Intelligent reasoning method and device for traditional Chinese medicine prescriptions and medium

Abstract

The invention discloses a traditional Chinese medicine prescription intelligent reasoning method and device and a medium, and relates to the cross technical field of artificial intelligence and the traditional Chinese medicine field. The method comprises the following steps: receiving patient diagnosis and treatment information submitted by a user, and constructing a corresponding multi-layer dynamic knowledge graph based on structured diagnosis and treatment information; positioning symptom nodes in the multilayer dynamic knowledge graph to calculate symptom feature vectors, and based on the symptom feature vectors, performing syndrome clustering analysis by using a preset syndrome-symptom causal inference network to determine to-be-analyzed syndromes with confidence greater than a preset threshold; and analyzing the to-be-analyzed syndromes through a preset treatment strategy generation network to generate a corresponding treatment strategy, and inputting the treatment strategy into the prescription generation model to determine a treatment prescription according with a monarch, minister, assistant and guide prescription principle.

Description

Technical Field

[0001] The present application relates to the interdisciplinary technical field of artificial intelligence and traditional Chinese medicine, and in particular to an intelligent reasoning method, device and medium for traditional Chinese medicine prescriptions. Background Art

[0002] In the process of TCM modernization, the deep integration of artificial intelligence technology and traditional TCM knowledge has become an important research direction. As a treasure of the Chinese nation, TCM contains rich theoretical knowledge and practical experience, but in practical applications, the expression of TCM knowledge, prescription generation, and clinical decision support still face many challenges.

[0003] The current intelligent diagnosis and treatment systems of traditional Chinese medicine have made some progress in the standardized expression of the theoretical system of traditional Chinese medicine and in assisting clinical decision-making through technologies such as knowledge graphs and deep learning. These systems can analyze the characteristics of syndromes and provide corresponding prescription suggestions. Thanks to advanced technologies such as knowledge graphs and deep generative models, these systems have shown good performance in handling standardized diagnosis and treatment tasks. However, when dealing with complex scenarios in clinical practice, the existing intelligent diagnosis and treatment systems of traditional Chinese medicine still have problems such as irrational prescription generation and insufficient adaptability to clinical practice.

[0004] Specifically, current methods often simplify prescription generation into a direct mapping from symptoms to prescriptions, ignoring the causal reasoning process in TCM syndrome differentiation and treatment, resulting in a lack of theoretical rationality in the generated prescriptions. In addition, due to insufficient consideration of individual differences among patients, the prescriptions generated by existing systems often fail to meet the actual needs of personalized diagnosis and treatment.

[0005] Therefore, how to effectively improve the ability to express complex TCM syndromes and improve the theoretical rationality and clinical applicability of prescription schemes has become a technical problem that needs to be solved urgently. Summary of the invention

[0006] The embodiments of the present application provide a method, device and medium for intelligent reasoning of TCM prescriptions to solve the following technical problems: how to effectively improve the ability to express complex TCM syndromes and improve the theoretical rationality and clinical applicability of prescription schemes.

[0007] In the first aspect, an embodiment of the present application provides an intelligent reasoning method for traditional Chinese medicine prescriptions, the method comprising: receiving patient diagnosis and treatment information submitted by a user, and constructing a corresponding multi-layer dynamic knowledge graph based on the structured diagnosis and treatment information; locating symptom nodes in the multi-layer dynamic knowledge graph to calculate symptom feature vectors, and based on the symptom feature vectors, using a preset syndrome-symptom causal inference network, performing syndrome clustering analysis to determine syndromes to be analyzed whose confidence is greater than a preset threshold; analyzing the syndromes to be analyzed through a preset treatment strategy generation network to generate corresponding treatment strategies, and inputting the treatment strategies into a prescription generation model to determine a treatment prescription that complies with the principle of monarch, minister, assistant and envoy prescription composition.

[0008] In one implementation of the present application, after receiving the patient diagnosis and treatment information submitted by the user, the method also includes: preprocessing the patient diagnosis and treatment information, specifically including: removing noise data, duplicate data, and obviously erroneous data in the diagnosis and treatment information; unifying and standardizing diagnosis and treatment information of different formats and standards, and converting unstructured diagnosis and treatment information into a structured data form.

