Traditional Chinese medicine advantage disease data management system and method

Abstract

This invention relates to the field of TCM knowledge management technology, specifically to a TCM advantageous disease data management system and method. The solution acquires disease data from terminal nodes; constructs a multi-level TCM domain model as a semantic standard; performs entity recognition, relation extraction, and attribute filling on the disease data to obtain knowledge units; reconstructs discrete diagnosis and treatment events according to a timeline to obtain structured time-series data; and constructs an evaluation model based on this data for trend analysis and threshold comparison to identify disease samples with consistently excellent or significantly improved performance, generating evaluation reports; integrates knowledge units and disease samples to construct a dynamic knowledge network for clinical and research services, thereby building a knowledge graph; and integrates the evaluation reports and knowledge graph into clinical workstations to provide treatment plan recommendations and risk warnings for multiple terminal nodes. This technical solution integrates TCM data into a related knowledge graph and integrates it with evaluation reports into the application, resulting in a high practical conversion rate.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traditional Chinese medicine knowledge management, and in particular to a traditional Chinese medicine dominant disease data management system and method. BACKGROUND

[0002] At present, the digitization and informatization of traditional Chinese medicine ancient books have made great progress, forming several large-scale digital ancient book databases and retrieval platforms, which provide convenience for the preservation and retrieval of ancient books. The existing technologies mainly embody the following aspects: 1. basic literature digitization, that is, through scanning, OCR recognition and other technologies, paper ancient books are converted into electronic texts or images, realizing the electronic storage of the content and simple full-text retrieval based on keywords; 2. preliminary structured database construction, for example, relational databases established for some classic prescriptions and traditional Chinese medicines, which record basic information such as composition and indications, and support field query; 3. some research attempts to use text mining technologies, such as word frequency statistics and co-occurrence analysis, to mine the association rules of some drugs or symptoms from ancient book corpus. These works lay a foundation for the utilization of traditional Chinese medicine ancient book resources, enable the massive literature to be preserved and spread in digital form, and researchers no longer need to rely completely on manual flipping of paper literature, and the retrieval efficiency is improved to a certain extent.

[0003] However, the existing technical solutions are mostly in the data stage and far from intelligentization, and it is difficult to effectively support the precise needs of modern clinical and scientific research of traditional Chinese medicine dominant diseases. The main problem is that the data is in a serious "information island" and "knowledge fragmentation" state. The existing databases are mostly simple lists or independent storage of entries. The complex logical relationships between diseases, syndromes, treatments, prescriptions and medicines, the academic thought context of medical experts, the time sequence correlation of diagnosis and treatment experience evolution and other deep knowledge contained in ancient books have not been effectively extracted and associated. Knowledge is discrete and static, and cannot form an organic whole. Moreover, there is no precise knowledge reconstruction for modern clinical diseases, and most of them are narrative or broad retrieval of ancient book content, without taking "dominant diseases" with clear modern clinical value as the framework, systematically tracing back, combing and organizing relevant discussions scattered in different classics, resulting in a huge cognitive gap between ancient book knowledge and modern clinical practice, and low practical conversion rate. SUMMARY

[0004] Therefore, the purpose of the present application is to provide a traditional Chinese medicine dominant disease data management system and method to solve the problem of simple list management of traditional Chinese medicine data in the prior art, without forming an associated whole, and low practical conversion rate.

[0005] According to a first aspect of the present application, a traditional Chinese medicine dominant disease data management method is provided, comprising:

[0006] establishing connection with multi-terminal nodes, obtaining disease data from terminal nodes when receiving data synchronization instruction sent by terminal nodes or data request event triggered by management end; constructing multi-level TCM field model as semantic standard, performing entity recognition, relation extraction and attribute filling on the disease data to obtain knowledge units of symptoms, tongue and pulse and syndrome elements; reconstructing discrete diagnosis and treatment events according to time line to output structured time sequence data; constructing evaluation model according to the structured time sequence data, using the evaluation model to perform trend analysis on disease diagnosis and treatment data from different patients, comparing with multiple dimensions respectively according to threshold value based on trend analysis result to identify disease sample with continuous excellence or significant improvement, and generating evaluation report; fusing the knowledge units and the identified disease sample with continuous excellence or significant improvement to construct dynamic knowledge network for serving clinic and scientific research, and constructing knowledge graph containing disease, syndrome, prescription, medicine, curative effect, cost and rich relationship thereof according to TCM ontology framework of the dynamic knowledge network and filled disease sample; integrating the evaluation report and the knowledge graph into clinical workstation to provide diagnosis and treatment scheme recommendation and risk warning for multi-terminal nodes.

