A traditional Chinese medicine auxiliary decision-making method, device, medium and product

Abstract

This application relates to the field of traditional Chinese medicine (TCM) information technology, and discloses a method, device, medium, and product for TCM-assisted decision-making. The method includes: receiving the patient's four diagnostic methods (inspection, auscultation and olfaction); obtaining a first treatment plan text based on the four diagnostic methods and an intuitive reasoning channel; the intuitive reasoning channel is implemented based on a pre-trained TCM domain large language model, and the first treatment plan text is used to represent preliminary treatment suggestion text; obtaining a second set of treatment plans based on the four diagnostic methods and a logical reasoning channel; the logical reasoning channel is implemented based on a pre-constructed TCM diagnosis and treatment knowledge graph, and the second set of treatment plans is used to represent multiple treatment plans in the form of logical links; and obtaining TCM-assisted decision-making information based on the first treatment plan text and the second set of treatment plans.

Description

Technical Field

[0001] This application relates to the field of traditional Chinese medicine information technology, and in particular to a method, device, medium and product for assisting decision-making in traditional Chinese medicine. Background Technology

[0002] With the modernization of Traditional Chinese Medicine (TCM), due to the scarcity of high-quality TCM resources and the relatively weak service capacity of TCM at the grassroots level, intelligent methods are often used to assist TCM decision-making. This aims to solve the unique challenges of fuzzy semantic understanding and complex dialectical logical reasoning in TCM, establish relatively objective quantitative standards, and improve doctors' diagnostic and treatment skills. Currently, there are two main methods for intelligent TCM decision-making assistance: one is to generate results through large language models, and the other is to complete the decision-making assistance through symbolic reasoning systems based on knowledge graphs.

[0003] The inventors discovered at least the following technical problems in the related technologies: 1. Large language models suffer from decision-making black box and credibility issues. Because the model generation process is untraceable, the treatment methods and prescription suggestions in the generated results lack clear medical logical support. Doctors cannot understand the reasons for the model's output diagnosis. This opacity makes it difficult to trust the output results in complex or critical cases, hindering clinical application. 2. Large language models struggle to guarantee the logical rigor of the reasoning process. The model may produce internal logical contradictions or outputs that contradict classical TCM theories. For example, there may be conflicts between diagnoses and prescriptions, or even incompatible combinations. Simple post-processing rules cannot fundamentally repair logical loopholes in the reasoning process. 3. Knowledge graph symbolic reasoning systems lack flexibility and have limited capacity. They struggle to handle the large amount of fuzzy, unstructured descriptions in TCM diagnosis and treatment. Constructing a comprehensive TCM knowledge graph is a massive undertaking, and the graph itself cannot capture the subtle dialectical techniques and flexible prescription adjustments of renowned doctors. In conclusion, there is a lack of a framework that can balance the flexibility of large language models with the logical rigor of knowledge graphs to meet the high standards required for TCM auxiliary decision-making. Summary of the Invention

[0004] One objective of this application is to provide a method for assisting decision-making in Traditional Chinese Medicine, which at least addresses the technical problems of insufficient logical rigor and credibility in large language models, as well as the lack of flexibility and limited capacity of knowledge graph symbolic reasoning systems.

[0005] To achieve the above objectives, some embodiments of this application provide the following aspects:

[0006] In a first aspect, some embodiments of this application provide a TCM-assisted decision-making method, the method comprising: receiving a patient's four diagnostic methods (inspection, auscultation and olfaction); obtaining a first treatment plan text based on the four diagnostic methods and an intuitive reasoning channel; the intuitive reasoning channel being implemented based on a pre-trained TCM domain large language model, the first treatment plan text being used to represent preliminary treatment suggestion text; obtaining a second set of treatment plans based on the four diagnostic methods and a logical reasoning channel; the logical reasoning channel being implemented based on a pre-constructed TCM diagnosis and treatment knowledge graph, the second set of treatment plans being used to represent multiple treatment plans in the form of logical links; and obtaining TCM-assisted decision-making information based on the first treatment plan text and the second set of treatment plans.

[0007] Secondly, some embodiments of this application also provide an electronic device, the electronic device comprising: one or more processors; and a memory storing computer program instructions, which, when executed, cause the processor to perform the steps of the method described above.

[0008] Thirdly, some embodiments of this application also provide a computer-readable medium having computer program instructions stored thereon, which can be executed by a processor to implement the method described above.

[0009] Fourthly, some embodiments of this application also provide a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of the method described above.

