Deep learning-based AI tongue picture health analysis and diagnosis method and diagnosis instrument

Abstract

The invention provides an AI tongue picture health analysis and diagnosis method and diagnostic apparatus based on deep learning, the AI technology is used for tongue picture health analysis and diagnosis, and multi-modal tongue picture parameters of a patient are analyzed and calculated by collecting tongue sensing data of the patient: tongue color, tongue shape, tongue coating thickness, crack index, tongue surface humidity and tongue temperature deviation. And calculating the physique score, disease risk index and health degree of the patient according to the multi-modal tongue picture parameters. And identifying tongue picture parameter characteristics of the patient by using an AI tongue picture health analysis model, outputting tongue diagnosis results and rehabilitation measures matched with the tongue picture parameter characteristics, and generating a tongue picture health analysis and diagnosis report. A patient can carry out tongue picture traditional Chinese medicine self-diagnosis or assist a doctor in diagnosis, the technical threshold is lowered, application of the intelligent technology in the traditional Chinese medicine tongue picture is enhanced, and the intelligent diagnosis capacity is improved. The patient is prevented from consuming time and the treatment efficiency is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of intelligent medical treatment, in particular to an AI tongue health analysis and diagnosis method based on deep learning, an AI tongue health diagnosis and analysis diagnosis instrument, an electronic device and a computer readable storage medium. BACKGROUND

[0002] The way of disease health analysis and diagnosis according to tongue in clinic mainly relies on traditional Chinese medicine theory for disease diagnosis.

[0003] Traditional Chinese medicine tongue diagnosis analyzes the health status and disease condition of the human body by observing the shape, color and texture of the tongue. Tongue appearance mainly includes tongue body and tongue fur:

[0004] 1. Tongue body: reflects the deficiency and excess of zang-fu organs and the abundance and decline of qi and blood. For example, red and ruddy tongue body indicates abundant qi and blood, and pale tongue body may indicate insufficient qi and blood.

[0005] 2. Tongue fur: reflects the nature, depth and growth and decline of pathogenic factors. For example, thin and white tongue fur is normal, and thick and greasy tongue fur may indicate internal retention of dampness.

[0006] In specific diagnosis, Chinese medicine will judge the nature of the disease according to different characteristics of tongue appearance. For example:

[0007] 1. Red on both sides of the tongue, which may indicate liver fire.

[0008] 2. Red tongue tip, which may indicate heart fire.

[0009] 3. Red tongue and cough, which may indicate lung fire.

[0010] 4. Red tongue, yellow fur and bad breath, which may indicate excessive stomach fire.

[0011] 5. Red tongue with little fur, which may indicate kidney fire or heat deficiency.

[0012] In addition, the shape of the tongue body, such as whether there are cracks and tooth marks, and the flexibility of the tongue, are also important observation contents of Chinese medicine tongue diagnosis.

[0013] With the rise of intelligent medical technology, the way of disease health analysis and diagnosis according to tongue in traditional clinic, although it can analyze the early diagnosis, treatment and prognosis of diseases by comprehensively analyzing tongue appearance characteristics such as tongue body and tongue fur, still has some application defects:

[0014] The way of analyzing tongue appearance and diagnosing diseases in traditional Chinese medicine in clinic requires that the diagnosing doctors have rich clinical experience and have high technical threshold, and a lot of practice is needed to learn Chinese medicine tongue appearance, so the learning difficulty is great;

[0015] The traditional Chinese medicine outpatient registration, queuing and other processes need to spend a lot of time, and sometimes patients need to wait for a long time to register (some users will continuously wait for the registration of some well-known Chinese doctors, and the waiting time is longer), so the efficiency is relatively low.