[0007] Preferably, the method for obtaining disease data from terminal nodes when receiving data synchronization instruction sent by terminal nodes or data request event triggered by management end comprises: establishing connection with multi-terminal nodes through pre-configured multi-type heterogeneous interface adapter; real-time monitoring of data synchronization instruction sent by each terminal node or data request event triggered by management end; when receiving data synchronization instruction, driving the adapter corresponding to the terminal node to collect disease data from the terminal node; when receiving data request event, driving the corresponding adapter according to the data request event to collect disease data from the corresponding terminal node; when collecting disease data, recording source, protocol type and triggering context of disease data, and injecting the disease data into unified receiving queue.

[0008] Preferably, the method further comprises: the obtained disease data comprises structured part, semi-structured part and unstructured part.

[0009] Preferably, the multi-level TCM field model comprises: the top layer of the multi-level TCM field model is TCM core theoretical framework, other layers are refined to specific disease, syndrome, prescription and medicine entity concepts, and semantic relationship between concepts is defined.

[0010] Preferably, entity recognition, relation extraction and attribute filling are performed on the disease data, including: A multi-level TCM field model is used as a semantic standard, and a natural language processing engine is used to analyze the text in the disease data: specific entities are identified and labeled, including symptoms, tongue and pulse signs; semantic relationships between syndromes and manifestations, and drugs and efficacy are extracted; and attribute information is extracted and filled into the corresponding entities.

[0011] Preferably, the disease sample with continuous excellent or significant improvement is identified by comparing with multiple dimensions respectively, including: The trend analysis result is automatically compared with a preset multi-dimensional threshold rule group; the multi-dimensional judgment conditions in the multi-dimensional threshold rule group include the improvement slope of curative effect, the stability of cost-benefit ratio, and the index compliance rate; According to the comparison result, the disease sample with continuous excellent or significant improvement is obtained.

[0012] Preferably, the trend analysis result is automatically compared with a preset multi-dimensional threshold rule group, including: The trend analysis result is automatically imported into a rule engine and compared with a preset multi-dimensional threshold rule group in the rule engine; through weighted calculation and logical judgment, the quantitative evaluation of whether each disease meets the advantage or significant improvement standard in each dimension is automatically completed.

[0013] Preferably, a dynamic knowledge network for serving clinical and scientific research is constructed, including: The entities of the knowledge unit are used as nodes, the semantic relationship is used as an edge, and the disease sample with continuous excellent or significant improvement is filled to construct a dynamic knowledge network.

[0014] Preferably, the method further comprises: The scheme adoption, actual curative effect and new clinical observation data are continuously collected from the application end of the multi-terminal node as feedback data; The built-in data comparison engine is used to calculate the difference between the recommended diagnosis and treatment scheme and the actual execution scheme in the key indicators in real time; According to the difference result, the parameter weight of the evaluation model is dynamically adjusted by using a reinforcement learning algorithm, and the confidence of the treatment-therapeutic effect relationship in the knowledge graph is optimized.

[0015] According to the second aspect of the present application, a TCM advantage disease data management system is provided, including: A data acquisition module is configured to establish a connection with a plurality of terminal nodes, and obtain disease data from the terminal nodes when receiving a data synchronization instruction sent by the terminal nodes or a data request event triggered by the management end; a data processing module, which constructs a multi-level TCM field model as a semantic standard, performs entity recognition, relation extraction and attribute filling on the disease data to obtain knowledge units of symptoms, tongue and pulse and syndrome elements, and reconstructs discrete diagnosis and treatment events according to a time line to output structured time sequence data; an evaluation identification module, which is configured to construct an evaluation model according to the structured time sequence data, perform trend analysis on disease diagnosis and treatment data from different patients by using the evaluation model, compare a threshold value with multiple dimensions respectively according to a trend analysis result, identify disease sample with sustained excellence or significant improvement, and generate an evaluation report; a knowledge graph module, which is configured to fuse the knowledge units and the identified disease sample with sustained excellence or significant improvement to construct a dynamic knowledge network for serving clinical and scientific research, and construct a knowledge graph containing diseases, syndromes, prescriptions, medicines, curative effects, costs and rich relationships thereof according to a TCM ontology skeleton of the dynamic knowledge network and the filled disease sample; an application integration module, which is configured to deeply integrate the evaluation report and the knowledge graph into a clinical workstation to provide diagnosis and treatment scheme recommendation and risk early warning for multiple terminal nodes.

