By leveraging the powerful information processing and learning capabilities of the large model, it can quickly and efficiently generate traditional Chinese medicine standard diagnosis and treatment paths and efficacy criteria. Compared with the traditional manual formulation method, the working cycle is greatly shortened, and the work efficiency is significantly improved, enabling limited human resources to be liberated from tedious repetitive work. 2) Adopting the method of retrieval-enhanced generation effectively makes up for the hallucination problem that may occur in the generation process of the large model. Through accurate retrieval of the authoritative knowledge base, it provides a solid and reliable data support for the generation process of the large model, ensuring that the generated diagnosis and treatment paths and efficacy criteria are closely in line with actual clinical data and medical knowledge, avoiding the occurrence of errors or inaccurate information caused by the limitations of the model itself, enhancing the credibility and practicality of the generated results, and making them more in line with the professional requirements and norms of the medical industry. Brief Description of the Drawings
[0016] Figure 1 It is a schematic flowchart of the traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method provided by the embodiments of the present invention. Detailed Embodiments
[0017] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solutions of the present invention and cannot be used to limit the protection scope of the present invention.
[0018] Embodiment 1:
[0019] As Figure 1 shown, the embodiments of the present invention provide a traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method, including the following steps: Step S1: Obtain text data related to traditional Chinese medicine diagnosis and treatment, and clean and screen the text data.
[0020] The text data related to traditional Chinese medicine diagnosis and treatment mainly includes two forms: paper documents and electronic documents. For paper documents, such as some classical documents, the paper content is scanned into an electronic image through a scanning device, and the optical character recognition technology is used for the electronic image to convert the text content in the electronic image into an electronic text, and the electronic text can be edited and corrected to ensure the accuracy and integrity of the text. For electronic documents, through authorization, access to corresponding websites (such as the official website of academic journals) and databases (such as medical databases such as PubMed and CNKI), the keyword search function is used to accurately retrieve the literature data related to traditional Chinese medicine diagnosis and treatment, and the retrieved literature data is saved in electronic form to provide comprehensive and accurate original materials for subsequent data processing and analysis.
[0021] Step S2: Based on the cleaned and screened text data, construct a knowledge graph using natural language processing technology.
[0022] Based on the cleaned and screened text data specifically includes: For each literature in the text data, use text editing tools and programming scripts to remove invalid information, convert it into a target encoding format, and eliminate the literature irrelevant to the traditional Chinese medicine diagnosis and treatment path and efficacy.
[0023] Specifically in this embodiment, removing invalid information includes removing headers, footers, page numbers, garbled characters, and special characters. According to actual needs, other invalid information can also be added.
[0024] The target encoding format selects the commonly used UTF-8. UTF-8 encoding ensures the compatibility and consistency of the text on different systems and platforms with its advantages such as global universality, backward compatibility, efficient storage, easy processing and transmission, flexibility and scalability, and wide application and support.
[0025] Based on information such as the title, abstract, and keywords of the literature, conduct a preliminary screening of the literature, and eliminate the literature that has nothing to do with traditional Chinese medicine diagnosis and treatment paths and efficacy.
[0026] Using natural language processing technology to construct a knowledge graph specifically includes: 1) Input the cleaned and screened text data into a large language model to identify target entities and relationships.
[0027] A large language model (LLM) is an artificial intelligence model based on deep learning technology, aiming to understand and generate human language, such as GPT-4.
[0028] By setting specific system instructions and prompts, guide the large language model to identify target entities and relationships. Specifically, in this embodiment, the target entities include but are not limited to diseases, drugs, and symptoms; the relationships include but are not limited to causal relationships, parallel relationships, and interaction relationships.
[0029] 2) For each pair of target entities, calculate their cosine similarity as the semantic association degree : ; In the formula, is the th target entity, is a pre-trained medical word embedding model, is the norm of the vector.
[0030] The word embedding model is a key technology in natural language processing (NLP), used to map words or phrases from a vocabulary (which may contain thousands or millions of words) to real number vectors with a fixed dimension. In this embodiment, the word embedding model is pre-trained with data in the medical field to better meet the computational requirements of this example.
[0031] 3) Count the frequency of each relationship between each pair of target entities, and calculate the weight of each relationship in combination with the semantic association degree : ; In the formula, is the th relationship of the th target entity, is the frequency of occurrence, is the weight of.
[0032] 4) Construct a knowledge graph based on entities, relationships, and the weights of relationships.
[0033] Knowledge graph , is a set of target entities, is a set of relationships, and the weight of the edge is determined by the weight .
[0034] The knowledge graph is stored in the Neo4j graph database in a standardized format, providing efficient support for subsequent retrieval and generation of standard treatment paths.
[0035] Step S3: Obtain the medical record, and retrieve relevant entries from the knowledge graph for the entities in the medical record based on semantic similarity.
[0036] Extract all entities from the medical record, such as diseases in the admission diagnosis, examination items in the diagnostic plan, prescriptions and formula explanations.
[0037] Form entity pairs from all entities in the medical record and all entities in the knowledge graph, and calculate the cosine similarity of the entity pairs as the semantic association degree; For each entity in the medical record, based on the semantic association degree of the entity pairs, retrieve multiple entries most relevant to it from the knowledge graph as relevant entries.
[0038] Relevant entries , is the number of selected entries, is the entity in the knowledge graph and the entity in the medical record 's semantic association degree, is the entity set of the medical record and the knowledge graph.