[0009] In one implementation of the present application, the multi-layer dynamic knowledge graph includes: a representation layer, a syndrome layer, a treatment layer and a prescription layer; the representation layer is used to store information such as the patient's external symptoms, signs, etc.; the syndrome layer is used to represent the comprehensive judgment of traditional Chinese medicine on the disease state; the treatment layer contains treatment principles and methods for different syndromes; the prescription layer is used to store specific prescription information; based on the structured diagnosis and treatment information, a corresponding multi-layer dynamic knowledge graph is constructed, specifically including: according to the entities in the structured diagnosis and treatment information, corresponding nodes are created in each level, and the association relationship between different nodes is determined.

[0010] In one implementation of the present application, symptom nodes are located in a multi-layer dynamic knowledge graph to calculate a symptom feature vector, specifically including: searching for matching symptom nodes in the representation layer of the multi-layer dynamic knowledge graph based on preprocessed patient symptom information; extracting adjacent node information of the symptom nodes in the knowledge graph, and analyzing the attributes and relationships of the adjacent nodes to obtain contextual information of the symptom nodes; using graph embedding technology to map the symptom nodes and their adjacent information to a low-dimensional vector space, and calculating the symptom feature vector.

[0011] In one implementation of the present application, based on the symptom feature vector, a preset syndrome-symptom causal inference network is used to perform syndrome clustering analysis to determine the syndromes to be analyzed whose confidence is greater than a preset threshold, specifically including: inputting the symptom feature vector into the syndrome-symptom causal inference network to calculate the posterior probability of the occurrence of each syndrome based on the prior probability and conditional probability; wherein the syndrome-symptom causal inference network is obtained through training based on the Bayesian causal network model, the prior probability is pre-set based on traditional Chinese medicine theory and a large amount of clinical data, and the conditional probability is dynamically calculated based on the symptom feature vector and the training parameters of the network; a density-based clustering algorithm is used to perform cluster analysis on the calculated posterior probability, and a confidence calculation is performed on each clustering result to determine the syndromes to be analyzed whose confidence is greater than a preset threshold.

[0012] In one implementation of the present application, the treatment strategy generation network adopts a deep neural network architecture and introduces a knowledge guidance module; the knowledge guidance module integrates expert knowledge and clinical experience in the medical field and stores them in the form of a knowledge matrix. The knowledge matrix includes a theoretical rule matrix, a compatibility relationship matrix and a clinical experience matrix. The theoretical rule matrix stores the correspondence between syndromes and treatments in traditional Chinese medicine theory, and the compatibility relationship matrix records the compatibility contraindications and synergistic effects between drugs to ensure the safety and effectiveness of the drug combination in the generated treatment strategy. The clinical experience matrix contains a large number of successful treatment strategies in actual clinical cases, providing a reference for the network to generate treatment strategies; through the preset treatment strategy generation network, the syndrome to be analyzed is analyzed to generate the corresponding treatment strategy, specifically including: inputting the characteristic vector of the syndrome to be analyzed into the treatment strategy generation network, and generating the corresponding treatment strategy according to the constraints of the knowledge guidance module and its own learning ability.

[0013] In one implementation of the present application, a prescription generation model is constructed using a deep generation framework in combination with graph neural networks and variational reasoning technology; the treatment strategy is input into the prescription generation model to determine a treatment prescription that conforms to the principle of monarch, minister, assistant and envoy prescription composition, specifically including: encoding the treatment strategy through a multimodal encoder to generate a unified feature representation, and using a hierarchical decoder to sample and generate a candidate set of monarch drugs from the latent space; in the candidate set, the monarch drug to be used is screened out according to the preset constraints of the knowledge guidance module; wherein the monarch drug is a drug that plays a major therapeutic role for the main disease or main symptom; based on the monarch drug to be used, an analysis is performed through a hierarchical decoder to determine the minister drug and adjuvant drug to be used; wherein the minister drug is a drug that assists the monarch drug to enhance the efficacy, or treats concurrent diseases or concurrent symptoms; the main function of the adjuvant and envoy drugs is to harmonize the medicinal properties, restrict the toxicity or side effects of the monarch drug and the minister drug, and guide the drugs directly to the site of the disease.

[0014] In one implementation of the present application, the method further includes: generating a detailed explanation for the generated therapeutic prescription based on TCM theory and information in the knowledge graph.

[0015] In a second aspect, an embodiment of the present application also provides an intelligent reasoning device for traditional Chinese medicine prescriptions, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute an intelligent reasoning method for traditional Chinese medicine prescriptions such as any one of the above items.

[0016] On the third aspect, an embodiment of the present application also provides a non-volatile computer storage medium for intelligent reasoning of traditional Chinese medicine prescriptions, which stores computer executable instructions. When the computer executable instructions are executed, a method for intelligent reasoning of traditional Chinese medicine prescriptions such as any of the above is implemented.

