A three-dimensional interactive diagnosis and treatment display method and system of traditional Chinese medicine treatment knowledge graph

Abstract

The present application relates to the field of traditional Chinese medicine information popularization technology, especially to a three-dimensional interactive diagnosis and treatment display method and system of traditional Chinese medicine treatment knowledge graph, which comprises constructing a multi-dimensional relationship treatment knowledge graph by fusing traditional Chinese medicine literature, clinical and pharmacological data; generating personalized recommended diagnosis and treatment schemes based on vector matching or recursive disassembly of user multi-modal feature information; and dynamically rendering the meridian and acupoint and drug action path in combination with the three-dimensional human digital twin model. The present application can realize the deep structuring, accurate intelligent recommendation and intuitive three-dimensional interactive display of traditional Chinese medicine data, and improve the diagnosis and treatment efficiency and visualization level.

Description

Technical Field

[0001] This invention belongs to the field of traditional Chinese medicine information popularization technology. Specifically, it relates to a three-dimensional interactive diagnosis and treatment display method and system for a knowledge graph of traditional Chinese medicine treatment. Background Technology

[0002] Traditional Chinese medicine (TCM) diagnosis relies on the four diagnostic methods of observation, auscultation and olfaction, inquiry, and palpation, with experienced TCM physicians making comprehensive judgments based on the patient's symptoms, tongue appearance, pulse, constitution, and other information. However, traditional TCM diagnosis has a certain degree of subjectivity; differences in the experience and cognitive abilities of different physicians can lead to inaccurate diagnostic results.

[0003] Currently, with the development of artificial intelligence, big data, and knowledge graph technologies, medical diagnosis is gradually moving towards intelligentization. However, current intelligent medical systems mainly focus on Western medicine, and the structured organization, intelligent matching, and precise recommendation of traditional Chinese medicine (TCM) data are still in the initial exploratory stage. Existing TCM intelligent diagnostic methods lack in-depth correlation analysis of multi-dimensional data such as TCM syndrome types, symptoms, prescriptions, and meridian treatments, making it difficult for the system to provide accurate and personalized treatment plans when dealing with complex conditions.

[0004] Therefore, there is an urgent need to invent a TCM diagnostic technology to efficiently construct a treatment knowledge graph and achieve precise matching between patient characteristics and TCM treatment plans. Summary of the Invention

[0005] In view of this, the present invention proposes a three-dimensional interactive diagnostic and treatment display method and system for a knowledge graph of traditional Chinese medicine treatment, aiming to solve the problems of existing TCM intelligent diagnostic methods lacking in-depth structured organization of TCM data and intelligent matching and accurate recommendation of multi-dimensional data, and lacking intuitive three-dimensional interactive display means, which leads to the inability to provide efficient and personalized diagnosis and treatment plans and visualized diagnosis and treatment guidance.

[0006] This invention proposes a three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment, including: Acquire traditional Chinese medicine information data, perform structured processing and feature extraction on the traditional Chinese medicine information data, and construct a multi-dimensional relational treatment knowledge graph; The system acquires multimodal feature information of the user to be treated, performs feature vector matching between the multimodal feature information and each preset feature information stored in the relational treatment knowledge graph, and determines the recommended treatment plan based on the matching calculation results. If the Euclidean distance between the multimodal feature information of the user to be treated and any feature vector among the preset feature information is less than a first preset threshold, then the treatment method mapped by the preset feature information in the relational treatment knowledge graph is determined to be a recommended treatment plan. If the Euclidean distance between the multimodal feature information of the user to be treated and each preset feature information is greater than or equal to the first preset threshold, then the multimodal feature information is decomposed according to the cosine similarity between it and each preset feature information to obtain preset feature information that matches each feature decomposition component, and a recommended treatment plan is reconstructed based on the topological relationship between the treatment methods mapped by each preset feature information in the relational treatment knowledge graph. The recommended treatment plan is obtained, and based on the meridian and acupoint data, pharmacokinetics data, and preset three-dimensional human digital twin model in the recommended treatment plan, three-dimensional interactive treatment display data is generated, and the three-dimensional interactive treatment display data is converted into a preset three-dimensional image output format for dynamic rendering output.

[0007] Furthermore, when acquiring information and data related to traditional Chinese medicine, this includes: Traditional Chinese medicine (TCM) literature information data is obtained through web crawler interfaces, clinical medical data is obtained from medical institution databases, and pharmacological data of Chinese medicinal herbs is obtained from pharmacopoeia databases. The TCM literature information data, clinical medical data, and pharmacological data of Chinese medicinal herbs are then aggregated to establish an original TCM information dataset.