[0016] The technical solution provided by the present application can include the following beneficial effects: It can be understood that the technical solution shown by the present application can obtain disease data from a terminal node, construct a multi-level TCM field model as a semantic standard to perform entity recognition, relation extraction and attribute filling on the disease data to obtain knowledge units, reconstruct discrete diagnosis and treatment events according to a time line to obtain structured time sequence data, construct an evaluation model according to the structured time sequence data to perform trend analysis and threshold value comparison, identify disease sample with sustained excellence or significant improvement, generate an evaluation report, fuse the knowledge units and the disease sample to construct a dynamic knowledge network for serving clinical and scientific research, and further construct a knowledge graph, and integrate the evaluation report and the knowledge graph into a clinical workstation to provide diagnosis and treatment scheme recommendation and risk early warning for multiple terminal nodes. The technical solution integrates TCM data into an associated knowledge graph, and integrates the evaluation report into an application end, so that the practical conversion rate is high.

[0017] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. BRIEF DESCRIPTION OF DRAWINGS

[0018] The accompanying drawings, which are incorporated into and form part of the specification, illustrate an embodiment consistent with the present application and, together with the specification, serve to explain the principles of the present application.

[0019] Figure 1 is a step schematic diagram of a TCM dominant disease data management method according to an exemplary embodiment; Figure 2 is a schematic block diagram of a traditional Chinese medicine dominant disease data management system according to an example embodiment. DETAILED DESCRIPTION

[0020] The example embodiments will be described in detail herein with reference to the attached drawings. In the following description, unless otherwise indicated, like numbers refer to like elements throughout the description and drawings. The following description of example embodiments does not represent all contemplated embodiments consistent with the present invention. Rather, they are merely examples of apparatus and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0021] In an embodiment, Figure 1 is a schematic block diagram of a traditional Chinese medicine dominant disease data management method according to an example embodiment, see Figure 1 A traditional Chinese medicine dominant disease data management method is provided, comprising: Step S11, a connection is established with a plurality of terminal nodes, and when a data synchronization instruction issued by a terminal node or a data request event triggered by a management end is received, disease data is obtained from the terminal node.

[0022] This step builds a flexible and scalable data access framework, which can seamlessly connect existing information systems of medical institutions with emerging data sources, ensuring real-time and continuous data collection, and providing a stable and high-quality data supply source for subsequent analysis steps.

[0023] Step S12, a multi-level traditional Chinese medicine field model is constructed as a semantic standard, entity recognition, relationship extraction and attribute filling are performed on the disease data, and knowledge units of symptoms, tongue and pulse and syndrome elements are obtained. This step realizes the qualitative change from raw data to computable knowledge.

[0024] Step S13, discrete diagnosis and treatment events are reconstructed according to a timeline, and structured time series data is output. The structured time series data output in this step not only contains entities and relationships, but also contains the dynamic rules of disease development, which is the core data form for precise efficacy evaluation and trend prediction.

[0025] Step S14, an evaluation model is constructed according to the structured time series data, the evaluation model is used to perform trend analysis on disease diagnosis and treatment data from different patients, and according to the trend analysis results, a plurality of dimensions are compared with threshold values respectively, disease samples with continuous excellent or significant improvement are identified, and an evaluation report is generated.

[0026] This step transforms qualitative assessments of advantages into quantitative, measurable, multi-dimensional comprehensive evaluations, freeing the identification of advantageous diseases from the traditional model of relying on expert subjective experience and conference discussions, and realizing a normalized, automated, data-driven dynamic selection process.

[0027] Step S15: Integrate the knowledge units with the identified disease samples that show continuous improvement or significant progress to construct a dynamic knowledge network for clinical and scientific research purposes; based on the TCM ontology framework of the dynamic knowledge network and the disease samples that are filled in, construct a knowledge graph that includes diseases, syndromes, prescriptions, medicines, efficacy, costs and their rich relationships.

[0028] Step S16: Integrate the assessment report and knowledge graph into the clinical workstation to provide treatment plan recommendations and risk warnings for multiple terminal nodes.

[0029] It is understood that the technical solution presented in this invention can acquire disease data from terminal nodes; construct a multi-level TCM domain model as a semantic standard to perform entity recognition, relation extraction, and attribute filling on disease data to obtain knowledge units; reconstruct discrete diagnosis and treatment events according to a timeline to obtain structured time-series data, and construct an evaluation model based on this to perform trend analysis and threshold comparison, identify disease samples that are consistently excellent or significantly improved, and generate evaluation reports; integrate knowledge units with disease samples to construct a dynamic knowledge network for clinical and research services, and then construct a knowledge graph; integrate the evaluation report and knowledge graph into a clinical workstation to provide treatment plan recommendations and risk warnings for multiple terminal nodes. This technical solution integrates TCM data into a related knowledge graph and integrates it with the evaluation report into the application, resulting in a high practical conversion rate.