[0010] Compared with related technologies, the solution provided in this application uses unstructured diagnostic information including chief complaint, symptoms, tongue and pulse characteristics, etc., as input, covering the patient's clinical information elements required for TCM syndrome differentiation and treatment. The intuitive reasoning channel utilizes the strong semantic generalization ability of a pre-trained TCM domain large language model to process the diagnostic information end-to-end, obtaining a first treatment plan text rich in clinical experience. The logical reasoning channel, based on the symbolic reasoning ability of a pre-constructed TCM diagnosis and treatment knowledge graph, extracts key information from the diagnostic information and performs path mapping of the graph to deduce a second set of treatment plans with rigorous logic. This dual-channel reasoning architecture, by integrating the two channels to independently produce complete treatment plans, obtains a TCM decision-making method, breaking the limitations of existing single models or serial process processing. It achieves a deep integration of data-driven TCM domain large language model and knowledge-driven knowledge graph, ensuring both the flexibility in processing unstructured diagnostic information and the rigor and interpretability of TCM theory. It solves the problems of credibility risk, lack of logical constraints, and rigidity of traditional graph systems that are difficult to adapt to changes in TCM clinical practice. Attached Figure Description

[0011] One or more embodiments are illustrated by way of example with reference numerals in the accompanying drawings. These illustrations do not constitute a limitation on the embodiments. Elements with the same reference numerals in the drawings are denoted as similar elements. Unless otherwise stated, the figures in the drawings are not to be limited by scale.

[0012] Figure 1 Exemplary flowcharts of TCM-assisted decision-making methods provided in some embodiments of this application;

[0013] Figure 2 This is a schematic diagram of an electronic device structure provided for some embodiments of this application. Detailed Implementation

[0014] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] The following terms are used in this document:

[0016] Dual-channel collaborative reasoning refers to the parallel processing architecture adopted by this system, which includes two independent and complementary reasoning paths.

[0017] Intuitive reasoning pathway, based on a large-scale pre-trained language model, outputs a diagnosis and treatment plan directly based on the input information through pattern recognition and generation capabilities. It is characterized by speed, flexibility, and the ability to handle unstructured descriptions, but the decision-making process is not transparent.

[0018] The logical reasoning channel, based on the reasoning path of structured knowledge graphs, derives the diagnosis and treatment plan by following the rule chain of "symptom-syndrome-treatment-prescription" through symbolic logic and relational traversal. It is characterized by rigor, traceability and conformity to explicit knowledge, but its flexibility is relatively insufficient.

[0019] The TCM diagnosis and treatment knowledge graph is a knowledge base in the field of TCM organized in a graph structure. Nodes represent entities, such as symptoms, syndromes, treatments, prescriptions, and Chinese medicines, while edges represent relationships between entities, such as "belongs to", "treatment", and "composition". Attributes such as weights can be attached, and it is the core knowledge source for logical reasoning channels.

[0020] Figure 1 A traditional Chinese medicine-assisted decision-making method is provided in some embodiments of this application, the method comprising:

[0021] S1. Receive the patient's four diagnostic methods information;

[0022] S2. Based on the information from the four diagnostic methods and the intuitive reasoning channel, the first treatment plan text is obtained; the intuitive reasoning channel is implemented based on a pre-trained large language model in the field of traditional Chinese medicine, and the first treatment plan text is used to represent the preliminary treatment suggestion text;

[0023] S3. Based on the information from the four diagnostic methods and the logical reasoning channel, a second set of treatment plans is obtained; the logical reasoning channel is implemented based on a pre-constructed TCM diagnosis and treatment knowledge graph, and the second set of treatment plans is used to represent multiple treatment plans in the form of logical links.

[0024] S4. Based on the first treatment plan text and the second treatment plan set, obtain TCM auxiliary decision-making information.

[0025] First, it should be noted that there are fundamental differences in the characteristics of diagnostic and treatment data between traditional Chinese medicine and Western medicine. Western medicine diagnosis and treatment mainly rely on objective quantitative indicators, such as blood pressure, blood sugar, and CT images, and the data structure is standardized and clear. In contrast, traditional Chinese medicine diagnosis and treatment rely on unstructured natural language descriptions, and its clinical terminology is ambiguous. In addition, traditional Chinese medicine theory is an abstract philosophical model, which is difficult to analyze through simple rules or calculations.

[0026] Specifically, regarding S1, the four diagnostic methods refer to the four methods of examining diseases in traditional Chinese medicine clinical practice: observation, auscultation, inquiry, and palpation. The information from the four diagnostic methods can include at least the chief complaint, medical history, symptoms, tongue appearance, and pulse appearance, and can cover the complete patient information required for traditional Chinese medicine diagnosis and treatment.

[0027] Specifically, for S2-S3, the intuitive reasoning channel receives information from the four diagnostic methods. Based on a pre-trained large language model in the field of traditional Chinese medicine, it performs end-to-end processing on the input information from the four diagnostic methods to generate a first treatment plan text that includes preliminary diagnostic conclusions, treatment principles and methods, recommended prescriptions and drugs.

[0028] The logical reasoning channel receives information from the four diagnostic methods (inspection, auscultation, palpation, and olfaction), performs symbolic reasoning based on a pre-built TCM diagnostic knowledge graph, extracts key information from the four diagnostic methods, maps the key information to graph nodes, performs path search in the graph, and constructs one or more diagnostic and treatment plans in the form of logical reasoning links. Each link represents a diagnostic and treatment possibility that conforms to classical TCM theory, and the collection of diagnostic and treatment plans serves as a second set of diagnostic and treatment plans.

