AI syndrome differentiation and treatment model system suitable for multi-dimensional difference environment

Abstract

The invention discloses an AI syndrome differentiation and treatment model system suitable for a multi-dimensional difference environment, which comprises a data integration layer and a three-cause characteristic processing and dynamic dialectical model, and is characterized in that the data integration layer comprises outpatient service data, hospitalization data, follow-up visit data, physical characteristics, pathogenesis solar terms / five-transportation six-qi and regional medicinal material relationships; the three-factor feature processing comprises human-factor feature processing, time-factor feature processing and place-factor feature processing; the dynamic dialectical model comprises multi-modal data fusion, three-factor weight dynamic allocation and curative effect correction syndrome reasoning. The syndrome differentiation accuracy of chronic epigastric pain patients reaches 89.2%, the clinical curative effect and acceptability of a treatment scheme are improved, the average iteration period of the model is shortened to 72 h through a feedback closed-loop system, the method deeply conforms to Chinese geographical population characteristics, and the method has the advantages of being high in adaptability and good in application prospect through an environment perception-physique evaluation-dynamic syndrome differentiation three-level linkage mechanism. Precise diagnosis and treatment of one person with one prescription at once is achieved, and core technical support is provided for modernization of traditional Chinese medicine.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence in traditional Chinese medicine, and in particular to an AI-based diagnostic and treatment model system suitable for multidimensional and diverse environments. Background Technology

[0002] Analysis of the current pain points in my country's national conditions:

[0003] The geographical environment is vast: the climate difference between the north and south (annual humidity difference of 40%) and the significant altitude gradient (0-8000 meters) lead to a prominent need for different treatments for the same disease.

[0004] The population has complex characteristics: there are differences in physical constitution among the 56 ethnic groups and differences in urban and rural living habits (for example, the proportion of people with high salt diets in the north reaches 62%).

[0005] Limitations of traditional AI models: Existing systems (such as CN118779293A and CN118141677A) only generate suggestions based on syndrome differentiation and prescription mapping relationships, without considering the corrective effects of patient constitution and environmental spatiotemporal variables on syndrome differentiation, and without establishing a regional efficacy database of medicinal materials, resulting in situations where, for example, the use of medicines is similar in the humid and hot eastern region and the dry and cold western region, but the efficacy is heterogeneous.

[0006] Therefore, providing a method for constructing an intelligent diagnostic model based on the integration of the TCM theory of "three factors" (person, time, and place) and building a TCM intelligent diagnostic and treatment system that integrates dynamic spatiotemporal parameters for TCM auxiliary diagnosis, primary healthcare, remote consultation, and cross-regional health management is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0007] In view of this, the present invention provides an AI dialectical treatment model system applicable to multidimensional and differentiated environments.

[0008] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0009] An AI-based dialectical treatment model system suitable for multidimensional and differentiated environments includes: a data integration layer, three-factor feature processing, and a dynamic dialectical model.

[0010] The data integration layer includes: outpatient data, inpatient data, follow-up data, physical characteristics, disease onset during the solar term / Five Elements and Six Qi, and the relationship between regional medicinal materials;

[0011] The three-factor feature processing includes: person-specific feature processing, time-specific feature processing, and location-specific feature processing;

[0012] The dynamic dialectical model includes: multimodal data fusion, dynamic allocation of three-factor weights, and efficacy-modified syndrome reasoning.

[0013] Preferably, the outpatient data includes: patient basic information, environmental information, four diagnostic methods information, onset time characteristics, symptom and sign information, prescription information, and efficacy feedback information;

[0014] The inpatient data includes: diagnostic information, examination indicators, environmental information, four diagnostic methods information, characteristics of onset time, symptoms and signs, prescription information, efficacy feedback, and discharge improvement.

[0015] The follow-up data includes: symptom improvement, physiological indicator trends, frequency of recurrence of key symptoms, dynamic living environment data, medication intervention, short-term follow-up, and long-term follow-up.