[0030] In a preferred embodiment, in step S11, establishing a connection with multiple terminal nodes, and when receiving a data synchronization command from a terminal node or a data request event triggered by the management terminal, obtaining disease data from the terminal node, including: S111. Establish connections with multiple terminal nodes through pre-configured heterogeneous interface adapters of various types.

[0031] For example, by configuring various heterogeneous interface adapters such as HL7, FHIR, and custom APIs, secure connections can be established with multiple terminal nodes of hospital information systems or research platforms.

[0032] S112. Monitor in real time the data synchronization commands issued by each terminal node, or the data request events triggered by the management terminal.

[0033] S113. When a data synchronization instruction is received, the adapter corresponding to the terminal node is driven to collect disease data from the terminal node.

[0034] This step involves real-time monitoring of data synchronization commands from each terminal node, driving the corresponding adapter to automatically initiate disease-related data collection tasks. This design achieves intelligent scheduling and on-demand triggering of collection tasks, avoiding resource waste caused by polling and ensuring connection security and stability in complex network environments.

[0035] S114. When a data request event is received, the corresponding adapter is driven according to the data request event to collect disease data from the corresponding terminal node.

[0036] After receiving a data request event triggered by the management terminal, this step parses the data request event and drives the corresponding adapter to perform the task of collecting disease data based on the terminal node indicated in it.

[0037] When collecting disease data, the source, protocol type and triggering context of the disease data are recorded, and the disease data is injected into a unified receiving queue.

[0038] In this embodiment, after a data acquisition task is triggered, the corresponding adapter extracts data from the target node according to its connection protocol specifications. The acquired raw data stream is then injected into a unified receiving queue. Before being injected into the unified receiving queue, its source, protocol type, and triggering context are recorded to ensure the traceability of the task status for each data acquisition. This establishes a complete data lineage graph, allowing any piece of data to be traced back to its source and triggering conditions, greatly enhancing data governance capabilities and the feasibility of quality auditing, and providing an underlying guarantee for data credibility.

[0039] Preferably, the method further includes: the acquired disease data includes structured parts, semi-structured parts, and unstructured parts.

[0040] In practice, disease data presents a complete spectrum in terms of type. Its structured portion explicitly includes diagnostic and drug standard codes and laboratory numerical fields; the semi-structured portion is reflected in the paragraph-based descriptions and tables in the medical records; and the unstructured portion extensively covers free text of medical records, tongue photographs, pulse waveforms, and imaging data. Together, they constitute a complete information set from defined indicators to subjective experience descriptions. This comprehensive data scope ensures a holistic and multi-faceted portrayal of disease states and treatment processes, especially incorporating informative but difficult-to-quantify experience descriptions and imaging features from traditional Chinese medicine diagnosis and treatment into the analysis system.

[0041] The complete information set is closely related in clinical sense, and includes structured diagnostic codes, semi-structured syndrome differentiation descriptions, and unstructured tongue coating images. The three corroborate each other to present a TCM diagnostic and treatment information map that combines disease, syndrome, and form, providing a diverse and three-dimensional data foundation for subsequent in-depth analysis. The correlation and integration of "disease-syndrome-form" information is the key to simulating the comprehensive judgment thinking of TCM experts, and creates the necessary conditions for building a computational model that can understand the holistic concept and syndrome differentiation and treatment ideas of TCM.

[0042] In a preferred embodiment, step S12 uses a multi-level TCM domain model as a semantic standard to drive the natural language processing engine to perform fine-grained entity recognition, relation extraction, and attribute filling on automatically acquired disease data, and accurately extract knowledge units of symptoms, tongue and pulse, and syndrome elements.

[0043] In step S12, a multi-level TCM domain model is constructed, including: the top layer of the multi-level TCM domain model is the core theoretical framework of TCM based on the viscera and meridians, and other layers are refined to specific entity concepts of diseases, syndromes, prescriptions, and medicines, and the semantic relationships between concepts are defined.

[0044] This TCM domain model acts as a "translator" and "benchmark," mapping the chaotic natural language expressions to a standardized knowledge system. It fundamentally solves the problems of ambiguity and synonymy in TCM terminology, providing a unified semantic foundation for machine understanding of TCM language.

[0045] In a preferred embodiment, entity recognition, relation extraction, and attribute filling are performed on the disease data, including: Using a multi-level TCM domain model as a semantic standard, a natural language processing engine is used to parse the text in the disease data: identify and label specific entities, including symptoms, tongue and pulse signs; extract the semantic relationships of syndrome-manifestation and drug-efficacy; extract attribute information and fill it into the corresponding entities.