[0017] The present application provides a method, device and medium for intelligent reasoning of traditional Chinese medicine prescriptions, which have the following features:

[0018] Beneficial effects:

[0019] 1. Improve knowledge processing and expression capabilities: (1) Construct accurate knowledge graphs: By preprocessing and structuring diagnosis and treatment information, a multi-layer dynamic knowledge graph including the representation layer, syndrome layer, treatment method layer and prescription layer is constructed. This graph can systematically and comprehensively integrate TCM knowledge, create nodes at each level and determine the association relationship based on the entities in the diagnosis and treatment information, and clearly show the multi-level and internal connection of TCM knowledge. For example, the patient's symptoms, signs and other representation information are matched with TCM syndrome judgment, treatment principles and specific prescriptions, which overcomes the defects of traditional knowledge graphs in expressing the systematic and dynamic nature of TCM knowledge and provides a solid foundation for subsequent accurate reasoning. (2) Enhance knowledge extraction and representation: When calculating the symptom feature vector, by searching for symptom nodes in the representation layer, analyzing the adjacent node information to obtain the context, and using graph embedding technology to map to a low-dimensional vector space, the key features of the symptoms can be effectively extracted and represented. Compared with traditional methods, this method can better capture the complex relationship between symptoms, provide richer and more accurate information for accurate syndrome judgment, and improve the quality of knowledge representation.

[0020] 2. Optimize the reasoning process and accuracy: (1) Syndrome analysis based on causal inference: The syndrome-symptom causal inference network trained based on the Bayesian causal network model combines the prior probability and the dynamically calculated conditional probability to calculate the posterior probability of each syndrome, and determines the syndrome to be analyzed through cluster analysis. This method simulates the clinical dialectical thinking process of traditional Chinese medicine, fully considers the causal relationship between symptoms and syndromes, and can more accurately judge the syndrome type of patients compared with the simple symptom-prescription direct mapping, providing a reliable basis for the subsequent formulation of reasonable treatment strategies. (2) Knowledge-guided treatment method generation: The treatment method strategy generation network adopts a deep neural network architecture and introduces a knowledge-guided module that integrates expert knowledge and clinical experience. This module stores theoretical rules, compatibility relationships and clinical experience in the form of a knowledge matrix. When generating treatment strategies, it analyzes according to the constraints of the knowledge-guided module and the network's own learning ability to ensure that the generated treatment strategy is consistent with traditional Chinese medicine theory and takes into account the safety and effectiveness of drug combinations. At the same time, it draws on a large number of clinical successful cases to improve the scientificity and practicality of the treatment strategy.

[0021] 3. Improve the quality of prescription generation: (1) Prescription construction that follows the principle of prescription composition: The prescription generation model uses a deep generative framework combined with graph neural networks and variational reasoning technology to generate therapeutic prescriptions based on the principle of monarch, minister, assistant and envoy prescription composition. Through a multimodal encoder and a hierarchical decoder, the monarch, minister and assistant drugs are determined in turn to ensure that each drug in the prescription plays a different role in treatment and cooperates with each other, which is in line with the theory of Chinese medicine prescription compatibility. This generation method improves the rationality and effectiveness of the prescription, making it more in line with actual clinical needs. (2) Prescription screening based on knowledge constraints: In the process of prescription generation, drugs are screened according to the preset constraints of the knowledge guidance module to further optimize the prescription. For example, the compatibility relationship matrix is ​​referred to to avoid drug incompatibility and ensure the safety of the prescription; the theoretical rule matrix and the clinical experience matrix are used to ensure the efficacy of the prescription, thereby improving the overall quality of the prescription.

[0022] 4. Improve clinical application value: (1) Meet personalized diagnosis and treatment needs: In the entire reasoning process, the patient's individual diagnosis and treatment information, including symptoms, signs, etc., is fully considered. Symptom analysis, treatment strategy formulation and prescription generation are carried out according to the specific conditions of different patients to achieve personalized diagnosis and treatment. For example, for patients of different ages and constitutions, even if the symptoms are similar, prescriptions suitable for their individual differences can be generated, which improves the treatment effect. (2) Enhance the explainability of diagnosis and treatment: Based on the information in traditional Chinese medicine theory and knowledge graphs, a detailed explanation is generated for the generated treatment prescription. This not only helps doctors understand the basis for the formulation of the prescription and improves their trust in the results of intelligent reasoning, but also makes it easier for patients to understand the treatment plan, enhances patients' compliance with treatment, and improves the explainability and credibility of intelligent diagnosis and treatment of traditional Chinese medicine.