[0008] Furthermore, data processing of the aforementioned traditional Chinese medicine information data includes: Based on the TCM literature information data, the clinical medical data, and the TCM pharmacology data, the following associations are established using entity relation extraction algorithms: disease-syndrome-treatment principle association, syndrome-prescription-TCM association, syndrome-food-medicine property association, prescription-drug-medicine property association, and meridian-acupoint-treatment association. The spatial distance metric between various correlations is obtained based on the Euclidean distance formula. A distance matrix is ​​established based on the spatial distance metric, and the correlation degree between each correlation formula is calculated based on the distance matrix to obtain the correlation degree weight value. Based on the aforementioned correlation weight values, each correlation is logically weighted to establish a set of treatment correlations.

[0009] Furthermore, when establishing various correlation formulas based on the aforementioned TCM literature information data, clinical medical data, and TCM pharmacological data, the following are included: Natural language processing technology is used to identify and remove incomplete text data and duplicate descriptive data from the TCM literature information data. Identify and remove duplicate case records and invalid case data that do not conform to preset logic from the clinical medical data; Identify and remove overlapping attribute data and missing parameter data from the pharmacological data of traditional Chinese medicine; Based on the mapping relationship between the TCM literature information data and the clinical medical data after data removal, establish the disease-syndrome-treatment principle association and the meridian-acupoint-treatment association; Based on the cross-reference relationships between the TCM literature information data, the clinical medical data, and the TCM pharmacology data after data removal, establish the syndrome type-prescription-TCM association and the syndrome type-food-drug property association; Based on the pharmacological data of the Chinese herbal medicines and the clinical medical data after data removal, a formula-drug-property correlation is established.

[0010] Furthermore, when constructing a relational therapy knowledge graph based on various treatment associations, it includes: Outlier detection is performed on each treatment association based on the isolated forest algorithm, and disease type clustering is performed on each treatment association based on the detection results; Extract the disease characteristic parameters of each treatment correlation in each disease type, and establish a linear regression model between each treatment correlation in the disease type; Based on the disease types, the treatment associations under each disease type, and the weight parameters determined by the linear regression model, a multi-level relational treatment knowledge graph is constructed.

[0011] Furthermore, when clustering various treatment associations for different disease types using the isolated forest algorithm, the following steps are included: Construct an isolated forest model consisting of multiple isolated trees, calculate the path length of each treatment correlation in the isolated forest model, and calculate the anomaly score based on the path length; When the abnormal score of the treatment association is greater than the preset abnormal threshold, the treatment association is determined to be an independent classification node, and the feature vector of the treatment association is extracted to establish a new disease type label. Obtain the disease information, physical characteristics information, tongue appearance information, and pulse appearance information of historical patients corresponding to the disease type label, and generate the central feature vector of the disease type through feature fusion algorithm; Calculate the cosine similarity between the pathological feature vector of each treatment association to be classified and the central feature vector of each disease type, and classify each treatment association into the disease type with the highest similarity based on the K-means clustering algorithm.

[0012] Furthermore, when obtaining the characteristic information of the user to be treated, specifically: The system acquires facial and tongue images of the patient to be treated through an image acquisition device, obtains pulse dynamics data through a pressure sensor array, and acquires the patient's medical history text information through a speech recognition device. Keyword extraction is performed on the described medical history text to obtain the physical characteristics information of the user to be treated.

[0013] Furthermore, when decomposing feature information according to its similarity to each preset feature information, it includes: Obtain the cosine similarity between each preset feature information vector and the feature information vector of the user to be treated; Multiple candidate preset feature information with similarity S greater than a preset similarity threshold are selected, and the multimodal feature information is decomposed into a first-level dimension based on the candidate preset feature information to obtain a first decomposed feature set; Each element in the first set of disassembly features is compared with each preset feature information, and the second level of disassembly is determined based on the comparison results.

[0014] Furthermore, when comparing the disassembled feature information with each preset feature information, and determining whether to further disassemble the disassembled feature information based on the comparison results, the process includes: If any element in the first set of disassembled features completely matches a feature vector in each of the preset feature information, then the disassembly operation of that element is stopped. If any element in the first set of decomposed features does not match any of the preset feature information, then the element is decomposed into a second-level dimension until the decomposed atomic feature information matches any of the preset feature information in the relational therapy knowledge graph.

[0015] Furthermore, when generating and outputting 3D interactive diagnostic and treatment display data, the following steps are included: The system calls upon a pre-defined three-dimensional human anatomy model library and, based on the meridian and acupoint information in the recommended treatment plan, performs spatial coordinate mapping on the three-dimensional human digital twin model to mark the target acupoint coordinates and meridian pathways. Based on the prescription composition and pharmacological data in the recommended treatment plan, a simulation animation of the diffusion of drug molecular structure in virtual human tissue is generated. The spatial coordinate mapping result and the diffusion simulation animation are synthesized by the rendering engine to generate a three-dimensional diagnosis and treatment view containing interactive hotspots; The system receives user interaction commands, adjusts the scaling, rotation angle, and display level of the 3D diagnostic view according to the commands, and outputs a real-time interactive diagnostic display screen.