[0016] The traditional Chinese medicine develops slowly, cannot be intelligently developed, cannot be combined with a deeper algorithm model to perform clinical auxiliary analysis and diagnosis, and cannot provide good auxiliary assistance for Chinese doctors to improve the diagnosis efficiency. SUMMARY

[0017] In order to solve the technical problems existing in the prior art, the present application provides the following technical solutions:

[0018] On the one hand, an AI tongue image health analysis and diagnosis method based on deep learning is provided, the method is realized by an electronic device, and the method comprises:

[0019] S1, collecting tongue sensing data of a patient and uploading and saving the tongue sensing data to a background server, wherein the tongue sensing data comprises a tongue picture and tongue surface temperature and humidity;

[0020] S2, the background server analyzes and calculates multi-modal tongue image parameters of the patient according to the tongue sensing data: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and calculates the constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters;

[0021] S3, inputting the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient into a pre-deployed AI tongue image health analysis model, identifying the tongue image parameter characteristics of the patient through the AI tongue image health analysis model, and outputting a tongue diagnosis result and a rehabilitation measure matched with the tongue image parameter characteristics;

[0022] S4, generating a tongue image health analysis and diagnosis report of the patient according to the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measure.

[0023] Preferably, S2 further comprises:

[0024] The background server pre-constructs an electronic medical record file of the patient in a HIS system;

[0025] When the tongue sensing data of the patient is received, the tongue sensing data is saved into the electronic medical record file of the patient.

[0026] Preferably, S3 further comprises:

[0027] The background server saves the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measure of the patient into the electronic medical record file.

[0028] Preferably, the method for generating the AI tongue health analysis model comprises:

[0029] Collecting tongue sensing data of a plurality of patients and preprocessing the data;

[0030] Performing feature engineering on the tongue sensing data to extract corresponding tongue parameter features: analyzing and calculating a plurality of modal tongue parameter of the patients, including tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity, and tongue temperature deviation, according to the tongue sensing data, and calculating a constitution score, a disease risk index, and a health degree of the patients according to the plurality of modal tongue parameter;

[0031] Performing feature labeling on the tongue parameter features to label all tongue diagnosis results consistent with the tongue parameter features and corresponding rehabilitation measures;

[0032] Statistically analyzing the labeled features of each patient to obtain a feature set, and dividing the feature set into a training set and a validation set according to a preset ratio;

[0033] Inputting the training set into a pre-deployed CNN-LSTM hybrid model to perform feature learning training, and generating an initial AI tongue health analysis model;

[0034] Verifying the recognition performance of the initial AI tongue health analysis model using the validation set:

[0035] If the verification is passed, deploying and applying the AI tongue health analysis model to a background server;

[0036] Otherwise, repeating the above steps.

[0037] Preferably, the constitution score is calculated as follows:

[0038] ,

[0039] Weight distribution (based on clinical data statistics):

[0040] ,

[0041] ;

[0042] Constitution type determination rule:

[0043] ;

[0044] Wherein:

[0045] C, S, T, F, H, and W represent tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity, and tongue temperature deviation, respectively, wherein:

[0046] ,

[0047] Red = 1.0, Pale white = 0.2, Purple = 0.8, Red for hot syndrome, pale white for deficiency cold, purple for blood stasis; based on clinical data statistics, wi is the weight of the ith color;

[0048] S = a ⋅ width-length ratio + b ⋅ tooth mark depth (fatty tongue > 0.7, thin tongue < 0.3), fatty tongue is mainly wet, thin tongue is mainly yin deficiency;

[0049] Thin fur < 0.3, thick fur > 0.7;

[0050] F = ∑ crack length ⋅ depth weight, no crack = 0, deep crack > 0.6;

[0051] H = (sensor humidity value - baseline value) / normal range, dry tongue < 0.3;

[0052] W = (measured temperature - 36.5) / 2, low temperature < 0, fever > 0.5.