[0046] In practice, this system can accurately identify and label specific symptoms, tongue and pulse signs from disease descriptions, and then extract semantic relationships between syndromes and manifestations, and between drugs and their effects. Simultaneously, attribute information such as dosage and frequency is used as attributes to populate the corresponding entities. This achieves high-precision conversion from free text to structured knowledge units. This process automates the previously manual work of organizing and structuring medical records by experts, significantly improving efficiency and ensuring the consistency and objectivity of knowledge extraction, thus making the construction of a large-scale, high-quality TCM knowledge base possible.

[0047] Step S13 introduces time series modeling, which reconstructs discrete diagnosis and treatment events into a continuous diagnosis and treatment journey according to the timeline, accurately depicts the dynamic evolution of disease progression and prescription, and outputs structured time series data.

[0048] Step S12 has already converted the disease data into knowledge units, and step S13, based on step S12, sorts the diagnosis and treatment events corresponding to all disease data in chronological order, thereby sorting the structured knowledge units into structured time-series data.

[0049] Step S14 generates an evaluation model through sequence analysis of structured time-series data to dynamically track indicator trends, and uses threshold judgment to automatically identify diseases that are consistently excellent or significantly improved from multiple dimensions, and generates an evaluation report marked as an advantageous disease or a potential advantageous disease.

[0050] In a preferred embodiment, step S14 involves comparing thresholds with multiple dimensions to identify disease samples that show consistently excellent or significantly improved outcomes, including: The trend analysis results are automatically compared with a preset multi-dimensional threshold rule set. The multi-dimensional judgment conditions in the multi-dimensional threshold rule set include the slope of efficacy improvement, cost-effectiveness stability, and indicator achievement rate. Based on the comparison results, disease samples with consistently excellent or significantly improved outcomes are obtained.

[0051] This step involves building and running an evaluation model based on structured time-series data. The model processes time-stamped efficacy and cost data from different patients using a unified time-series alignment method. Through trend fitting and pattern recognition, it dynamically tracks the changes and evolution of core indicators for each disease across different institutions and time periods. This dynamic tracking capability based on time series can keenly capture subtle changes in the effectiveness, plateau, or fluctuations of treatment regimens, thereby more accurately assessing the short-term and long-term impacts of treatment measures, rather than relying solely on a simple comparison of start and end points.

[0052] The trend analysis results output by the evaluation model are automatically compared with a set of preset multi-dimensional threshold rules. Furthermore, by comprehensively considering factors such as the slope of efficacy improvement, cost-effectiveness stability, and indicator achievement rate, the model automatically filters and identifies target subjects that meet the definitions of advantageous or potential diseases from disease groups that consistently demonstrate excellent performance or show significant positive trends. The introduction of multi-dimensional threshold rules shifts the evaluation from solely focusing on "efficacy" to taking into account both health economics value and the quality of medical services, guiding the evaluation of advantageous diseases towards a more scientific and comprehensive direction and supporting management decision-making.

[0053] In a preferred embodiment, the automatic comparison with a preset multi-dimensional threshold rule set includes: The trend analysis results are automatically imported into the rule engine, where they are compared with a preset multi-dimensional threshold rule group. Through weighted calculation and logical judgment, the quantitative assessment of whether each disease meets the criteria for advantage or significant improvement in each dimension is automatically completed.

[0054] The use of the rules engine makes the assessment logic transparent and configurable. Managers or experts can flexibly adjust the weights and judgment thresholds of each dimension according to policy guidance or changes in clinical understanding, so that the assessment system has good adaptability and guidance.

[0055] Preferably, after completing automated comparison and comprehensive evaluation, the comprehensive evaluation supports experts in adjusting the parameters and threshold weights of the assessment model online. The results are centrally presented in a visual dashboard, showing the overall situation and rankings. Users can further select any disease to view its specific assessment dimension scores and the underlying typical case group data, achieving a fundamental shift from relying on subjective experience to dynamic measurement based on objective data. The visualization and downward selection functions provide an intuitive and interactive decision support interface, not only presenting the results but also revealing the data basis behind them, achieving data transparency and enhancing the persuasiveness of the assessment conclusions and the pertinence of management measures.

[0056] Step S15 integrates knowledge units with automatically identified diseases showing sustained improvement or significant progress, constructing a dynamic knowledge network serving clinical practice and research. This dynamic knowledge network uses traditional Chinese medicine (TCM) as its framework and patient case data as its filler, building a knowledge graph encompassing diseases, syndromes, prescriptions, medications, efficacy, costs, and their rich relationships. This step creates a "smart brain" that integrates static theoretical knowledge with dynamic practical evidence; it is not only a repository of knowledge but also a production tool capable of making connections, reasoning, and discovering new knowledge.