[0029] By employing a dual-channel architecture where each channel operates independently, the flexibility of the intuitive reasoning channel and the rigor of the logical reasoning channel complement each other. On one hand, the intuitive reasoning channel compensates for the limitations of the TCM diagnostic knowledge graph in handling unknown terms or complex subjective descriptions; on the other hand, the logical reasoning channel compensates for potential logical leaps in the intuitive reasoning channel, providing verifiable objective evidence. This architectural design can handle complex and non-standard descriptions of illnesses while ensuring that the output conforms to the basic theories and logic of TCM, effectively avoiding hallucinatory outputs and common-sense errors.

[0030] Specifically, for S4, the results from the two channels are fused to generate an interpretable diagnosis and treatment report as TCM-assisted decision-making information. This TCM-assisted decision-making information may include: a final treatment plan, listing the diagnostic conclusions, treatment principles and methods, recommended prescription names, drug composition, and dosage suggestions; a multi-granularity interpretation path diagram, where the macro-reasoning path is displayed as a flowchart from the input symptoms to the final prescription, and micro-evidence associations are displayed at each node of the path providing supporting evidence; channel contribution indicators, visually identifying the main sources of each part of the final plan (such as intuitive reasoning channels, logical reasoning channels, or fused generation); and, if conflict resolution occurs, a dedicated area explaining the content of the conflict and the reasons for adopting the current plan. The report output conforms to modern medical record standards and can be directly used as part of an electronic medical record, fundamentally transforming black-box decision-making into a transparent, auditable, and trustworthy record of clinical thinking, used to assist doctors in decision-making and doctor-patient communication.

[0031] For example, the macro-level reasoning path could be [symptom A, B] → [syndrome C] → [treatment D] → [prescription E] → [drug F, G, H]. The micro-level evidence connection is at the "syndrome C" node, indicating which key symptoms (A, B) are associated through the symptom-syndrome edge in the graph. At the "drug F" node, its efficacy against "symptom A" and its weight in the graph are noted. If there is a conflict resolution, the initial intuitive solution suggested by drug X will be output in a special area. However, this conflict with the incompatibilities Y recorded in the graph, so drug H is replaced according to the graph.

[0032] Compared with related technologies, the solution provided in this application uses unstructured diagnostic information including chief complaint, symptoms, tongue and pulse characteristics, etc., as input, covering the clinical information elements of patients required for TCM syndrome differentiation and treatment. The intuitive reasoning channel utilizes the strong semantic generalization ability of a pre-trained TCM domain large language model to process the diagnostic information end-to-end, obtaining a first treatment plan text rich in clinical experience. The logical reasoning channel, based on the symbolic reasoning ability of a pre-constructed TCM diagnosis and treatment knowledge graph, extracts key information from the diagnostic information and performs path mapping of the graph to deduce a second treatment plan with rigorous logic. This dual-channel reasoning architecture, by integrating two channels to independently produce complete diagnosis and treatment plans, yields a TCM decision-making method. It breaks through the limitations of existing single models or serial process processing, achieving a deep integration of data-driven approaches from a large language model and knowledge-driven approaches from a knowledge graph in the TCM field. This ensures both the flexibility in processing unstructured diagnostic information and the rigor and interpretability of TCM theory. By outputting complete TCM auxiliary decision-making information that demonstrates the reasoning process, it transforms the decision-making black box into a visualized thinking process, enabling TCM doctors to examine the basis of each inference and greatly enhancing their confidence in clinical application.

[0033] In one embodiment, obtaining the second set of treatment plans based on the four diagnostic methods and logical reasoning includes:

[0034] Extract the set of key symptoms from the four diagnostic methods;

[0035] In the logical reasoning channel, based on the pre-constructed TCM diagnosis and treatment knowledge graph, the set of key symptoms is mapped to one or more syndrome hypotheses;

[0036] For each syndrome hypothesis, based on the pre-constructed TCM diagnosis and treatment knowledge graph, the corresponding treatment method node and prescription node are derived to form a reasoning link consisting of the syndrome node corresponding to the syndrome hypothesis, the treatment method node and the prescription node.

[0037] Associate the prescription nodes in each reasoning link and output the standard drug composition predefined in the knowledge graph for the prescription nodes;

[0038] One or more inference links that are associated with standard drugs are aggregated to form the second set of treatment options.

[0039] Specifically, the information from the four diagnostic methods (inspection, auscultation and olfaction, inquiry, and palpation) can be processed first, filtering out irrelevant modifiers and extracting symptoms with TCM significance as a set of key symptoms. Secondly, in a pre-constructed TCM diagnostic knowledge graph, a mapping from symptoms to syndromes is performed. Specifically, a search is conducted using predefined <symptom, manifestation, syndrome> relationship edges in the graph. Since the same symptom may correspond to multiple syndromes, one or more syndrome hypotheses will be obtained. Thirdly, in the pre-constructed TCM diagnostic knowledge graph, for each syndrome hypothesis, following the structure of <syndrome-corresponding treatment-commonly used formula>, deduction is performed along the association paths of <syndrome-corresponding treatment> and <corresponding treatment-commonly used formula> in the graph, generating one or more logical reasoning chains consisting of syndrome nodes, treatment nodes, and formula nodes corresponding to the syndrome hypothesis, and attaching the confidence level supported by the graph. Finally, for the prescription node at the end of the logical reasoning link, further query the <prescription-drug> relationship to obtain the standard drug composition under the prescription node, and package and summarize all the logical reasoning links generated above and their corresponding drug compositions to form a second set of treatment plans.