[0016] The physical characteristics include those collected according to nine physical constitution classification standards;

[0017] The aforementioned disease-causing solar terms / Five Elements and Six Qi information includes: the disease-causing solar terms and Five Elements and Six Qi information provide time-dimensional decision support for the model; wherein...

[0018] Solar Terms: Disease Warning During Periods of Sudden Climate Change;

[0019] Five Elements: Predicting the weak points of the internal organs throughout the year;

[0020] The Six Qi: Guiding the prevention and control of stage-specific diseases.

[0021] Preferably, the individualized feature processing includes: processing all information related to the patient's physical condition and disease onset;

[0022] The time-dependent feature processing includes: processing information on solar terms and the Five Elements and Six Qi related to the disease;

[0023] The location-specific processing includes processing information related to the patient's living environment and the characteristics of medication use in the place of residence.

[0024] Preferably, the dynamic dialectical model uses a BiLSTM-Transformer hybrid network to process temporal features and combines knowledge graph reasoning to generate the evidence type;

[0025] The multimodal data fusion includes: multimodal data fusion refers to integrating data from different sources and of different types, extracting complementary information through algorithms, and generating a comprehensive judgment that transcends a single data dimension;

[0026] This mainly includes textual semantic modalities, i.e., patient complaints; visual sensory modalities, i.e., tongue and pulse; temporal signal modalities, i.e., heart rate and blood pressure; environmental sensory modalities, i.e., temperature and humidity; and biochemical indicator modalities, i.e., examination indicators.

[0027] The dynamic allocation of the three factors includes: judging the efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up visits. In essence, the efficacy-driven allocation of the three factors is a dynamic optimization process.

[0028] The efficacy-modified syndrome reasoning includes: judging efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up visits. In essence, the efficacy-driven weight allocation of the three factors is a dynamic optimization process.

[0029] Preferred, individualized module: Based on the physical constitution identification model, dynamically adjust the diagnostic weights and develop an adaptive classification algorithm for physical constitution characteristics of multiple ethnic groups;

[0030] The timing module introduces the seasonal changes of disease onset, and uses the Five Elements and Six Qi algorithm to generate time-series corrected diagnostic weights;

[0031] Location-specific module: Establish a region-medicinal herb meridian association map to construct a mapping model of China's geography-climate-medicinal herb efficacy.

[0032] Preferably, the constitution-based identification model includes nine constitution classifiers;

[0033] The construction of the nine constitution classifiers requires multi-dimensional data fusion, algorithm adaptation, and embedding of traditional Chinese medicine theory, ultimately realizing the transformation from "experience-based judgment" to "data-driven".

[0034] The typical technical path is as follows:

[0035] Data collection: Standardized constitution questionnaire + digital tongue and pulse diagnosis;

[0036] Feature extraction: Extracting semantic features from traditional Chinese medicine;

[0037] Model training: combining machine learning with domain knowledge constraints;

[0038] Dynamic optimization: Iterative updates based on clinical feedback.

[0039] Preferably, the development of the multi-ethnic physical characteristics adaptive classification algorithm is to construct a classification model that can identify the unique physical characteristics of the Miao and Tibetan ethnic minorities. The model is trained based on case data from minority-populated areas, and by integrating traditional Chinese medicine theory and ethnic medical knowledge, it can achieve precise and personalized cross-ethnic physical identification.

[0040] The present invention achieves the following technical effects compared to the prior art:

[0041] (1) Clinical tests of this invention show that the accuracy rate of diagnosis for patients with chronic stomach pain is 89.2% (65.5% for the traditional model);

[0042] (2) By dynamically adjusting the dosage of medication (e.g., the dosage of ephedra for patients in the north is automatically reduced by 15%-20%), the clinical efficacy and acceptance of the treatment plan have been improved.

[0043] (3) The feedback closed-loop system reduces the average iteration cycle of the model to 72 hours (the traditional method requires more than 3 months);

[0044] (4) This plan is deeply aligned with China’s geographical and demographic characteristics. Through a three-level linkage mechanism of environmental perception, physical fitness assessment and dynamic diagnosis, it achieves precise diagnosis and treatment of “one person, one time, one prescription”, providing core technical support for the modernization of traditional Chinese medicine. Attached Figure Description

[0045] Figure 1 This is a system structure diagram of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, this invention discloses an AI-based dialectical treatment model system suitable for multidimensional and differentiated environments, comprising: a data integration layer, three-factor feature processing, and a dynamic dialectical model.