[0057] In this step, a dynamic knowledge network for clinical and scientific research purposes is constructed, including: using the entities of the knowledge units as nodes, semantic relationships as edges, and identifying disease samples that show continuous improvement or significant improvement as fillers to construct the dynamic knowledge network.

[0058] In practice, deeply structured knowledge units are integrated with automatically identified advantageous diseases and their multidimensional assessment conclusions to construct a dynamic knowledge network with entities as nodes and semantic and efficacy-related multi-dimensional relationships as edges. This dynamic knowledge network directly serves clinical decision support and research hypothesis discovery. This integration ensures that each "treatment-efficacy" relationship in the knowledge graph is accompanied by strong evidence (such as effectiveness rate and cost-effectiveness ratio), upgrading traditional qualitative experience associations to quantitative evidence associations, greatly enhancing the reference value of knowledge in clinical decision-making.

[0059] Understandably, the knowledge graph constructed in step S15 not only supports basic "disease-symptom-prescription" queries, but also performs complex multi-hop reasoning. For example, new diagnostic and treatment data is continuously imported after processing, driving the graph algorithm to automatically discover or verify potential relationships. At the same time, the evaluation conclusions of advantageous diseases are fed back to the graph as strong evidence labels, enhancing the weight and credibility of specific relationships in real time. This dynamic evolution mechanism makes the knowledge graph a "living" organism that can self-update and self-improve with the accumulation of new evidence, constantly approaching the true diagnostic and treatment patterns, effectively solving the problem of traditional knowledge bases easily becoming outdated and rigid.

[0060] Step S16 deeply integrates the assessment report and knowledge graph into the clinical workstation, providing real-time, personalized, and interpretable treatment recommendations and risk warnings for multiple terminal nodes.

[0061] This step also includes: continuously collecting data on treatment plan adoption, actual efficacy, and new clinical observations from application terminals across multiple terminals as feedback data; using a built-in data comparison engine to calculate in real time the differences between the recommended treatment plan and the actual implemented plan on key indicators; and based on the differences, dynamically adjusting the parameter weights of the evaluation model using reinforcement learning algorithms to optimize the confidence of relevant treatment-efficacy relationships in the knowledge graph. This continuously improves the accuracy of recommendations and the level of system intelligence.

[0062] This technical solution completes the final closed loop from data analysis to clinical value realization, and uses the feedback generated by the application as fuel for system evolution, building an intelligent ecosystem that can continuously learn and optimize.

[0063] In practice, the clinical workstation continuously collects data on protocol adoption, actual efficacy, and new clinical observations as feedback data. This feedback data is then re-integrated into the various multi-terminal nodes of the front-end data collection layer via a secure channel, forming a closed-loop data flow. A built-in data comparison engine calculates in real-time the differences between the recommended and actual protocols on key indicators. This feedback loop ensures that the system remains closely synchronized with real-world clinical practice, using doctors' actual choices and patients' actual outcomes as the most important optimization signals, ensuring that the system's optimization direction always serves to improve actual treatment outcomes.

[0064] Based on the calculated differential data, reinforcement learning algorithms are used to dynamically adjust the parameter weights of the evaluation model and optimize the confidence of relevant treatment-efficacy relationships in the knowledge graph. This allows the evaluation system and knowledge network to continuously adapt to the complex changes in the real world, constantly improving the accuracy of subsequent recommendations and the overall adaptive intelligence level of the system. The introduction of reinforcement learning endows the system with the ability to "learn from practice," enabling its recommendation strategy to be dynamically adjusted based on the success and failure of historical recommendations. The ultimate goal is to make the system's recommendations consistent with best clinical practice, achieving precision and personalization in AI-assisted diagnosis and treatment.

[0065] This technical solution breaks through the foundational data level by constructing an intelligent closed loop of assessment, knowledge, and optimization, achieving a leap from data management to intelligent decision-making. It integrates a dynamic quantitative assessment model with multi-dimensional indicators and causal inference, and for the first time comprehensively models dimensions such as efficacy, economics, and unique advantages. This fundamentally transforms the identification of advantageous diseases from expert experience-based judgment to data-driven identification. The constructed dynamic knowledge graph integrating efficacy evidence and the closed-loop optimization mechanism of reinforcement learning enable the system to continuously evolve based on real-world feedback. On the one hand, it provides clinicians with interpretable and personalized precise recommendations, directly improving the quality and consistency of diagnosis and treatment; on the other hand, it provides hospital administrators and researchers with powerful tools for dynamic monitoring, evidence-based decision-making, and the discovery of new knowledge, comprehensively promoting the scientific management, inheritance, innovation, and value transformation of advantageous diseases in traditional Chinese medicine.