[0039] The selected relevant entries contain detailed information related to the entities in the medical record, such as drug instructions, treatment experience, etc. In the specific implementation, the vector representation uses the pre-trained medical word embedding model BioBERT, and filters out low-correlation entries by setting a similarity threshold. Finally, the top k entries retrieved are combined with the medical record to form a high-correlation information set of medical data, providing accurate support for subsequent generation of standard treatment paths.
[0040] Step S4: Combine the relevant entries, the treatment records in the medical record, and the preset question model and input them into the large language model to obtain the corresponding standard treatment path and efficacy standard.
[0041] The question model is "Generate the standard treatment path and efficacy standard for this disease according to the following treatment records and relevant entries". The large language model outputs the final result through context understanding and generation ability.
[0042] In order to verify the accuracy of the generated results, this embodiment sets evaluation indicators: ; ; ; ; Wherein, is the standard diagnosis and treatment path and efficacy standard, respectively represent the i-th node in the real path and the i-th node in the generated standard diagnosis and treatment path, respectively represent the keyword sets in the real diagnosis and treatment path and the text of the generated standard diagnosis and treatment path, represents the number of elements in the intersection of the two word sets, represents the number of elements in the union of the two word sets.
[0043] Example Two:
[0044] Based on the traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method proposed in Example One, an embodiment of the present invention further provides a traditional Chinese medicine standard diagnosis and treatment path and efficacy generation device, including: A data processing module, configured to obtain text data related to traditional Chinese medicine diagnosis and treatment, and clean and filter the text data; A knowledge graph module, configured to construct a knowledge graph based on the cleaned and filtered text data by using natural language processing technology; An item retrieval module, configured to obtain a medical record, and retrieve relevant items from the knowledge graph for the entities in the medical record based on semantic similarity; A question and answer execution module, configured to merge the relevant items, the diagnosis and treatment records in the medical record, and a preset question model and input them into a large language model to obtain the corresponding standard diagnosis and treatment path and efficacy standard.
[0045] Example Three:
[0046] Based on the traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method proposed in Example One, an embodiment of the present invention further provides an electronic device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the above method.
[0047] Example Four:
[0048] Based on the traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method proposed in Example One, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the above method are implemented.
[0049] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0050] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0051] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that implement the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0052] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.
[0053] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.
Claims
1. A traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method, characterized in that, It includes: Obtain text data related to traditional Chinese medicine (TCM) diagnosis and treatment, and clean and screen the text data; Based on the text data after cleaning and screening, construct a knowledge graph using natural language processing technology; Obtain the case record, and retrieve relevant entries from the knowledge graph for the entities in the case record based on semantic similarity; Merge the relevant entries, the diagnosis and treatment records in the case record, and a preset question model and input them into a large language model to obtain corresponding standard diagnosis and treatment paths and efficacy criteria.
2. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 1, wherein The obtaining of the text data related to TCM diagnosis and treatment includes: For paper documents related to TCM diagnosis and treatment, scan the paper content into an electronic image through a scanning device, and use optical character recognition technology on the electronic image to convert the text content in the electronic image into electronic text; For electronic documents related to TCM diagnosis and treatment, download the electronic text by authorizing and accessing the corresponding websites and databases.
3. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 1, wherein The cleaning and screening of the text data includes: For each document in the text data, use text editing tools and programming scripts to remove invalid information, convert it into a target encoding format, and eliminate documents irrelevant to the TCM diagnosis and treatment path and efficacy.
4. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 3, wherein, The removal of invalid information includes removing headers, footers, page numbers, garbled characters, and special characters.
5. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 1, wherein The construction of the knowledge graph using natural language processing technology includes: Input the text data after cleaning and screening into a large language model to identify target entities and relationships; For each pair of the target entities, calculate the cosine similarity as the semantic association degree : ； In the formula, is the th target entity, is a pre-trained medical word embedding model, is the norm of the vector; Count the frequency of occurrence of each relationship between each pair of the target entities, and calculate the weight of each relationship in combination with the semantic association degree : ； In the formula, is the th relationship of the th target entity, is the frequency of occurrence, is the weight of. Construct a knowledge graph based on the entities, relationships, and weights of the relationships.
6. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 5, characterized in that The target entities include diseases, drugs, and symptoms; the relationships include causal relationships, parallel relationships, and interaction relationships.
7. The traditional Chinese medicine standard diagnosis and treatment path and efficacy generation method according to claim 1, characterized in that The retrieval of relevant entries from the knowledge graph for the entities in the case record based on semantic similarity includes: Form entity pairs by combining all entities in the case record and all entities in the knowledge graph, and calculate the cosine similarity of the entity pairs as the semantic association degree; For each entity in the case record, based on the semantic association degree of the entity pairs, retrieve multiple entries most relevant to it from the knowledge graph as relevant entries.
8. A traditional Chinese medicine standard diagnosis and treatment path and curative effect generation device, characterized in that, It includes: A data processing module configured to obtain text data related to TCM diagnosis and treatment and clean and screen the text data; A knowledge graph module configured to construct a knowledge graph using natural language processing technology based on the text data after cleaning and screening; An entry retrieval module configured to obtain the case record and retrieve relevant entries from the knowledge graph for the entities in the case record based on semantic similarity; A question and answer execution module configured to merge the relevant entries, the diagnosis and treatment records in the case record, and a preset question model and input them into a large language model to obtain corresponding standard diagnosis and treatment paths and efficacy criteria.
9. An electronic device, characterized in that, It includes a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1-7.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it realizes the steps of the method according to any one of claims 1-7.