[0016] Compared with existing technologies, the beneficial effects of this invention are as follows: First, by constructing a multi-dimensional relational treatment knowledge graph containing TCM literature, clinical data, and pharmacological data, it achieves deep structural integration of massive unstructured data in the field of TCM, providing high-dimensional underlying data support for subsequent accurate diagnosis. Second, by employing the isolated forest algorithm for disease type clustering, it can automatically identify and extract abnormal correlations in feature distribution, thereby achieving accurate classification of rare diseases or complex syndromes and improving the logical completeness of the knowledge graph in handling difficult and complicated cases. Furthermore, through a recursive decomposition and reconstruction mechanism of feature dimensions, it solves the problem of diagnostic gaps when the feature information of the user to be treated does not completely match the preset knowledge base, outputting personalized recommended treatment plans based on atomic feature synthesis, enhancing the system's diagnostic flexibility and coverage. Finally, by introducing a three-dimensional human digital twin model and image generation technology, it transforms abstract meridians, acupoints, and pharmacological effects into a three-dimensional view with clear spatial coordinates and dynamic interactivity, providing users with an intuitive treatment guidance interface, reducing reliance on the operator's TCM professional background, and improving the visualization of the treatment process.

[0017] On the other hand, this application also provides a three-dimensional interactive diagnosis and treatment display system for a knowledge graph of traditional Chinese medicine treatment, including: The model building module is configured to acquire TCM information data, and perform data cleaning, association analysis and clustering on the TCM information data to construct a multi-level relational treatment knowledge graph; The acquisition module is electrically connected to the model establishment module. The acquisition module is configured to acquire multimodal feature information of the user to be treated, match and calculate the multimodal feature information with preset feature information in the relational treatment knowledge graph, and output recommended treatment plan data. If the matching degree between the multimodal feature information and any preset feature information exceeds a preset threshold, the acquisition module outputs the treatment plan associated with the preset feature information; if it does not exceed the preset threshold, the acquisition module decomposes the multimodal feature information into multiple atomic features through a recursive decomposition algorithm, and synthesizes a recommended treatment plan based on the topological relationship of the atomic features in the relational treatment knowledge graph. The 3D display module is signal-connected to the acquisition module. The 3D display module is configured to receive the recommended treatment plan data, call the 3D rendering engine and digital twin model, generate 3D dynamic interactive images of meridians, acupoints and pharmacological pathways, and transmit them to a preset display terminal for display.

[0018] It is understood that the three-dimensional interactive diagnosis and treatment display method and system for a traditional Chinese medicine treatment knowledge graph in the above embodiments of the present invention have the same beneficial effects, and will not be described in detail here. Attached Figure Description

[0019] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of a three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment provided in an embodiment of the present invention; Figure 3 This is a structural block diagram of a three-dimensional interactive diagnostic and treatment display system for a knowledge graph of traditional Chinese medicine treatment, provided in an embodiment of the present invention. Detailed Implementation

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] like Figure 1-2 As shown in some embodiments of this application, this embodiment provides a three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment, including: Step S100: Obtain TCM information data, and perform structured processing and feature extraction on the TCM information data to construct a multi-dimensional relational treatment knowledge graph.

[0022] Specifically, acquiring traditional Chinese medicine (TCM) information data includes: obtaining TCM literature information data through web crawler interfaces, obtaining clinical medical data from medical institution databases, and obtaining TCM pharmacological data from pharmacopoeia databases. The TCM literature information data, the clinical medical data, and the TCM pharmacological data are then aggregated to establish an original TCM information dataset. The clinical medical data includes patient medical records, clinical observation data, and laboratory test indicators. The TCM pharmacological data includes drug components, molecular structures, metabolic pathway data, and efficacy evaluation indicators.

[0023] Understandably, by collecting data from multiple channels—namely, integrating TCM literature, clinical records, and pharmacological parameters—a foundational data source covering the entire lifecycle of TCM is formed. Specifically, by integrating unstructured text, semi-structured medical records, and structured pharmacological parameters from different physical storage media, a raw dataset with a high-dimensional feature space is established, providing underlying data support for subsequent entity extraction, relation modeling, and the dynamic evolution of knowledge graphs.

[0024] Specifically, data processing of the aforementioned traditional Chinese medicine information data includes: establishing, using entity relation extraction algorithms, associations for disease-syndrome-treatment principle, syndrome-prescription-traditional Chinese medicine, syndrome-food-medicinal property, prescription-drug-medicinal property, and meridian-acupoint-treatment based on the aforementioned traditional Chinese medicine literature information data, clinical medical data, and traditional Chinese medicine pharmacological data. Spatial distance metrics between these associations are then calculated using the Euclidean distance formula. d(P,Q)=\sqrt{\sum_{i=1}^{n}(p_i-q_i)^2}; Where P and Q represent feature vectors of different correlations, and n represents the feature dimension. A distance matrix is ​​established based on the spatial distance metric, and the correlation degree between each correlation is calculated based on the distance matrix to obtain correlation degree weight values. Based on the correlation degree weight values, each correlation is logically weighted to establish a set of treatment correlations.