[0066] On the other hand, see Figure 2 This provides a data management system for diseases with advantages in traditional Chinese medicine, including: The data acquisition module 101 is used to establish a connection with multiple terminal nodes. When it receives a data synchronization command from a terminal node or a data request event triggered by the management terminal, it acquires disease data from the terminal node. Data processing module 102 constructs a multi-level TCM domain model as a semantic standard, performs entity recognition, relation extraction and attribute filling on the disease data, and obtains knowledge units of symptoms, tongue and pulse and syndrome elements; it reconstructs discrete diagnosis and treatment events according to timeline and outputs structured time series data. The assessment and identification module 103 is used to construct an assessment model based on the structured time-series data, use the assessment model to perform trend analysis on disease diagnosis and treatment data from different patients, compare the trend analysis results with thresholds in multiple dimensions, identify disease samples that are consistently excellent or significantly improved, and generate an assessment report. The knowledge graph module 104 is used to integrate the knowledge units with the identified disease samples that are consistently excellent or significantly improved, and to construct a dynamic knowledge network for clinical and scientific research purposes; based on the TCM ontology framework of the dynamic knowledge network and the disease samples that are filled in, a knowledge graph containing diseases, syndromes, prescriptions, medicines, efficacy, costs and their rich relationships is constructed. Application integration module 105 is used to deeply integrate assessment reports and knowledge graphs into clinical workstations, providing treatment plan recommendations and risk warnings for multiple terminal nodes.

[0067] In summary, this technical solution, through the construction of a dynamic normalization acquisition mechanism based on multiple adapters such as HL7 and FHIR, achieves automated aggregation and real-time cleaning of multi-source heterogeneous data, including in-hospital medical records, scientific research data, ancient books and documents, and IoT devices. It fundamentally solves the problems of scattered data sources and inconsistent standards. With the help of deep structured processing based on fusion time-series modeling, it transforms unstructured medical case texts and tongue and pulse images into machine-understandable, time-stamped knowledge units, laying a precise data foundation for objective analysis. This not only breaks the limitations of traditional reliance on manual sorting and subjective review, but also enables massive and messy TCM data to be transformed into calculable and traceable high-value assets.

[0068] At the foundational data level, by constructing an intelligent closed loop of assessment, knowledge, and optimization, a leap from data management to intelligent decision-making has been achieved. A dynamic quantitative assessment model integrating multi-dimensional indicators and causal inference has, for the first time, comprehensively modeled dimensions such as efficacy, economics, and unique advantages, fundamentally transforming the identification of advantageous diseases from expert experience-based judgment to data-driven identification. The constructed dynamic knowledge graph integrating efficacy evidence and the closed-loop optimization mechanism of reinforcement learning enable the system to continuously evolve based on real-world feedback. On the one hand, it provides clinicians with interpretable and personalized precise recommendations, directly improving the quality and consistency of diagnosis and treatment; on the other hand, it provides hospital administrators and researchers with powerful tools for dynamic monitoring, evidence-based decision-making, and the discovery of new knowledge, comprehensively promoting the scientific management, inheritance, innovation, and value transformation of advantageous diseases in traditional Chinese medicine.

[0069] It is understood that the same or similar parts in the above embodiments can be referred to each other, and the contents not described in detail in some embodiments can be referred to the same or similar contents in other embodiments.

[0070] It should be noted that in the description of this invention, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance. Furthermore, in the description of this invention, unless otherwise stated, "a plurality of" means at least two.

[0071] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

[0072] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0073] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0074] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0075] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0076] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0077] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A data management method for traditional Chinese medicine advantageous diseases, characterized in that, include: Establish connections with multiple terminal nodes, and obtain disease data from the terminal nodes when a data synchronization command is received from a terminal node or a data request event is triggered by the management terminal. A multi-level TCM domain model is constructed as a semantic standard. Entity recognition, relation extraction, and attribute filling are performed on the disease data to obtain knowledge units of symptoms, tongue and pulse, and syndrome elements. Discrete medical events are reconstructed according to the timeline to output structured time-series data; An evaluation model is constructed based on the structured time-series data, and the evaluation model is used to perform trend analysis on disease diagnosis and treatment data from different patients. Based on the trend analysis results, threshold comparisons are performed with multiple dimensions to identify disease samples that show sustained excellence or significant improvement, and an evaluation report is generated. The knowledge units are integrated with the identified disease samples that show continuous improvement or significant progress to construct a dynamic knowledge network for clinical and scientific research purposes. Based on the TCM ontology framework of the dynamic knowledge network and the disease samples that fill it, a knowledge graph containing diseases, syndromes, prescriptions, medicines, efficacy, costs and their rich relationships is constructed. The assessment report and knowledge graph are integrated into the clinical workstation to provide treatment plan recommendations and risk warnings for multiple terminal nodes.