[0040] In this embodiment, a pre-constructed TCM diagnosis and treatment knowledge graph is used to ensure that each step of the reasoning process is supported by a clear graph path, and the reasoning process is entirely based on the TCM theoretical knowledge pre-entered in the graph, without any random probability.

[0041] In one embodiment, obtaining TCM-assisted decision-making information based on the first treatment plan text and the second treatment plan set, i.e., S4, includes:

[0042] S41. Perform structured parsing on the first treatment plan text to obtain a first structure tuple; each reasoning link in the second treatment plan set corresponds to a second structure tuple;

[0043] S42. Based on the first set of structural tuples and the second set of structural tuples, obtain the set of drugs to be tested;

[0044] S43. For the set of drugs to be tested, call the Traditional Chinese Medicine Contraindication Rule Engine to identify conflicts;

[0045] S44. Based on the conflict identification result, the first set of structural tuples and the second set of structural tuples, perform conflict resolution and / or consistency determination to obtain the TCM auxiliary decision-making information.

[0046] Specifically, for S41, the first treatment plan text can be presented using natural language, while the second treatment plan set is presented using a database-formatted link. Since their formats are different, they cannot be directly compared. Therefore, the first treatment plan text and the second treatment plan set need to be structurally aligned. The first treatment plan text generated by the intuitive channel is parsed into a structured [syndrome, treatment method, prescription, drug list] tuple through entity recognition and relation extraction, and this serves as the first structured tuple. Since the logical channel itself generates a reasoning link, field mapping and node expansion can be performed to obtain the structured [syndrome, treatment method, prescription, drug list] tuple.

[0047] Specifically, for S42, the drugs in the first structural tuple and the drugs in the second structural tuple are extracted. The resulting drug set to be tested not only includes the drugs within each tuple, but also the drug union after the two are combined. The drug union is used to detect the risk of poisoning caused by the combined drugs.

[0048] Specifically, for S43, this step performs conflict detection. By calling an independent Traditional Chinese Medicine (TCM) contraindication rule engine in parallel, the set of drugs to be tested is scanned. If the set of drugs to be tested triggers a TCM contraindication rule, the identification result is a conflict; if the set of drugs to be tested does not trigger a TCM contraindication rule, the identification result is no conflict. The TCM contraindication rules include classic drug compatibility rules, drug antagonism rules, and special population contraindication rules. The specific conflict triggering mechanism can be: if any drug suggested by any channel or a combination of drugs from both channels triggers a contraindication rule, it is considered a conflict.

[0049] For example, classic drug compatibility rules may include "licorice is incompatible with Euphorbia kansui" and "veratrum is incompatible with ginseng." Rules regarding the combination of conflicting medicinal properties may include "using extremely cold or extremely hot drugs together without restraint." Contraindications for specific populations may include "pregnant women should not use blood-activating or blood-breaking drugs." When the intuitive channel suggests adding "licorice," while the logical channel deduces a formula containing "Euphorbia kansui," and the TCM contraindication rules identify "licorice is incompatible with Euphorbia kansui," then the identification result is a conflict.

[0050] Specifically, for S44, the conflict identification result of S43 is obtained. When the result is a conflict, conflict resolution is performed on the first set of structural tuples and the second set of structural tuples to obtain the TCM auxiliary decision information. When the result is not a conflict, consistency determination is performed on the first set of structural tuples and the second set of structural tuples to obtain the TCM auxiliary decision information.

[0051] Conflict resolution is a decision-making process that occurs when two conflicting approaches cannot be directly fused. Upon detection of a conflict, a multi-dimensional intelligent resolution strategy is employed, rather than simple rejection or reliance on manual intervention. This involves a comprehensive assessment of the evidence strength of each approach, combined with an adaptive judgment based on the individual patient, to generate a fusion or arbitration plan. This decision-making process facilitates making optimal decisions that align with clinical logic amidst disagreements, demonstrating high-level collaborative intelligence.

[0052] Among them, the consistency determination is used to quantitatively evaluate the similarity of the two-channel schemes in order to decide whether to directly integrate and adopt the two-channel scheme.

[0053] In this embodiment, structured alignment solves the problem of direct comparison and verification of data formats generated by dual channels. Traditional Chinese medicine (TCM) contraindication rules are placed at the forefront as the first standard for verifying the dual-channel generation scheme, ensuring a safety baseline for clinical medication. Finally, based on the conflict identification results of the TCM contraindication rules, a flexible processing flow—either conflict resolution or consistency determination—can be selected, adapting to complex TCM clinical decision-making scenarios.