[0025] Specifically, when establishing various association formulas based on the TCM literature information data, the clinical medical data, and the TCM pharmacology data, the process includes: using natural language processing technology to identify and remove incomplete text data and duplicate descriptive data from the TCM literature information data; identifying and removing duplicate case records and invalid case data that do not conform to preset logic from the clinical medical data; identifying and removing overlapping attribute data and missing parameter data from the TCM pharmacology data; establishing disease-syndrome-treatment principle association formulas and meridian-acupoint-treatment association formulas based on the mapping relationship between the TCM literature information data and the clinical medical data after data removal; establishing syndrome-prescription-TCM association formulas and syndrome-food-drug property association formulas based on the cross-reference relationship between the TCM literature information data, the clinical medical data, and the TCM pharmacology data after data removal; and establishing prescription-drug-drug property association formulas based on the pharmacodynamic feedback relationship between the TCM pharmacology data and the clinical medical data after data removal.

[0026] Understandably, by performing cleaning and denoising operations on multi-source heterogeneous data, redundant information and logical errors in the original data are eliminated, ensuring the purity of the dataset. Based on this, a multi-dimensional topological structure covering "disease, syndrome, prescription, medicine, meridian, and acupoint" is established through entity extraction and attribute association. Euclidean distance is used to quantify various diagnostic and treatment logics within the feature space, constructing a distance matrix reflecting the inherent laws of traditional Chinese medicine. By calculating the matrix weights, the correlation strength between different diagnostic and treatment elements is determined, ultimately forming a logically rigorous set of treatment associations, providing a quantitative basis for subsequent intelligent reasoning.

[0027] Specifically, when establishing a relational therapy knowledge graph based on each treatment association, the process includes: performing outlier detection on each treatment association using the isolated forest algorithm, and clustering each treatment association by disease type based on the detection results; extracting disease feature parameters from each treatment association within each disease type, and establishing a linear regression model between each treatment association within that disease type; and constructing a multi-level relational therapy knowledge graph based on each disease type, each treatment association under each disease type, and the weight parameters determined by the linear regression model.

[0028] Specifically, when clustering treatment associations for different disease types based on the isolated forest algorithm, the process includes: constructing an isolated forest model composed of multiple isolated trees, and calculating the path length of each treatment association within the isolated forest model. s(a)=2^{-\frac{E(h(a))}{c(n)}}; Where s(a) represents the outlier score of data a, E(h(a)) represents the average path length of data a in all isolated trees in the isolated forest, and c(n) represents the normalized path length. When the outlier score of the treatment association is greater than a preset outlier threshold, the treatment association is determined as an independent classification node, and the feature vector of the treatment association is extracted to establish a new disease type label. The disease information, physical characteristics, tongue appearance, and pulse appearance of the historical patients corresponding to the disease type label are obtained, and the central feature vector of the disease type is generated by the feature fusion algorithm. The cosine similarity between the pathological feature vector of each treatment association to be classified and the central feature vector of each disease type is calculated, and each treatment association is classified into the disease type with the highest similarity based on the K-means clustering algorithm.

[0029] Understandably, by introducing the Isolation Forest model, the automatic identification of marginal features in the set of treatment associations is achieved, classifying significantly different treatment patterns as independent disease types. Subsequently, through feature extraction and fusion of historical multimodal data, centralized representations of various diseases are constructed. Utilizing the synergistic effect of cosine similarity measurement and the K-means algorithm, massive amounts of treatment associations are categorized into corresponding disease categories. Finally, through parameter weights determined by a linear regression model, these classification results are transformed into a relational treatment knowledge graph with hierarchical structure and logical weights.

[0030] Step S200: Obtain the multimodal feature information of the user to be treated, and perform feature vector matching between the multimodal feature information and each preset feature information stored in the relational treatment knowledge graph, and determine the recommended treatment plan based on the matching calculation results.

[0031] Specifically, the process of acquiring the characteristic information of the user to be treated involves: acquiring facial and tongue images of the user through an image acquisition device; acquiring pulse dynamics data through a pressure sensor array; and acquiring the user's written medical history through a speech recognition device. Keyword extraction is then performed on the written medical history to obtain the user's physical characteristics.

[0032] Step S210: If the Euclidean distance between the multimodal feature information of the user to be treated and any feature vector among the preset feature information is less than a first preset threshold, then the treatment method mapped by the preset feature information in the relational treatment knowledge graph is determined to be a recommended treatment plan.

[0033] Step S220: If the Euclidean distance between the multimodal feature information of the user to be treated and each preset feature information is greater than or equal to the first preset threshold, then the multimodal feature information is decomposed according to the cosine similarity between it and each preset feature information to obtain preset feature information that matches each feature decomposition component, and a recommended treatment plan is reconstructed based on the topological relationship between the treatment methods mapped by each preset feature information in the relational treatment knowledge graph.