[0023] 5. Promote the inheritance and development of TCM: (1) Assist in the inheritance of TCM knowledge: The method of the present invention combines TCM theoretical knowledge with modern technology, integrates TCM knowledge in a structured and digital way, and forms a knowledge map that can be processed and used by computers. This provides a new model for the inheritance of TCM knowledge, facilitates future generations to learn and study TCM's dialectical treatment thinking and prescription compatibility rules, and promotes the inheritance and development of TCM knowledge. (2) Promote the intelligentization of TCM: The application of this method can improve the efficiency and accuracy of TCM diagnosis and treatment, and provide technical support for the development of TCM intelligent diagnosis and treatment systems. In scenarios such as TCM clinics and telemedicine, it helps doctors quickly obtain accurate diagnosis and treatment recommendations, improves the quality and coverage of TCM medical services, and promotes the innovative development of TCM in the era of intelligent medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 A flow chart of an intelligent reasoning method for a traditional Chinese medicine prescription provided in an embodiment of the present application;

[0026] Figure 2 A schematic diagram of the internal structure of a TCM prescription intelligent reasoning device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be clearly and completely described below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0028] The embodiments of the present application provide a method, device and medium for intelligent reasoning of TCM prescriptions to solve the following technical problems: how to effectively improve the ability to express complex TCM syndromes and improve the theoretical rationality and clinical applicability of prescription schemes.

[0029] The technical solution proposed in the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0030] Figure 1 A flow chart of an intelligent reasoning method for TCM prescriptions provided in the embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides an intelligent reasoning method for traditional Chinese medicine prescriptions, which specifically includes the following steps:

[0031] Step 101: Receive patient diagnosis and treatment information submitted by the user, and build a corresponding multi-layer dynamic knowledge graph based on the structured diagnosis and treatment information.

[0032] In one embodiment of the present application, the system receives patient diagnosis and treatment information through a user interaction interface. The information may come from a hospital information system, an online diagnosis and treatment platform, etc., and is in various formats, such as text, images, electronic medical records, etc. For example, an electronic medical record containing basic patient information (name, gender, age), symptom description (such as cough, sputum, fatigue, etc.), and examination report (tongue image photo, pulse data, etc.) is received.

[0033] Furthermore, the diagnosis and treatment information is preprocessed. Use data cleaning algorithms to remove noise data, such as irrelevant characters entered by mistake and data in the wrong format; identify and delete duplicate data through duplicate checking algorithms; correct or delete obviously erroneous data according to preset rules, such as negative age. At the same time, use natural language processing technology, data format conversion tools, etc. to convert unstructured text information (such as symptom descriptions) into structured data, unify data of different formats and standards, such as standardizing the names and units of test indicators in different hospitals.

[0034] In one embodiment of the present application, the multi-layer dynamic knowledge graph includes: a representation layer, a syndrome layer, a treatment layer and a prescription layer; the representation layer is used to store information such as the patient's external symptoms and signs; the syndrome layer is used to represent the comprehensive judgment of the disease state by traditional Chinese medicine; the treatment layer contains treatment principles and methods for different syndromes; the prescription layer is used to store specific prescription information; based on the structured diagnosis and treatment information, a corresponding multi-layer dynamic knowledge graph is constructed, specifically including: according to the entities in the structured diagnosis and treatment information, corresponding nodes are created in each level, and the association relationship between different nodes is determined

[0035] In one embodiment, a multi-layer dynamic knowledge graph is constructed based on the pre-processed structured diagnosis and treatment information. Assume that the diagnosis and treatment information contains the symptom of "cough", the syndrome of "lung qi deficiency", the treatment method of "tonifying the lungs and replenishing qi", and the prescription of "Yupingfeng Powder". Create a "cough" node in the representation layer to record the symptom-related attributes (such as cough frequency, degree, etc.); create a "lung qi deficiency" node in the syndrome layer, and associate the "cough" and other related symptom nodes in the representation layer; create a "tonifying the lungs and replenishing qi" node in the treatment method layer, and establish a "targeted treatment" relationship with the "lung qi deficiency" node; create a "Yupingfeng Powder" node in the prescription layer, and connect it to the "tonifying the lungs and replenishing qi" treatment method node, indicating that the prescription is used to implement this treatment method, thereby constructing nodes at each level and their associations.