[0054] In one embodiment, the step of performing conflict resolution based on the first set of structured tuples and the second set of structured tuples to obtain the TCM auxiliary decision-making information specifically includes:

[0055] Calculate the confidence score of the first treatment plan text and the path weight of the inference link in the second treatment plan set;

[0056] The path weights are adjusted by incorporating patient characteristic information to obtain the adjusted target path weights;

[0057] The TCM auxiliary decision-making information is obtained based on the target path weight and the confidence score.

[0058] Specifically, this embodiment illustrates the specific conflict resolution process. First, the credibility scores of the results generated by each of the two channels are calculated, that is, the supporting evidence strength of the results generated by the intuitive reasoning channel and the logical reasoning channel are evaluated respectively. On one hand, for the intuitive reasoning channel, the matching degree between the keywords of the first treatment plan text and the relevant patterns in the training corpus of the TCM domain large language model is analyzed, or the confidence score of the TCM domain large language model during reasoning is directly used; wherein, the higher the matching degree, the higher the reliability of the first treatment plan text, the higher the confidence score, and the more confident the TCM domain large language model. On the other hand, for the logical channel, since each logical link consists of multiple edges, each edge has a corresponding weight in the knowledge graph, and the weight represents the association strength or classicity. The edge weights can be multiplied or weighted to calculate the path weight of the reasoning link; the shorter the link, the higher the edge weight, and the more reliable the reasoning link.

[0059] Secondly, context-adaptive judgments and dynamic adjustments are made based on specific patient information. Specifically, based on patient tags containing information such as age, physical condition, and medical history, the solutions in the reasoning chain are checked to see if they fit the patient's individualized situation. The path weights are then dynamically adjusted to obtain the adjusted target path weights. For example, if the patient is elderly, although a reasoning chain containing a "severe purgative" formula may be logically correct, it is unsuitable for the patient, so the overall weight of that reasoning chain is reduced.

[0060] Next, for the quantitative scores of the two channels obtained in the first two steps, a decision-making operation is performed: if the target path weight is greater than the preset weight threshold, the reasoning link in the second set of treatment plans corresponding to the target path weight is adopted to generate the TCM auxiliary decision-making information;

[0061] If the confidence score is higher than the preset score threshold, and the target path weight is less than or equal to the weight threshold, then the first treatment plan text is corrected based on the TCM diagnosis and treatment knowledge graph to generate the TCM auxiliary decision-making information;

[0062] If the target path weight is less than or equal to the weight threshold, and the confidence score is less than or equal to the score threshold, then the TCM auxiliary decision-making information is generated based on a third-party knowledge base or manual review.

[0063] Among them, the weight threshold is used to measure the quality of the reasoning link obtained by the logical reasoning channel, and the scoring threshold is used to measure the reliability of the content generated by the intuitive reasoning channel.

[0064] Finally, a conflict and resolution log is recorded, documenting the content of the conflict, the basis for resolution, and the final decision throughout the process, for use in model iteration and auditing.

[0065] In this embodiment, the credibility scores of the results generated by each of the two channels are first calculated. Then, the score of the logical reasoning channel is adaptively corrected in combination with the patient profile. Finally, the corresponding decision-making operations are executed according to three different conflict scenarios. The three strategies of single-channel result, local correction or external assistance are flexibly selected to ensure the flexibility and safety of the conflict resolution process.

[0066] In one embodiment, the step of performing a consistency determination based on the first set of structured tuples and the second set of structured tuples to obtain the TCM auxiliary decision-making information specifically includes:

[0067] Obtain the dynamic consistency threshold in the scenario;

[0068] Calculate the semantic similarity and key element matching degree between the first structural tuple and the corresponding tuples in the second structural tuple set, and perform attribute consistency verification to obtain the attribute consistency result.

[0069] Based on the dynamic consistency threshold, combined with the semantic similarity, the key element matching degree, and the attribute consistency result, a collaborative consistency determination is performed to obtain the determination result;

[0070] If the determination result is consistent, then the corresponding tuple in the first set of structural tuples and / or the second set of structural tuples will be determined as the TCM auxiliary decision-making information.

[0071] If the determination result is inconsistent, conflict resolution is triggered, and the TCM auxiliary decision-making information is generated based on the first set of structural tuples and the second set of structural tuples.

[0072] Specifically, this embodiment illustrates the specific consistency determination process. The dynamic consistency threshold is used to determine the minimum score for consistency between the output results of the intuitive reasoning channel and the logical reasoning channel. The dynamic consistency threshold is not fixed, but is determined jointly based on the severity of the illness, age, and the confidence level of the knowledge graph.