[0048] The data integration layer includes: outpatient data, inpatient data, follow-up data, physical characteristics, seasonal factors of onset / Five Elements and Six Qi, and regional relationships with medicinal materials;

[0049] Three-factor feature processing includes: person-based feature processing, time-based feature processing, and location-based feature processing;

[0050] The dynamic dialectical model includes: multimodal data fusion, dynamic allocation of three-factor weights, and syndrome reasoning based on efficacy correction.

[0051] Outpatient data includes: patient basic information, environmental information, four diagnostic methods information, characteristics of onset time, symptoms and signs, prescription information, and efficacy feedback information;

[0052] Inpatient data includes: diagnostic information, examination indicators, environmental information, four diagnostic methods information, characteristics of onset time, symptoms and signs, prescription information, efficacy feedback, and discharge improvement.

[0053] Follow-up data includes: symptom improvement, physiological indicator trends, frequency of recurrence of key symptoms, dynamic living environment data, medication intervention, short-term follow-up, and long-term follow-up.

[0054] Physical characteristics include: data collected according to the nine-type physical constitution classification standard;

[0055] The onset of disease during specific solar terms / Five Elements and Six Qi includes: information on the onset of disease during specific solar terms and Five Elements and Six Qi provides time-dimensional decision support for the model; among which...

[0056] Solar Terms: Disease Warning During Periods of Sudden Climate Change;

[0057] Five Elements: Predicting the weak points of the internal organs throughout the year;

[0058] The Six Qi: Guiding the prevention and control of stage-specific diseases.

[0059] Application Cases

[0060] Patient: Female, 48 years old, with a history of asthma, visited the clinic in March 2025 (Spring Equinox).

[0061] Spatiotemporal parameters:

[0062] Solar Term: Spring Equinox (Second Phase: Lesser Yin Fire (Main Qi) + Greater Yang Cold Water (Guest Qi))

[0063] Five Elements and Six Qi: Year of Yi Si (Metal element is insufficient, Jueyin Wind Wood governs the sky)

[0064] Location: Beijing (dry and windy)

[0065] Guiding dialectical decision-making:

[0066] Core syndrome pattern: Wind-evil invading the lungs, internal stirring of liver wind

[0067] Recommended formula: Saposhnikovia divaricata, Stellaria dichotoma, Buthus martensii, Scolopendra subspinipes

[0068] Dynamic adjustment:

[0069] Due to the timing: The dosage of Saposhnikovia divaricata is increased to 12g (based on 9g) at the Spring Equinox.

[0070] Due to the dry climate in Beijing, add 15g of Ophiopogon japonicus to moisten the lungs.

[0071] The regional relationships of medicinal materials include: the source of the medicine, batch number, place of origin of the medicinal material, harvesting season, processing method, and feedback on efficacy.

[0072] Personalized processing includes processing all information related to the patient's physical condition and the course of the disease;

[0073] Time-specific processing includes: processing information related to the solar terms and the Five Elements and Six Qi in relation to the disease;

[0074] Site-specific treatment includes processing information related to the patient's living environment and the characteristics of medication use in the place of residence.

[0075] The dynamic dialectical model uses a BiLSTM-Transformer hybrid network to process temporal features and combines knowledge graph reasoning to generate the evidence type.

[0076] Multimodal data fusion includes: Multimodal data fusion refers to integrating data from different sources and of different types, extracting complementary information through algorithms, and generating comprehensive judgments that transcend a single data dimension;

[0077] This mainly includes textual semantic modalities, i.e., patient complaints; visual sensory modalities, i.e., tongue and pulse; temporal signal modalities, i.e., heart rate and blood pressure; environmental sensory modalities, i.e., temperature and humidity; and biochemical indicator modalities, i.e., examination indicators.