[0036] Step 102: locate the symptom node in the multi-layer dynamic knowledge graph to calculate the symptom feature vector, and based on the symptom feature vector, use the preset syndrome-symptom causal inference network to perform syndrome clustering analysis to determine the syndrome to be analyzed whose confidence is greater than a preset threshold.

[0037] In one embodiment of the present application, after constructing the corresponding multi-layer dynamic knowledge graph, the symptom nodes are located in the multi-layer dynamic knowledge graph to calculate the symptom feature vector.

[0038] Specifically, according to the preprocessed patient symptom information, the matching symptom nodes are searched in the representation layer of the multi-layer dynamic knowledge graph; the adjacent node information of the symptom nodes in the knowledge graph is extracted, and the attributes and relationships of the adjacent nodes are analyzed to obtain the contextual information of the symptom nodes; the graph embedding technology is used to map the symptom nodes and their adjacent information to the low-dimensional vector space, and the symptom feature vector is calculated.

[0039] Furthermore, based on the symptom feature vector, a syndrome clustering analysis is performed using a preset syndrome-symptom causal inference network to determine the syndromes to be analyzed whose confidence level is greater than a preset threshold.

[0040] Specifically, the symptom feature vector is input into the syndrome-symptom causal inference network to calculate the posterior probability of each syndrome according to the prior probability and conditional probability; wherein, the syndrome-symptom causal inference network is trained based on the Bayesian causal network model, the prior probability is pre-set based on traditional Chinese medicine theory and a large amount of clinical data, and the conditional probability is dynamically calculated according to the symptom feature vector and the training parameters of the network; a density-based clustering algorithm is used to perform cluster analysis on the calculated posterior probability, and a confidence calculation is performed on each clustering result to determine the syndromes to be analyzed whose confidence is greater than a preset threshold.

[0041] In one embodiment, according to the pre-processed patient symptom information, the symptom nodes matching it are searched in the representation layer of the multi-layer dynamic knowledge graph. If the patient has the symptoms of "cough" and "shortness of breath", the corresponding "cough" and "shortness of breath" nodes are found in the representation layer. The adjacent node information of the symptom node is extracted, and its attributes and relationships are analyzed to obtain context information. The adjacent node of the "cough" node may have a "lung qi deficiency syndrome" syndrome node. By analyzing the "prompt possible suffering from" relationship between the two and the "lung qi deficiency syndrome" node attributes, the association information between cough and lung qi deficiency syndrome is obtained. Using graph embedding technology (such as Node2Vec algorithm), the symptom nodes and their adjacent information are mapped to a low-dimensional vector space, and the symptom feature vector is calculated. For example, the symptom nodes such as "cough" and "shortness of breath" and their adjacent information are mapped to low-dimensional vectors such as [0.1, 0.3, -0.2, ...] as features for subsequent analysis. The symptom feature vector is input into the syndrome-symptom causal inference network obtained by training based on the Bayesian causal network model. Assuming that the probability of "lung qi deficiency syndrome" in a certain population is 0.1 in the prior probability, the conditional probability is dynamically calculated based on the symptom feature vector and the network training parameters, such as the conditional probability of "lung qi deficiency syndrome" under the symptoms of "cough" and "shortness of breath". Then calculate the posterior probability of each syndrome. Use a density-based clustering algorithm (such as the DBSCAN algorithm) to perform cluster analysis on the calculated posterior probability, and calculate the confidence of each clustering result based on the compactness and distribution of each syndrome in the clustering result. Set the confidence threshold to 0.7. If the confidence of the clustering result of "lung qi deficiency syndrome" is 0.8, it is determined to be the syndrome to be analyzed.

[0042] Step 103: Analyze the syndrome to be analyzed through the preset treatment strategy generation network to generate the corresponding treatment strategy, and input the treatment strategy into the prescription generation model to determine the treatment prescription that conforms to the principle of monarch, minister, assistant and envoy prescription composition.

[0043] In one embodiment of the present application, the treatment strategy generation network adopts a deep neural network architecture and introduces a knowledge guidance module; the knowledge guidance module integrates expert knowledge and clinical experience in the field of medicine and stores it in the form of a knowledge matrix. The knowledge matrix includes a theoretical rule matrix, a compatibility relationship matrix, and a clinical experience matrix. The theoretical rule matrix stores the correspondence between syndromes and treatments in traditional Chinese medicine theory, and the compatibility relationship matrix records the compatibility contraindications and synergies between drugs to ensure the safety and effectiveness of the drug combination in the generated treatment strategy. The clinical experience matrix contains a large number of successful treatment strategies in actual clinical cases, providing a reference for the network to generate treatment strategies. The prescription generation model adopts a deep generation framework and combines graph neural networks and variational reasoning technology to build.