[0073] After obtaining the dynamic consistency threshold, the output results of the two channels are compared from three dimensions: semantic similarity, key element matching degree, and attribute consistency. Semantic similarity can be calculated by calculating the cosine similarity of text vectors; the key element matching degree is mainly compared with structured fields, such as whether the syndrome names are the same or contain each other, and whether the main prescription is consistent; attribute consistency verification is used for qualitative detection in traditional Chinese medicine, and the two channels are compared to see if there are attribute conflicts by constructing TCM attribute conflict rules. Among them, the TCM attribute conflict rules can include treatment method attribute rules, formula / drug attribute rules, and syndrome attribute rules. For example, treatment method attributes can be warming, clearing, tonifying, purging, ascending, descending, dispersing, astringing, unblocking, and astringing; formula / drug attributes can be the four natures (cold, hot, warm, cool), the five flavors (pungent, sweet, sour, bitter, salty), the meridian tropism, and the efficacy category (warming yang, nourishing yin, clearing heat and drying dampness, etc.); syndrome attributes can be yin and yang, exterior and interior, cold and heat, deficiency and excess (eight principles), as well as disease location and nature labels such as viscera, qi, blood, and body fluids. In the TCM attribute conflict rules, warming yang and nourishing yin are mutually exclusive. Cold syndromes cannot use cold and cooling drugs, and excess heat syndromes should not be strongly tonified.

[0074] The obtained semantic similarity, key element matching degree, and dynamic consistency threshold are compared, and consistency is determined in conjunction with the attribute consistency verification results. When the semantic similarity is higher than the dynamic consistency threshold, it proves that the two channels are correct in their description; when the key element matching degree is higher than the dynamic consistency threshold, it proves that the two channels are correct in their entity representation; if there is no attribute conflict, the TCM auxiliary decision-making information is obtained directly, but if there is an attribute conflict, conflict resolution must be performed.

[0075] In this embodiment, by introducing a dynamic consistency threshold, the decision-making logic can flexibly switch between high fault tolerance and high security under different clinical urgency levels. By calculating semantic similarity and key element matching degree, the consistency of the results generated by the two channels is objectively quantified. Through attribute consistency verification, logical errors that appear to be correct on the surface but do not actually conform to traditional Chinese medicine theory can be identified. The collaborative and consistent judgment logic improves decision-making efficiency from the above three dimensions.

[0076] In one embodiment, the dynamic consistency threshold is calculated as follows:

[0077]

[0078] Among them, the basic threshold represents the initial judgment value set according to clinical guidelines or historical data; A factor indicating the severity of the illness, adjusted according to the urgency of the disease; This indicates an age-sensitive factor, which is adjusted for specific age groups; This represents the confidence level of the knowledge graph upon which the logical channel is based, reflecting the reliability of the reasoning path based on the knowledge graph, and its value ranges from [0,1]. , and This represents a configurable weighting coefficient used to balance the impact of different factors on the dynamic consistency threshold.

[0079] Specifically, the severity factor quantifies the urgency of the condition; the more severe the condition, the higher the value, leading to a higher threshold and stricter review. For example, the value for acute and critical illnesses is higher than that for chronic conditions or sub-health states. Furthermore, chronic conditions or sub-health states allow for greater flexibility, encouraging the intuitive approach to leverage empirical treatment advantages. The age sensitivity factor quantifies the risk associated with the patient's age; the value is higher for elderly or pediatric patients than for adult patients. The knowledge graph confidence score represents the extent or certainty of the knowledge graph's coverage of the current disease. Therefore, higher risks require higher consistency between the two channels, thereby reducing the risk of misdiagnosis.

[0080] In one embodiment, the step of performing a collaborative consistency determination based on the dynamic consistency threshold, combined with the semantic similarity, the key element matching degree, and the attribute consistency result, to obtain a determination result includes:

[0081] If the semantic similarity and key element matching degree of at least one tuple in the second set of structured tuples with the first set of structured tuples both exceed the dynamic consistency threshold, and the attribute consistency result is non-conflicting, then the determination result is consistent.

[0082] If the semantic similarity and key element matching degree of at least one tuple in the second structural tuple set with the first structural tuple both exceed the dynamic consistency threshold, and the attribute consistency result is in conflict, the determination result is inconsistent;

[0083] If the semantic similarity of all tuples in the second structural tuple set with the first structural tuple is lower than the dynamic consistency threshold, the determination result is inconsistent.

[0084] For example, taking an elderly patient with wind-cold cold as an example, first, according to the risk characteristics of his advanced age and physical weakness, a relatively high dynamic consistency threshold is obtained. The diagnosis and treatment plan obtained by the intuitive reasoning channel recommends using "Jingfang Baidu Powder", while the prescription deduced by the logical reasoning channel is "Jingfang Dabiao Decoction". When making a consistency determination, although their prescription names are different, their semantic similarity and key element matching degree both exceed the dynamic consistency threshold, and in the attribute consistency verification, both belong to the "pungent-warm exterior-resolving" method, and their medicinal properties are both "warm", without opposing conflicts. Therefore, it meets the condition that the determination result is consistent, and the fused diagnosis and treatment report is directly output as the traditional Chinese medicine auxiliary decision-making information; on the contrary, if the intuitive channel misjudges it as wind-heat cold and recommends the cold-natured "Yinqiao Powder", even if the score is high, it will be determined as in conflict due to the attribute conflict of "cold-heat property conflict", thus triggering conflict resolution, and the diagnosis and treatment report is output as the traditional Chinese medicine auxiliary decision-making information according to the result after conflict resolution.