[0034] Specifically, when decomposing the multimodal feature information according to the cosine similarity with each preset feature information, the following steps are included: Calculate the cosine similarity S between each preset feature information vector and the feature information vector of the user to be treated: S=\frac{\mathbf{A}\cdot\mathbf{B}}{\|\mathbf{A}\|\|\mathbf{B}\|}; Where A is a preset feature information vector, and B is a feature information vector of the user to be treated. Multiple candidate preset feature information vectors with a similarity S greater than a preset similarity threshold are selected. The multimodal feature information is then decomposed into a first-level dimensionality based on these candidate preset feature information vectors, resulting in a first decomposed feature set. Each element in the first decomposed feature set is compared with each preset feature information vector, and a second-level decomposition is determined based on the comparison results.

[0035] Specifically, when comparing the decomposed feature information with each preset feature information, and determining whether to further decompose the decomposed feature information based on the comparison results, the process includes: if any element in the first decomposed feature set completely matches a feature vector in each preset feature information, then the decomposition operation on that element is stopped. If any element in the first decomposed feature set does not match any of the preset feature information, then a second-level dimensional decomposition is performed on that element until the decomposed atomic feature information matches any preset feature information in the relational therapy knowledge graph.

[0036] Understandably, the system constructs a complete user feature vector space by collecting multi-source modal data such as user images, stress levels, and voice recordings. When this vector has a direct mapping relationship with a preset node in the knowledge graph, the corresponding treatment plan is output. When no direct match exists, the system initiates a recursive decomposition mechanism, using cosine similarity as a guide to progressively decompose the complex multimodal features into smaller-grained atomic features. By recombinating these atomic features within the knowledge graph's topology, the logical reconstruction of non-standard cases is achieved. This atomic feature-based synthesis mechanism ensures that even when faced with complex or rare symptoms, the system can still generate scientifically based recommendations based on the underlying logic of the knowledge graph.

[0037] As can be seen, the aforementioned multimodal feature matching and recursive decomposition mechanism achieves deep adaptation to individual differences. When user features exhibit high uncertainty, the system finds locally optimal matches through dimensional decomposition and uses topological relationships in the knowledge graph for logical completion. This approach not only improves the error tolerance of the diagnostic process but also generates composite treatment plans including acupuncture, medication, and dietary therapy through the recombination of underlying atomic features. Furthermore, the cosine similarity-based screening mechanism ensures the rigor of the reconstructed plan in medical logic, providing accurate data input for subsequent 3D visualization.

[0038] The recommended treatment plan is obtained, and based on the meridian and acupoint data, pharmacokinetics data, and preset three-dimensional human digital twin model in the recommended treatment plan, three-dimensional interactive treatment display data is generated, and the three-dimensional interactive treatment display data is converted into a preset three-dimensional image output format for dynamic rendering output.

[0039] Understandably, the system calls upon a pre-defined 3D human anatomical model library and, based on the spatial coordinates of meridians and acupoints in the recommended treatment plan, performs point-to-path mapping and annotation on the digital twin model. Simultaneously, it utilizes pharmacokinetic parameters to generate a simulated fluid animation of drug diffusion within a virtual tissue environment. The rendering engine then synthesizes the static spatial coordinates with the dynamic simulation animation to generate a 3D diagnostic view with interactive hotspots. Users can send interactive commands via external input devices to perform zooming, rotation, or layered display operations on the view. This 3D interactive display method transforms complex traditional Chinese medicine theories into intuitive visual feedback, achieving visualized guidance of the diagnostic and treatment process and improving the accuracy of plan execution.

[0040] In the above embodiments, by constructing a multi-dimensional relational treatment knowledge graph containing TCM literature, clinical data, and pharmacological data, a deep structural integration of massive unstructured data in the field of TCM is achieved, providing high-dimensional underlying data support for subsequent accurate diagnosis. Secondly, the use of the isolated forest algorithm for symptom type clustering can automatically identify and extract abnormal correlations in feature distribution, thereby achieving accurate classification of rare diseases or complex syndromes and improving the logical completeness of the knowledge graph in handling difficult and complex cases. Furthermore, through a recursive decomposition and reconstruction mechanism of feature dimensions, the problem of diagnostic gaps when the user's feature information does not completely match the preset knowledge base is solved, outputting personalized recommended treatment plans based on atomic feature synthesis, enhancing the system's diagnostic flexibility and coverage. Finally, the introduction of a three-dimensional human digital twin model and image generation technology transforms abstract meridians, acupoints, and pharmacological effects into a three-dimensional view with clear spatial coordinates and dynamic interactivity, providing users with an intuitive treatment guidance interface, reducing reliance on the operator's TCM professional background, and improving the visualization of the treatment process.