[0044] In one embodiment of the present application, after the syndrome to be analyzed, the characteristic vector of the syndrome to be analyzed is input into the treatment strategy generation network, and the corresponding treatment strategy is generated according to the constraints of the knowledge guidance module and its own learning ability.

[0045] Furthermore, the treatment strategy is input into the prescription generation model to determine the treatment prescription that conforms to the principle of monarch, minister, assistant and envoy prescription composition.

[0046] Specifically, the treatment strategy is encoded through a multimodal encoder to generate a unified feature representation, and a hierarchical decoder is used to sample and generate a candidate set of monarch drugs from the latent space; in the candidate set, the monarch drug to be used is screened out according to the preset constraints of the knowledge guidance module; wherein, the monarch drug is a drug that plays a major therapeutic role for the main disease or main syndrome; based on the monarch drug to be used, analysis is performed through a hierarchical decoder to determine the minister drug and adjuvant drug to be used; wherein, the minister drug is a drug that assists the monarch drug to enhance the efficacy, or treats concurrent diseases or concurrent syndromes; the main function of the adjuvant drug is to harmonize the medicinal properties, restrict the toxicity or side effects of the monarch drug and the minister drug, and guide the drugs directly to the site of the disease.

[0047] In one embodiment, the treatment strategy generation network adopts a deep neural network architecture and introduces a knowledge guidance module. The feature vector of the syndrome "lung qi deficiency syndrome" to be analyzed is input into the treatment strategy generation network. The theoretical rule matrix in the knowledge guidance module indicates that "lung qi deficiency syndrome" corresponds to the "tonifying lung and replenishing qi" treatment method; the compatibility relationship matrix ensures that the subsequent drug combination is safe and effective; and the clinical experience matrix provides a reference for successful treatment strategies in similar cases. Based on these constraints and its own learning ability, the network generates a treatment strategy of "tonifying lung and replenishing qi, relieving cough and relieving asthma". The prescription generation model adopts a deep generation framework and combines graph neural network and variational reasoning technology to build. The "tonifying lung and replenishing qi, relieving cough and relieving asthma" treatment strategy is encoded by a multimodal encoder to generate a unified feature representation, and a hierarchical decoder is used to sample from the latent space to generate a set of monarch drug candidates, such as ["Huangqi" and "Ginseng"]. According to the preset constraints of the knowledge guidance module (such as theoretical rules, drug characteristics, etc.), "Huangqi" is selected from the candidate set as the monarch drug because of its outstanding effect in tonifying lung and replenishing qi. Based on the main drug "Huangqi", the hierarchical decoder is used to further analyze and determine the ministerial drugs and adjuvant drugs. For example, "Baizhu" is determined as the ministerial drug to assist Huangqi in enhancing the effect of nourishing the lungs and replenishing qi, and "Saposhnikovia divaricata" is the adjuvant drug, which can not only restrain the warming and drying nature of Huangqi, but also guide the drug to the lung meridian.

[0048] In one embodiment of the present application, the method further includes: generating a detailed explanation for the generated therapeutic prescription based on TCM theory and information in the knowledge graph.

[0049] In one embodiment, combined with TCM theory, it is explained that "Huangqi" as the main drug greatly replenishes lung qi; "Baizhu" as the minister drug helps Huangqi replenish qi and strengthen the spleen, cultivate soil and produce gold; "Saposhnikovia" is an adjuvant drug, which dispels wind and relieves exterior symptoms and harmonizes the medicinal properties. The association between each drug and syndrome and treatment method is obtained from the knowledge graph, such as the correspondence between "Huangqi", "Baizhu", "Saposhnikovia" and "Lung Qi Deficiency Syndrome" and "Tonifying Lung and Invigorating Qi" treatment methods, and a detailed explanation is generated: "This prescription is for lung qi deficiency syndrome, and adopts the treatment method of tonifying lung and replenishing qi. The main drug Huangqi greatly replenishes lung qi, the minister drug Baizhu assists Huangqi to enhance the power of replenishing qi, and the adjuvant Saposhnikovia dispels wind and relieves exterior symptoms and harmonizes the medicinal properties. The combination of these drugs has the effect of tonifying lung and replenishing qi, relieving cough and relieving asthma."