[0085] The step divisions of the above various methods are only for clear description. When implemented, they can be combined into one step or some steps can be split into multiple steps. As long as the same logical relationship is included, it is within the protection scope of this application; adding insignificant modifications to the algorithm or process or introducing insignificant designs, but not changing the core design of its algorithm and process are all within the protection scope of this application.

[0086] In addition, some embodiments of this application also provide an electronic device. The electronic device can be various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and so on. The electronic device can also be various forms of mobile devices, such as cellular phones, smart phones, wearable devices, and other similar computing devices.

[0087] The electronic device includes: one or more processors; and a memory storing computer program instructions, and when the computer program instructions are executed, the processors execute the steps of the method provided in any one or more of the above embodiments. Figure 2An exemplary structural diagram of the electronic device is disclosed. The electronic device includes one or more processors 1101, a memory 1102, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components are interconnected via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the electronic device, including instructions stored in or on memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple electronic devices can be connected, each providing some of the necessary operations. The components, their connections and relationships, and their functions shown herein are merely examples and are not intended to limit the implementation of the present application described and / or claimed herein.

[0088] The electronic device may further include an input device 1103 and an output device 1104. The processor 1101, memory 1102, input device 1103 and output device 1104 may be connected by a bus or other means, as shown in the figure, which is connected by a bus.

[0089] Input device 1103 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the electronic device, such as a touch screen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 1104 may include a display device, auxiliary lighting device (e.g., LED), and haptic feedback device (e.g., vibration motor). The display device may include, but is not limited to, a liquid crystal display, a light-emitting diode display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0090] To provide interaction with the user, the electronic device can be a computer. The computer has: a display device (e.g., a cathode ray tube or LCD monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback); and input from the user can be received in any form (e.g., voice input or tactile input).

[0091] In this embodiment, a computer-readable medium stores a computer program / instructions that, when executed by a processor, implement the steps of the methods provided in any one or more of the above embodiments. This computer-readable medium may be included in the electronic device described in the above embodiments; or it may exist independently and not assembled into that device. The aforementioned computer-readable medium carries one or more computer-readable instructions.

[0092] The memory 1102 can serve as a non-transitory computer-readable storage medium, used to store non-transitory software programs, non-transitory computer-executable programs, and modules. The processor 1101 executes various functional applications and data processing of the server by running the non-transitory software programs, instructions, and modules stored in the memory 1102, thereby implementing the program instructions / modules corresponding to the methods provided in any one or more of the embodiments described above in this application.

[0093] The memory 1102 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device. Furthermore, the memory 1102 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, the memory 1102 may optionally include memory remotely located relative to the processor 1101, and these remote memories can be connected to the electronic device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0094] It should be noted that the computer-readable medium described in this application can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. Computer-readable media can be, for example, but not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections having one or more wires, portable computer disks, hard disks, random access memory, read-only memory, erasable programmable read-only memory, optical fibers, portable compact disk read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In this application, a computer-readable medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0095] Computer-readable media include permanent and non-permanent, removable and non-removable media, which can store information by any method or technology. 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, static random access memory, dynamic random access memory, other types of random access memory, read-only memory, electrically erasable programmable read-only memory, flash memory or other memory technologies, read-only optical discs, digital versatile optical discs or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0096] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including local area networks (LANs) or wide area networks (WANs), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0097] In the above embodiments, all or part of the implementation can be achieved through software, hardware, firmware, or any combination thereof. For example, it can be implemented using an application-specific integrated circuit (ASIC), a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of this application can be executed by a processor to implement the above steps or functions. Similarly, the software program of this application (including related data structures) can be stored in a computer-readable recording medium, such as RAM memory, magnetic or optical drives, floppy disks, and similar devices. In addition, some steps or functions of this application can be implemented in hardware, for example, as circuitry that cooperates with a processor to perform the various steps or functions.

[0098] The computer program product provided in this application includes one or more computer programs / instructions. When executed by a processor, these computer programs / instructions generate, in whole or in part, the processes or functions described in this application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium may be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive), etc.

[0099] The flowcharts or block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of devices, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-specific system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0100] The scope of this application is defined by the appended claims rather than the foregoing description, and is therefore intended to encompass all variations falling within the meaning and scope of equivalents of the claims. No reference numerals in the claims should be construed as limiting the scope of the claims. Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a device claim may also be implemented by a single unit or device in software or hardware. Terms such as "first," "second," etc., are used only for distinguishing descriptions and do not indicate any particular order, nor should they be construed as indicating or implying relative importance.

[0101] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily made by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims, and the above embodiments should be regarded as exemplary and non-limiting.

Claims

1. A TCM-assisted decision-making method, characterized in that, The method includes: Receive the patient's information from the four diagnostic methods (inspection, auscultation, palpation, and olfaction); Based on the information from the four diagnostic methods and the intuitive reasoning channel, the first treatment plan text is obtained; the intuitive reasoning channel is implemented based on a pre-trained large language model in the field of traditional Chinese medicine, and the first treatment plan text is used to represent the preliminary treatment suggestion text; Based on the information from the four diagnostic methods and the logical reasoning channel, a second set of treatment plans is obtained; the logical reasoning channel is implemented based on a pre-constructed TCM diagnosis and treatment knowledge graph, and the second set of treatment plans is used to represent multiple treatment plans in the form of logical links; Based on the first treatment plan text and the second treatment plan set, TCM auxiliary decision-making information is obtained.