2. The method according to claim 1, characterized in that, Establish connections with multiple terminal nodes. When a data synchronization command is received from a terminal node or a data request event triggered by the management terminal, obtain disease data from the terminal node, including: Establish connections with multiple terminal nodes through pre-configured heterogeneous interface adapters of various types; Real-time monitoring of data synchronization commands issued by each terminal node, or data request events triggered by the management terminal; When a data synchronization command is received, the adapter corresponding to the terminal node is driven to collect disease data from the terminal node. When a data request event is received, the corresponding adapter is driven according to the data request event to collect disease data from the corresponding terminal node; When collecting disease data, the source, protocol type and triggering context of the disease data are recorded, and the disease data is injected into a unified receiving queue.

3. The method according to claim 1, characterized in that, Also includes: The acquired disease data includes structured, semi-structured, and unstructured components.

4. The method according to claim 1, characterized in that, Constructing a multi-level model for the field of Traditional Chinese Medicine, including: The top layer of the multi-level TCM domain model is the core theoretical framework of TCM, while other layers are refined to specific entity concepts of disease, syndrome, prescription, and medicine, and the semantic relationships between concepts are defined.

5. The method according to claim 4, characterized in that, The data on the aforementioned diseases is subjected to entity recognition, relation extraction, and attribute filling, including: Using a multi-level TCM domain model as a semantic standard, a natural language processing engine is used to parse the text in the disease data: identify and label specific entities, including symptoms, tongue and pulse signs; extract the semantic relationships of syndrome-manifestation and drug-efficacy; extract attribute information and fill it into the corresponding entities.

6. The method according to claim 1, characterized in that, Threshold comparisons were performed against multiple dimensions to identify disease samples that showed consistently excellent or significantly improved outcomes, including: The trend analysis results are automatically compared with a preset multi-dimensional threshold rule set; the multi-dimensional judgment conditions in the multi-dimensional threshold rule set include the slope of efficacy improvement, cost-effectiveness stability, and indicator achievement rate. Based on the comparison results, disease samples with consistently excellent or significantly improved outcomes were obtained.

7. The method according to claim 6, characterized in that, Automated comparison with preset multi-dimensional threshold rule sets, including: The trend analysis results are automatically imported into the rule engine, where they are compared with a preset multi-dimensional threshold rule group. Through weighted calculation and logical judgment, the quantitative assessment of whether each disease meets the criteria for advantage or significant improvement in each dimension is automatically completed.

8. The method according to claim 5, characterized in that, Constructing a dynamic knowledge network to serve clinical practice and research, including: By using the entities of the knowledge units as nodes, semantic relationships as edges, and identified disease samples that show continuous improvement or significant enhancement as fillers, a dynamic knowledge network is constructed.

9. The method according to claim 1, characterized in that, Also includes: We continuously collect data on protocol adoption, actual efficacy, and new clinical observations from multiple terminal nodes as feedback data. The built-in data comparison engine calculates in real time the differences in key indicators between the recommended treatment plan and the actual implementation plan; Based on the discrepancy results, the parameter weights of the evaluation model are dynamically adjusted using reinforcement learning algorithms to optimize the confidence of the relevant treatment-efficacy relationship in the knowledge graph.

10. A data management system for traditional Chinese medicine's advantageous diseases, characterized in that, include: The data acquisition module is used to establish connections with multiple terminal nodes. When it receives a data synchronization command from a terminal node or a data request event triggered by the management terminal, it obtains disease data from the terminal node. The data processing module constructs a multi-level TCM domain model as a semantic standard, performs entity recognition, relation extraction, and attribute filling on the disease data, and obtains knowledge units of symptoms, tongue and pulse, and syndrome elements. Discrete medical events are reconstructed according to the timeline to output structured time-series data; An assessment and identification module is used to construct an assessment model based on the structured time-series data and to perform trend analysis on disease diagnosis and treatment data from different patients using the assessment model. Based on the trend analysis results, threshold comparisons are performed with multiple dimensions to identify disease samples that show sustained excellence or significant improvement, and an evaluation report is generated. The knowledge graph module is used to integrate the knowledge units with the identified disease samples that show continuous improvement or significant progress, and to construct a dynamic knowledge network for clinical and scientific research purposes. Based on the TCM ontology framework of the dynamic knowledge network and the disease samples that fill it, a knowledge graph containing diseases, syndromes, prescriptions, medicines, efficacy, costs and their rich relationships is constructed. The application integration module is used to deeply integrate assessment reports and knowledge graphs into clinical workstations, providing treatment plan recommendations and risk warnings for multiple terminal nodes.

Images