[0050] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, this embodiment of the application also provides a Chinese medicine prescription intelligent reasoning device, whose structure is as follows: Figure 2 shown.

[0051] Figure 2 This is a schematic diagram of the internal structure of a TCM prescription intelligent reasoning device provided in the embodiment of the present application. Figure 2 As shown, the device includes:

[0052] at least one processor 201;

[0053] and, a memory 202 communicatively connected to the at least one processor;

[0054] The memory 202 stores instructions that can be executed by at least one processor, and the instructions are executed by at least one processor 201 to enable at least one processor 201 to:

[0055] Receive patient diagnosis and treatment information submitted by users, and build a corresponding multi-layer dynamic knowledge graph based on the structured diagnosis and treatment information;

[0056] Locate symptom nodes in the multi-layer dynamic knowledge graph to calculate symptom feature vectors, and based on the symptom feature vectors, use the preset syndrome-symptom causal inference network to perform syndrome clustering analysis to determine the syndromes to be analyzed whose confidence is greater than the preset threshold;

[0057] Through the preset treatment strategy generation network, the syndrome to be analyzed is analyzed to generate the corresponding treatment strategy, and the treatment strategy is input into the prescription generation model to determine the treatment prescription that conforms to the principle of monarch, minister, assistant and envoy prescription composition.

[0058] Some embodiments of the present application provide corresponding Figure 1 A non-volatile computer storage medium for intelligent reasoning of a traditional Chinese medicine prescription stores computer executable instructions, wherein the computer executable instructions are set as follows:

[0059] Receive patient diagnosis and treatment information submitted by users, and build a corresponding multi-layer dynamic knowledge graph based on the structured diagnosis and treatment information;

[0060] Locate symptom nodes in the multi-layer dynamic knowledge graph to calculate symptom feature vectors, and based on the symptom feature vectors, use the preset syndrome-symptom causal inference network to perform syndrome clustering analysis to determine the syndromes to be analyzed whose confidence is greater than the preset threshold;

[0061] Through the preset treatment strategy generation network, the syndrome to be analyzed is analyzed to generate the corresponding treatment strategy, and the treatment strategy is input into the prescription generation model to determine the treatment prescription that conforms to the principle of monarch, minister, assistant and envoy prescription composition.

[0062] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the IoT device and medium embodiments, since they are basically similar to the method embodiments, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiments.

[0063] The system and medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the system and medium also have similar beneficial technical effects to the corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0064] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0065] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0066] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0067] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0068] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0069] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0070] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0071] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0072] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for intelligent reasoning of traditional Chinese medicine prescriptions, characterized in that: The method comprises: Receive patient diagnosis and treatment information submitted by users, and build a corresponding multi-layer dynamic knowledge graph based on the patient diagnosis and treatment information; Locating symptom nodes in the multi-layer dynamic knowledge graph to calculate symptom feature vectors, and based on the symptom feature vectors, using a preset syndrome-symptom causal inference network, performing syndrome clustering analysis to determine syndromes to be analyzed whose confidence is greater than a preset threshold; The syndrome to be analyzed is analyzed through a preset treatment strategy generation network to generate a corresponding treatment strategy, and the treatment strategy is input into a prescription generation model to determine a treatment prescription that complies with the principle of monarch, minister, assistant and envoy prescription composition.

2. According to claim 1, a method for intelligent reasoning of traditional Chinese medicine prescriptions is characterized in that: After receiving the patient diagnosis and treatment information submitted by the user, the method further includes: Preprocessing the patient's diagnosis and treatment information specifically includes: Remove noisy data, duplicate data, and obviously erroneous data from diagnosis and treatment information; Unify and standardize medical information of different formats and standards, and convert unstructured medical information into structured data form.

3. According to claim 1, a method for intelligent reasoning of traditional Chinese medicine prescriptions is characterized in that: The multi-layer dynamic knowledge graph includes: a surface layer, a syndrome layer, a treatment layer and a prescription layer; the surface layer is used to store information such as external symptoms and signs of patients; the syndrome layer is used to represent the comprehensive judgment of TCM on the disease state; the treatment layer contains treatment principles and methods for different syndromes; the prescription layer is used to store specific prescription information; Based on the patient diagnosis and treatment information, a corresponding multi-layer dynamic knowledge graph is constructed, specifically including: According to the entities in the structured diagnosis and treatment information, corresponding nodes are created in each level, and the association relationship between different nodes is determined.