[0041] In another preferred embodiment based on the above embodiments, such as Figure 3As shown, this embodiment provides a three-dimensional interactive diagnosis and treatment display system based on a traditional Chinese medicine (TCM) treatment knowledge graph. The system includes: a model building module configured to acquire TCM information data, perform data cleaning, association analysis, and clustering on the TCM information data, and construct a multi-level relational treatment knowledge graph. An acquisition module electrically connected to the model building module is configured to acquire multimodal feature information of the user to be treated, match the multimodal feature information with preset feature information in the relational treatment knowledge graph, and output recommended treatment plan data. A three-dimensional display module signal-connected to the acquisition module is configured to receive the recommended treatment plan data, call a three-dimensional rendering engine and a digital twin model, generate three-dimensional dynamic interactive images of meridians, acupoints, and pharmacological pathways, and transmit them to a preset display terminal for display.

[0042] Wherein: if the matching degree between the multimodal feature information and any preset feature information exceeds a preset threshold, the acquisition module outputs the treatment plan associated with the preset feature information; if it does not exceed the preset threshold, the acquisition module decomposes the multimodal feature information into multiple atomic features through a recursive decomposition algorithm, and synthesizes a recommended treatment plan based on the topological relationship of the atomic features in the relational treatment knowledge graph.

[0043] To enable those skilled in the art to fully understand and implement this invention, the specific implementation principles of this invention are further supplemented below with a specific application scenario.

[0044] In a real-world clinical setting, a patient seeking treatment at a TCM intelligent diagnostic terminal experienced symptoms such as chronic insomnia, palpitations, and dry mouth. The system first acquires facial and tongue images using an image acquisition device, showing a red tongue with little coating. Simultaneously, a pressure sensor array collects pulse dynamics data to obtain characteristic parameters of a thin, rapid pulse. The voice recognition device simultaneously records the patient's complaints of "difficulty falling asleep, frequent awakenings at night, and irritability," and after keyword extraction, generates the constitution characteristic information as "Yin deficiency with excessive fire."

[0045] The aforementioned multimodal feature information is integrated into a feature vector B, which is then input into the constructed relational therapy knowledge graph for matching. The system first calculates the relationship between B and each preset feature vector A1, A2, ..., A in the knowledge graph. Euclidean distance between them: ; Calculations showed that all distance values ​​were greater than a first preset threshold (e.g., 0.35), indicating that the patient's characteristics did not completely match any existing standard syndrome nodes. At this point, the system initiated the feature dimension decomposition process, calculating the distances between B and each A. Cosine similarity: ; Three candidate feature vectors with a similarity greater than a preset similarity threshold (e.g., 0.85) are selected: A1 (corresponding to "Heart Yin Deficiency Syndrome"), A3 (corresponding to "Liver and Kidney Yin Deficiency Syndrome"), and A7 (corresponding to "Deficiency Fire Rising Syndrome"). Based on these three candidate vectors, the system performs a first-level dimensional decomposition on the original feature vector B, separating sub-feature sets such as dry mouth and tongue, red tongue with little coating, and thin and rapid pulse.

[0046] Subsequently, the system compared the sub-feature "red tongue with little coating" with the atomic feature library in the knowledge graph and found that it completely matched the tongue atomic features under "heart yin deficiency syndrome"; "thready and rapid pulse" completely matched the pulse atomic features under "liver and kidney yin deficiency syndrome"; and "irritability" matched the emotional feature atomic features under "deficiency fire rising syndrome". Since the corresponding atomic feature nodes have been found for each component, the system stopped further decomposition.

[0047] Based on the topological relationships stored in the knowledge graph, the system retrieved the following: "Heart Yin Deficiency Syndrome" is associated with the formula Tianwang Buxin Dan; "Liver and Kidney Yin Deficiency Syndrome" is associated with the formula Liuwei Dihuang Wan; "Deficiency Fire Rising Syndrome" is associated with the dietary therapy plan Baihe Lianzi Porridge; and the associated meridian therapy plan is acupuncture at Shenmen (HT7), Sanyinjiao (SP6), and Taixi (KI3). Based on the weight coefficients of each atomic feature in the original vector, the system calls the weight parameters determined by the linear regression model to perform a weighted fusion of the above three treatment paths, generating a composite recommended treatment plan: using Tianwang Buxin Dan as the base formula, adding Rehmannia glutinosa and Cornus officinalis to nourish the liver and kidneys, combined with daily consumption of Baihe Lianzi Porridge, and recommending acupuncture three times a week at Shenmen, Sanyinjiao, and Taixi acupoints.

[0048] The recommended treatment plan is transmitted to the 3D display module. The system calls a preset 3D human digital twin model and, based on the spatial coordinate data in the meridian-acupoint-treatment correlation, precisely marks three acupoints on the model: Shenmen (located at the ulnar end of the wrist crease, in the depression on the radial side of the flexor carpi ulnaris tendon), Sanyinjiao (3 cun above the tip of the medial malleolus, posterior to the medial border of the tibia), and Taixi (posterior to the medial malleolus, in the depression between the tip of the medial malleolus and the Achilles tendon). The system also highlights the pathways of the Hand Shaoyin Heart Meridian, Foot Taiyin Spleen Meridian, and Foot Shaoyin Kidney Meridian. Simultaneously, based on the pharmacokinetic parameters of ingredients such as Danshen, Danggui, and Suanzaoren in Tianwang Buxin Dan, a fluid simulation animation of drug molecule diffusion is generated in the virtual heart and brain regions, simulating the dynamic process of the drug acting on the mind and blood.