[0078] The dynamic allocation of the three factors includes: judging the efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up. In essence, the efficacy-driven allocation of the three factors is a dynamic optimization process.

[0079] The efficacy-modified syndrome reasoning includes judging efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up. In essence, the efficacy-driven weight allocation of the three factors is a dynamic optimization process.

[0080] Individualized module: Based on the physical constitution identification model, dynamically adjust the diagnostic weights and develop an adaptive classification algorithm for physical constitution characteristics of multiple ethnic groups;

[0081] The timing module introduces the seasonal changes of disease onset, and uses the Five Elements and Six Qi algorithm to generate time-series corrected diagnostic weights;

[0082] Location-specific module: Establish a region-medicinal herb meridian association map to construct a mapping model of China's geography-climate-medicinal herb efficacy.

[0083] The constitution identification model includes nine constitution classifiers;

[0084] The construction of the nine constitution classifiers requires multi-dimensional data fusion, algorithm adaptation and embedding of traditional Chinese medicine theory, ultimately realizing the transformation from "experience judgment" to "data-driven".

[0085] The typical technical path is as follows:

[0086] Data collection: Standardized constitution questionnaire + digital tongue and pulse diagnosis;

[0087] Feature extraction: Extracting semantic features from traditional Chinese medicine;

[0088] Model training: combining machine learning with domain knowledge constraints;

[0089] Dynamic optimization: Iterative updates based on clinical feedback.

[0090] The purpose of developing an adaptive classification algorithm for multi-ethnic physical characteristics is to construct a classification model that can identify the unique physical characteristics of the Miao and Tibetan ethnic minorities. The model is trained based on case data from minority-populated areas, and by integrating traditional Chinese medicine theory and ethnic medical knowledge, it can achieve precise and personalized cross-ethnic physical identification.

[0091] The constitution-syndrome-prescription are shown in Table 1.

[0092] Table 1: Constitution-Syndrome Type-Prescription

[0093]

[0094]

[0095] Among them, the enhancement of the solar term-main-common syndrome type is shown in Table 2;

[0096] Table 2: Solar Terms - Main Themes - Common Syndromes Enhancement

[0097] Solar Terms Main Qi Enhancement of common syndrome types Big Cold-Vernal Equinox Jueyin Wind Wood Liver diseases weighted +30% Lesser Fullness (Xiaoman) - Greater Heat (Dashu) Lesser Yang Fire The threshold for damp-heat syndrome has been lowered by 15%.

[0098] Table 3 shows the drug dosage adjustments for the treatment plan to match the solar terms.

[0099] Table 3: Adjustment of drug dosage according to the solar term in the treatment plan

[0100]

[0101] Among them, Fritillaria cirrhosa (Sichuan Aba Prefecture vs. Tibet Linzhi) is shown in Table 4;

[0102] Table 4: Fritillaria cirrhosa (Aba Prefecture, Sichuan vs. Linzhi, Tibet)

[0103]

[0104] The efficacy of treatment is judged based on the improvement of patients' symptoms and the trend of test indicators during follow-up. Addressing the issue of inconsistent quality of medicinal materials from different institutions in certain regions where traditional Chinese medicines are used is essentially a dynamic optimization process driven by efficacy.

[0105] Among them, Chinese yam (from Jiaozuo, Henan vs. Heze, Shandong) is shown in Table 5;

[0106] Table 5: Chinese yam (Jiaozuo, Henan vs. Heze, Shandong)

[0107]

[0108] The efficacy of treatment is judged based on the improvement of patients' symptoms and the trend of test indicators during follow-up. Addressing the issue of inconsistent quality of medicinal materials from different institutions in certain regions where traditional Chinese medicines are used is essentially a dynamic optimization process driven by efficacy.

[0109] The data type-data type-TCM association features are shown in Table 6.

[0110] Table 6: Data Type-Data Type-TCM Correlation Features

[0111]

[0112]

[0113] After processing and analyzing multimodal data, a treatment plan is proposed based on the weight allocation of the three factors. Then, the patient's symptom improvement and the trend of the test indicators are followed up to judge the efficacy. In essence, the efficacy-driven weight allocation of the three factors is a dynamic optimization process.