4. According to claim 2, a method for intelligent reasoning of traditional Chinese medicine prescriptions is characterized in that: Locate symptom nodes in the multi-layer dynamic knowledge graph to calculate symptom feature vectors, including: According to the pre-processed patient symptom information, search for matching symptom nodes in the representation layer of the multi-layer dynamic knowledge graph; Extracting the adjacent node information of the symptom node in the knowledge graph, and analyzing the attributes and relationships of the adjacent nodes to obtain the context information of the symptom node; Graph embedding technology is used to map symptom nodes and their adjacency information into a low-dimensional vector space, and the symptom feature vector is calculated.

5. The intelligent reasoning method for traditional Chinese medicine prescription according to claim 1, characterized in that: Based on the symptom feature vector, a syndrome cluster analysis is performed using a preset syndrome-symptom causal inference network to determine the syndromes to be analyzed whose confidence is greater than a preset threshold, specifically including: The symptom feature vector is input into the syndrome-symptom causal inference network to calculate the posterior probability of each syndrome according to the prior probability and conditional probability; wherein the syndrome-symptom causal inference network is obtained by training based on the Bayesian causal network model, the prior probability is pre-set based on traditional Chinese medicine theory and a large amount of clinical data, and the conditional probability is dynamically calculated according to the symptom feature vector and the training parameters of the network; A density-based clustering algorithm is used to perform cluster analysis on the calculated posterior probability, and a confidence calculation is performed on each clustering result to determine the syndrome to be analyzed whose confidence is greater than a preset threshold.

6. The intelligent reasoning method for traditional Chinese medicine prescription according to claim 1, characterized in that: The governance strategy generation network adopts a deep neural network architecture and introduces a knowledge guidance module; The knowledge guidance module integrates expert knowledge and clinical experience in the field of medicine and stores them in the form of a knowledge matrix. The knowledge matrix includes a theoretical rule matrix, a compatibility relationship matrix and a clinical experience matrix. The theoretical rule matrix stores the correspondence between syndromes and treatment methods in TCM theory. The compatibility relationship matrix records the compatibility contraindications and synergies between drugs to ensure the safety and effectiveness of the drug combination in the generated treatment strategy. The clinical experience matrix contains a large number of successful treatment strategies in actual clinical cases, providing a reference for the network to generate treatment strategies. The syndrome to be analyzed is analyzed through a preset treatment strategy generation network to generate a corresponding treatment strategy, which specifically includes: The characteristic vector of the syndrome to be analyzed is input into the treatment strategy generation network, and the corresponding treatment strategy is generated according to the constraints of the knowledge guidance module and its own learning ability.

7. The intelligent reasoning method for traditional Chinese medicine prescription according to claim 1, characterized in that: The prescription generation model is constructed using a deep generation framework combined with graph neural networks and variational reasoning techniques; The treatment strategy is input into the prescription generation model to determine the treatment prescription that complies with the principle of monarch, minister, assistant and envoy prescription composition, specifically including: The treatment strategy is encoded through a multimodal encoder to generate a unified feature representation, and a hierarchical decoder is used to sample the candidate set of monarch drugs from the latent space; In the candidate set, the monarch drug to be used is screened out according to the preset constraints of the knowledge guidance module; wherein the monarch drug is a drug that plays a major therapeutic role for the main disease or main symptom; Based on the monarch drug to be used, analysis is performed through a hierarchical decoder to determine the minister drug and adjuvant drug to be used; wherein, the minister drug is a drug that assists the monarch drug to enhance the efficacy, or treats concurrent diseases or concurrent syndromes; the main function of the adjuvant drug is to harmonize the medicinal properties, restrict the toxicity or side effects of the monarch drug and the minister drug, and guide the drugs directly to the site of the disease.

8. The intelligent reasoning method for traditional Chinese medicine prescription according to claim 1, characterized in that: The method further comprises: Generate detailed explanations for the generated treatment prescriptions based on TCM theory and information in the knowledge graph.

9. A TCM prescription intelligent reasoning device, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; Wherein, the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a traditional Chinese medicine prescription intelligent reasoning method as described in any one of claims 1-8.

10. A non-volatile computer storage medium for intelligent reasoning of traditional Chinese medicine prescriptions, storing computer executable instructions, characterized in that: When the computer executable instructions are executed, an intelligent reasoning method for traditional Chinese medicine prescriptions as described in any one of claims 1 to 8 is implemented.

Images