[0049] Finally, the system uses a rendering engine to synthesize static acupoint annotations, dynamic meridian paths, and drug diffusion animations, outputting an interactive 3D diagnostic view. Patients or physicians can rotate the human body model via the touchscreen, click on any acupoint to view its location, indications, and acupuncture depth description, or switch to "pharmacology mode" to observe the distribution sequence of drugs in the body. After receiving the user's "zoom in on the head area" command, the system adjusts the camera perspective in real time, focusing on the tongue and facial feature areas, and simultaneously displays the corresponding TCM diagnostic text labels for that area.

[0050] Throughout the process, the Isolation Forest algorithm identified the "Yin Deficiency with Internal Heat" marginal syndrome pattern during the knowledge graph construction phase and incorporated it as an independent symptom type node into the clustering system, with an anomaly score s(a)=2. -E(h(a)) / c(n) The calculated result was 0.87, exceeding the preset abnormality threshold of 0.8. Therefore, the system created a dedicated central feature vector for it. The cosine similarity between the patient's feature vector B and this central vector reached 0.91. Based on this, the K-means clustering algorithm classified it into this class, ensuring effective coverage of complex syndromes.

[0051] It is understood that the three-dimensional interactive diagnosis and treatment display system and method of traditional Chinese medicine treatment knowledge graph in the above embodiments of the present invention have the same beneficial effects, and will not be described again.

[0052] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0053] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0054] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0055] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment, characterized in that, include: Acquire traditional Chinese medicine information data, perform structured processing and feature extraction on the traditional Chinese medicine information data, and construct a multi-dimensional relational treatment knowledge graph; The system acquires multimodal feature information of the user to be treated, performs feature vector matching between the multimodal feature information and each preset feature information stored in the relational treatment knowledge graph, and determines the recommended treatment plan based on the matching calculation results. If the Euclidean distance between the multimodal feature information of the user to be treated and any feature vector among the preset feature information is less than a first preset threshold, then the treatment method mapped by the preset feature information in the relational treatment knowledge graph is determined to be a recommended treatment plan. If the Euclidean distance between the multimodal feature information of the user to be treated and each preset feature information is greater than or equal to the first preset threshold, then the multimodal feature information is decomposed according to the cosine similarity between it and each preset feature information to obtain preset feature information that matches each feature decomposition component, and a recommended treatment plan is reconstructed based on the topological relationship between the treatment methods mapped by each preset feature information in the relational treatment knowledge graph. The recommended treatment plan is obtained, and based on the meridian and acupoint data, pharmacokinetics data, and preset three-dimensional human digital twin model in the recommended treatment plan, three-dimensional interactive treatment display data is generated, and the three-dimensional interactive treatment display data is converted into a preset three-dimensional image output format for dynamic rendering output.

2. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 1, characterized in that, When acquiring information and data related to traditional Chinese medicine, the following should be included: Traditional Chinese medicine (TCM) literature information data is obtained through web crawler interfaces, clinical medical data is obtained from medical institution databases, and pharmacological data of Chinese medicinal herbs is obtained from pharmacopoeia databases. The TCM literature information data, clinical medical data, and pharmacological data of Chinese medicinal herbs are then aggregated to establish an original TCM information dataset.

3. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 2, characterized in that, Data processing of the aforementioned traditional Chinese medicine information data includes: Based on the TCM literature information data, the clinical medical data, and the TCM pharmacology data, the following associations are established using entity relation extraction algorithms: disease-syndrome-treatment principle association, syndrome-prescription-TCM association, syndrome-food-medicine property association, prescription-drug-medicine property association, and meridian-acupoint-treatment association. The spatial distance metric between various correlations is obtained based on the Euclidean distance formula. A distance matrix is ​​established based on the spatial distance metric, and the correlation degree between each correlation formula is calculated based on the distance matrix to obtain the correlation degree weight value. Based on the aforementioned correlation weight values, each correlation is logically weighted to establish a set of treatment correlations.

4. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 3, characterized in that, When establishing various correlation equations based on the aforementioned TCM literature information data, clinical medical data, and TCM pharmacological data, the following are included: Based on natural language processing technology, incomplete text data and duplicate descriptive data are identified and removed from the TCM literature information data. Identify and remove duplicate case records and invalid case data that do not conform to preset logic from the clinical medical data; Identify and remove overlapping attribute data and missing parameter data from the pharmacological data of traditional Chinese medicine; Based on the mapping relationship between the TCM literature information data and the clinical medical data after data removal, establish the disease-syndrome-treatment principle association and the meridian-acupoint-treatment association; Based on the cross-reference relationships between the TCM literature information data, the clinical medical data, and the TCM pharmacology data after data removal, establish the syndrome type-prescription-TCM association and the syndrome type-food-drug property association; Based on the pharmacological data of the Chinese herbal medicines and the clinical medical data after data removal, a formula-drug-property correlation is established.

5. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 4, characterized in that, When constructing a relational therapy knowledge graph based on various treatment associations, the following should be included: Outlier detection is performed on each treatment association based on the isolated forest algorithm, and disease type clustering is performed on each treatment association based on the detection results; Extract the disease characteristic parameters of each treatment correlation in each disease type, and establish a linear regression model between each treatment correlation in the disease type; Based on the disease types, the treatment associations under each disease type, and the weight parameters determined by the linear regression model, a multi-level relational treatment knowledge graph is constructed.

6. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 5, characterized in that, When clustering various disease types based on treatment associations using the isolated forest algorithm, the following applies: An isolated forest model consisting of multiple isolated trees is constructed. The path length of each treatment correlation in the isolated forest model is calculated, and the anomaly score is calculated based on the path length, where: When the abnormal score of the treatment association is greater than the preset abnormal threshold, the treatment association is determined to be an independent classification node, and the feature vector of the treatment association is extracted to establish a new disease type label. Obtain the disease information, physical characteristics information, tongue appearance information, and pulse appearance information of historical patients corresponding to the disease type label, and generate the central feature vector of the disease type through feature fusion algorithm; Obtain the cosine similarity between the pathological feature vectors of each treatment association to be classified and the central feature vectors of each disease type, and classify each treatment association into the disease type with the highest similarity based on the K-means clustering algorithm.

7. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 6, characterized in that, When obtaining the characteristic information of the user to be treated, the specific steps are as follows: The data acquisition module acquires facial images, tongue images, and pulse dynamics data. Acquire the textual medical history of the user to be treated based on a speech recognition device; Keyword extraction is performed on the described medical history text to obtain the physical characteristics information of the user to be treated.

8. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 7, characterized in that, When decomposing feature information according to its similarity to each preset feature information, it includes: Obtain the cosine similarity S between each preset feature information vector and the feature information vector of the user to be treated: Multiple candidate preset feature information with similarity S greater than a preset similarity threshold are selected, and the multimodal feature information is decomposed into a first-level dimension based on the candidate preset feature information to obtain a first decomposed feature set; Each element in the first set of disassembly features is compared with each preset feature information, and a decision is made based on the comparison results as to whether to perform the second-level disassembly, wherein: If any element in the first set of disassembled features completely matches a feature vector in each of the preset feature information, then the disassembly operation of that element is stopped. If any element in the first set of decomposed features does not match any of the preset feature information, then the element is decomposed into a second-level dimension until the decomposed atomic feature information matches any of the preset feature information in the relational therapy knowledge graph.

9. The three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in claim 8, characterized in that, When generating and outputting 3D interactive diagnostic and treatment display data, the following are included: The system calls upon a pre-defined three-dimensional human anatomy model library and, based on the meridian and acupoint information in the recommended treatment plan, performs spatial coordinate mapping on the three-dimensional human digital twin model to mark the target acupoint coordinates and meridian pathways. Based on the prescription composition and pharmacological data in the recommended treatment plan, a simulation animation of the diffusion of drug molecular structure in virtual human tissue is generated. The spatial coordinate mapping result and the diffusion simulation animation are synthesized by the rendering engine to generate a three-dimensional diagnosis and treatment view containing interactive hotspots; The system receives user interaction commands, adjusts the scaling, rotation angle, and display level of the 3D diagnostic view according to the commands, and outputs a real-time interactive diagnostic display screen.

10. A three-dimensional interactive diagnostic and treatment display system for a knowledge graph of traditional Chinese medicine treatment, employing the three-dimensional interactive diagnostic and treatment display method for a knowledge graph of traditional Chinese medicine treatment as described in any one of claims 1-9, characterized in that, include: The model building module is configured to acquire TCM information data, and perform data cleaning, association analysis and clustering on the TCM information data to construct a multi-level relational treatment knowledge graph; The acquisition module is electrically connected to the model establishment module. The acquisition module is configured to acquire multimodal feature information of the user to be treated, match and calculate the multimodal feature information with preset feature information in the relational treatment knowledge graph, and output recommended treatment plan data. If the matching degree between the multimodal feature information and any preset feature information exceeds a preset threshold, the acquisition module outputs the treatment plan associated with the preset feature information; if it does not exceed the preset threshold, the acquisition module decomposes the multimodal feature information into multiple atomic features through a recursive decomposition algorithm, and synthesizes a recommended treatment plan based on the topological relationship of the atomic features in the relational treatment knowledge graph. The 3D display module is signal-connected to the acquisition module. The 3D display module is configured to receive the recommended treatment plan data, call the 3D rendering engine and digital twin model, generate 3D dynamic interactive images of meridians, acupoints and pharmacological pathways, and transmit them to a preset display terminal for display.

Images