2. The method according to claim 1, characterized in that, The second set of treatment plans, derived from the information obtained through the four diagnostic methods and logical reasoning, includes: Extract the set of key symptoms from the four diagnostic methods; In the logical reasoning channel, based on the pre-constructed TCM diagnosis and treatment knowledge graph, the set of key symptoms is mapped to one or more syndrome hypotheses; For each syndrome hypothesis, based on the pre-constructed TCM diagnosis and treatment knowledge graph, the corresponding treatment method node and prescription node are derived to form a reasoning link consisting of the syndrome node corresponding to the syndrome hypothesis, the treatment method node, and the prescription node. Associate the prescription nodes in each reasoning link and output the standard drug composition predefined in the knowledge graph for the prescription nodes; One or more inference links that are associated with standard drugs are aggregated to form the second set of treatment options.

3. The method according to claim 1, characterized in that, The step of obtaining TCM-assisted decision-making information based on the first treatment plan text and the second treatment plan set includes: The first treatment plan text is parsed in a structured manner to obtain a first structure tuple; each reasoning link in the second treatment plan set corresponds to a second structure tuple; Based on the first set of structural tuples and the second set of structural tuples, the set of drugs to be tested is obtained; For the set of drugs to be tested, a traditional Chinese medicine contraindication rule engine is invoked to identify conflicts; Based on the conflict identification results, the first set of structural tuples, and the second set of structural tuples, conflict resolution and / or consistency determination are performed to obtain the TCM auxiliary decision-making information.

4. The method according to claim 3, characterized in that, The step of performing conflict resolution or consistency determination based on the conflict identification result, the first set of structured tuples, and the second set of structured tuples to obtain the TCM auxiliary decision-making information includes: If the conflict identification result indicates that a conflict exists, then conflict resolution is performed based on the first set of structural tuples and the second set of structural tuples to obtain the TCM auxiliary decision-making information; If the conflict identification result indicates that there is no conflict, then a consistency determination is performed based on the first set of structural tuples and the second set of structural tuples to obtain the TCM auxiliary decision-making information.

5. The method according to claim 4, characterized in that, The process of performing conflict resolution based on the first set of structured tuples and the second set of structured tuples to obtain the TCM auxiliary decision-making information specifically includes: Calculate the confidence score of the first treatment plan text and the path weight of the inference link in the second treatment plan set; The path weights are adjusted by incorporating patient characteristic information to obtain the adjusted target path weights; The TCM auxiliary decision-making information is obtained based on the target path weight and the confidence score.

6. The method according to claim 4, characterized in that, The process of performing a consistency determination based on the first set of structured tuples and the second set of structured tuples to obtain the TCM auxiliary decision-making information specifically includes: Obtain the dynamic consistency threshold in the scenario; Calculate the semantic similarity and key element matching degree between the first structural tuple and the corresponding tuples in the second structural tuple set, and perform attribute consistency verification to obtain the attribute consistency result. Based on the dynamic consistency threshold, combined with the semantic similarity, the key element matching degree, and the attribute consistency result, a collaborative consistency determination is performed to obtain the determination result; If the determination result is consistent, then the corresponding tuple in the first set of structural tuples and / or the second set of structural tuples will be determined as the TCM auxiliary decision-making information. If the determination result is inconsistent, conflict resolution is triggered, and the TCM auxiliary decision-making information is generated based on the first set of structural tuples and the second set of structural tuples.

7. The method according to claim 6, characterized in that, The dynamic consistency threshold is calculated as follows: Among them, the basic threshold represents the initial judgment value set according to clinical guidelines or historical data; A factor indicating the severity of the illness, adjusted according to the urgency of the disease; This indicates an age-sensitive factor, which is adjusted for specific age groups; This represents the confidence level of the knowledge graph upon which the logical channel is based, reflecting the reliability of the reasoning path based on the knowledge graph, and its value ranges from [0,1]. , and This represents a configurable weighting coefficient used to balance the impact of different factors on the dynamic consistency threshold.

8. An electronic device, characterized in that, The electronic device includes: Field-programmable gate arrays; and A memory storing configuration data, wherein the field-programmable gate array is configured by the configuration data to form hardware logic circuitry to perform the steps of the method as described in any one of claims 1 to 7.

9. A computer-readable medium storing configuration data thereon, characterized in that, When the configuration data is loaded into the field-programmable gate array (FPGA), the internal configuration of the FPGA forms a logic circuit, and the steps of the method as described in any one of claims 1 to 7 are executed.

10. A computer program product, comprising hardware description language code or a netlist, characterized in that, The hardware description language code or netlist is used to generate configuration data for configuring the field-programmable gate array to perform the steps of the method according to any one of claims 1 to 7.

Images