[0114] Example 1: Personalized Treatment of Hypertension (North vs. South)

[0115] A patient from Shenyang (52 years old, phlegm-dampness constitution, winter):

[0116] Environmental parameters: temperature -15℃, PM2.5 85μg / m3;

[0117] System output: Increase the dosage of earthworm to 12g (normally 9g), and combine with perilla seeds to resolve smog toxins. Guangzhou patient (48 years old, Yin deficiency constitution, summer):

[0118] Environmental parameters: Temperature 32℃, Humidity 92%;

[0119] System output: Reduce the dosage of Aconitum carmichaelii to 3g (normally 6g), and add Buzhaye, a characteristic medicinal herb of Lingnan.

[0120] Example 2: Multi-ethnic physical fitness adaptation

[0121] Tibetan patient (high-altitude polycythemia):

[0122] Physical constitution label: High altitude stagnation type (unique subtype);

[0123] The system activates the Tibetan medicine module: Rhodiola rosea dosage is linked to blood oxygen saturation sensor data.

[0124] Example 3:

[0125] Gender: Male

[0126] Age: 52

[0127] Chief complaint: Recurrent dizziness and fatigue for 2 years, worsening in the past week.

[0128] Location: Chengdu, Sichuan Province (30.6°N, 500m altitude, 80% average annual humidity)

[0129] Consultation time: Great Heat solar term (July 23rd in the Gregorian calendar)

[0130] Physical characteristics: Phlegm-dampness constitution (questionnaire score 82 / 100, yellow and greasy tongue coating, slippery and rapid pulse); data input and three-factor feature extraction are shown in Table 6;

[0131] Table 6: Data Input and Three-Factor Feature Extraction

[0132]

[0133] The results of the dialectical analysis and the generated solutions are shown in Table 7.

[0134] Table 7: Diagnostic Results and Scheme Generation

[0135]

[0136]

[0137] Data showing a symptom improvement rate greater than 30% during expected treatment follow-up were included in the revised data to improve the model.

[0138] The evaluation results are shown in Table 8.

[0139] Table 8: Evaluation Indicator Results

[0140] Evaluation indicators Dynamic Dialectical Model Group Traditional dialectical group Increase Diagnostic accuracy 92％ 68％ +35.3％ Symptom improvement rate 89％ 62％ +43.5％ Blood pressure control rate 85％ 58％ +46.6％ Recurrence rate (6 months) 18％ 42％ -57.1％

[0141] Example 4:

[0142] Case: Obese patient with phlegm-dampness constitution (BMI 28.5)

[0143] Constitution assessment: Phlegm-dampness constitution, 85 points (threshold ≥ 60). Tongue appearance: Large tongue with white, greasy coating (CNN classification confidence 91%). Pulse appearance: Slippery pulse (waveform entropy value 2.1, normal < 1.8).

[0144] Recommended medication:

[0145] Core formula: Er Chen Tang (Pinellia ternata, Citrus reticulata peel, Poria cocos). Suggested modifications: Add 15g of hawthorn (to lower lipids) and 9g of Alisma plantago-aquatica (to promote diuresis). Avoid rich, fatty, and sweet foods. It is recommended to exercise for 30 minutes daily.

[0146] Treatment feedback: After 3 months, BMI dropped to 25.2, and the model automatically strengthened the "hawthorn-lipid-lowering" association weight by 5%.

[0147] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the technical scope of the present invention. Therefore, any minor modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention shall still fall within the scope of the technical solution of the present invention.

Claims

1. An AI-based dialectical treatment model system suitable for multidimensional and differentiated environments, characterized in that, include: Data integration layer, three-factor feature processing, and dynamic dialectical model. The data integration layer includes: outpatient data, inpatient data, follow-up data, physical characteristics, disease onset during the solar term / Five Elements and Six Qi, and the relationship between regional medicinal materials; The three-factor feature processing includes: person-specific feature processing, time-specific feature processing, and location-specific feature processing; The dynamic dialectical model includes: multimodal data fusion, dynamic allocation of three-factor weights, and efficacy-modified syndrome reasoning.

2. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 1, characterized in that, The outpatient data includes: patient basic information, environmental information, four diagnostic methods information, onset time characteristics, symptom and sign information, prescription information, and efficacy feedback information; The inpatient data includes: diagnostic information, examination indicators, environmental information, four diagnostic methods information, characteristics of onset time, symptoms and signs, prescription information, efficacy feedback, and discharge improvement. The follow-up data includes: symptom improvement, physiological indicator trends, frequency of recurrence of key symptoms, dynamic living environment data, medication intervention, short-term follow-up, and long-term follow-up. The physical characteristics include those collected according to nine physical constitution classification standards; The aforementioned disease-causing solar terms / Five Elements and Six Qi information includes: the disease-causing solar terms and Five Elements and Six Qi information provide time-dimensional decision support for the model; wherein... Solar Terms: Disease Warning During Periods of Sudden Climate Change; Five Elements: Predicting the weak points of the internal organs throughout the year; The Six Qi: Guiding the prevention and control of stage-specific diseases.

3. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 1, characterized in that, The individualized characteristic processing includes: processing all information related to the patient's physical condition and disease onset; The time-dependent feature processing includes: processing information on solar terms and the Five Elements and Six Qi related to the disease; The location-specific processing includes processing information related to the patient's living environment and the characteristics of medication use in the place of residence.

4. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 1, characterized in that, The dynamic dialectical model uses a BiLSTM-Transformer hybrid network to process temporal features and combines knowledge graph reasoning to generate the evidence type. The multimodal data fusion includes: multimodal data fusion refers to integrating data from different sources and of different types, extracting complementary information through algorithms, and generating a comprehensive judgment that transcends a single data dimension; This mainly includes textual semantic modalities, i.e., patient complaints; visual sensory modalities, i.e., tongue and pulse; temporal signal modalities, i.e., heart rate and blood pressure; environmental sensory modalities, i.e., temperature and humidity; and biochemical indicator modalities, i.e., examination indicators. The dynamic allocation of the three factors includes: judging the efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up visits. In essence, the efficacy-driven allocation of the three factors is a dynamic optimization process. The efficacy-modified syndrome reasoning includes: judging efficacy based on the improvement of patients' symptoms and the trend of examination indicators during follow-up visits. In essence, the efficacy-driven weight allocation of the three factors is a dynamic optimization process.

5. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 1, characterized in that, Individualized module: Based on the physical constitution identification model, dynamically adjust the diagnostic weights and develop an adaptive classification algorithm for physical constitution characteristics of multiple ethnic groups; The timing module introduces the seasonal changes of disease onset, and uses the Five Elements and Six Qi algorithm to generate time-series corrected diagnostic weights; Location-specific module: Establish a region-medicinal herb meridian association map to construct a mapping model of China's geography-climate-medicinal herb efficacy.

6. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 5, characterized in that, The constitution-based identification model includes nine constitution classifiers; The construction of the nine constitution classifiers requires multi-dimensional data fusion, algorithm adaptation, and embedding of traditional Chinese medicine theory, ultimately realizing the transformation from "experience-based judgment" to "data-driven". The typical technical path is as follows: Data collection: Standardized constitution questionnaire + digital tongue and pulse diagnosis; Feature extraction: Extracting semantic features from traditional Chinese medicine; Model training: combining machine learning with domain knowledge constraints; Dynamic optimization: Iterative updates based on clinical feedback.

7. The AI ​​dialectical treatment model system applicable to multidimensional and differentiated environments according to claim 5, characterized in that, The purpose of developing an adaptive classification algorithm for multi-ethnic physical characteristics is to construct a classification model that can identify the unique physical characteristics of the Miao and Tibetan ethnic minorities. The model is trained based on case data from minority-populated areas and achieves precise and personalized cross-ethnic physical identification by integrating traditional Chinese medicine theory and ethnic medical knowledge.

Images