[0053] Preferably, the disease risk index is calculated as follows:

[0054] ,

[0055] Wherein, the index is statistically obtained by a logistic regression model:

[0056] k1 = 1.2 (red / purple tongue associated with inflammation), k2 = 0.8 (crack associated with chronic disease), k3 = 1.0 (high temperature associated with infection);

[0057] Risk level:

[0058] Low risk (< 0.3): normal or sub-health;

[0059] Medium risk (0.3-0.6): suggest digestive / metabolic problems;

[0060] High risk (> 0.6): cardiovascular / immune system warning.

[0061] Preferably, the health degree is calculated as follows:

[0062] ,

[0063] Wherein, the health degree is statistically obtained by a time series model:

[0064] Parameter change rate i: indicates the 24-hour fluctuation standard deviation of tongue color, humidity or temperature, i represents tongue color, humidity or temperature;

[0065] Threshold setting:

[0066] AC > 0.3 (inflammation activity);

[0067] AH < -0.2 (dehydration warning);

[0068] AW > 0.4 (acute fever).

[0069] Verification method: ROC curve analysis needs to be carried out through clinical trials (such as 1000 cases of tongue image data) to ensure that the sensitivity is > 85% and the specificity is > 90%.

[0070] In another aspect, an AI tongue health diagnosis analysis diagnosis instrument based on deep learning is provided, which is used to implement the above-mentioned AI tongue health analysis diagnosis method based on deep learning, and the AI tongue health diagnosis analysis diagnosis instrument comprises:

[0071] A camera module is configured to collect tongue pictures of a patient and feed back to a processor;

[0072] A temperature and humidity sensor is configured to collect tongue surface temperature and humidity of the patient and feed back to the processor;

[0073] A processor is configured to preprocess the tongue pictures and the tongue surface temperature and humidity of the patient, and perform local storage and uploading;

[0074] A memory is configured to buffer the tongue pictures and the tongue surface temperature and humidity of the patient;

[0075] A touch screen is configured to input control parameters;

[0076] A wireless module is configured to upload the tongue pictures and the tongue surface temperature and humidity of the patient to a background server;

[0077] A power supply is configured to supply power;

[0078] The camera module, the temperature and humidity sensor, the memory, the touch screen, the wireless module and the power supply are respectively electrically connected with the processor;

[0079] The AI tongue health diagnosis analysis diagnostic instrument is connected with the background server in communication through a wireless module, uploads the tongue picture and the tongue surface temperature and humidity of the patient to the background server, analyzes and calculates the multi-modal tongue image parameters of the patient, including tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, according to the tongue sensing data of the background server, and calculates the constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters; the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient are input into the pre-deployed AI tongue health analysis model, the tongue image parameter characteristics of the patient are identified through the AI tongue health analysis model, and the tongue diagnosis result and rehabilitation measures matched with the tongue image parameter characteristics are output; and a tongue health analysis diagnosis report of the patient is generated according to the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measures.

[0080] In another aspect, an electronic device is provided, including: a processor; a memory having computer readable instructions stored thereon, which, when executed by the processor, implement any one of the above deep learning-based AI tongue health analysis diagnosis methods.

[0081] In another aspect, a computer readable storage medium is provided, which stores at least one instruction, which is loaded and executed by a processor to implement any one of the above deep learning-based AI tongue health analysis diagnosis methods.

[0082] The technical scheme provided by the embodiments of the present application has at least the following beneficial effects:

[0083] The present application applies AI technology to tongue health analysis and diagnosis, collects tongue sensing data of a patient, analyzes and calculates multi-modal tongue image parameters of the patient, including tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and calculates constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters. Then, the AI tongue health analysis model is used to identify tongue image parameter characteristics of the patient, and outputs tongue diagnosis result and rehabilitation measures matched with the tongue image parameter characteristics, and generates a tongue health analysis diagnosis report. This can enable the patient to perform tongue image TCM self-diagnosis or assist doctors in diagnosis, reduce the technical threshold, strengthen the application of intelligent technology in TCM tongue image, and improve the intelligent diagnosis capability. This can avoid wasting time of the patient and improve the efficiency of medical treatment. BRIEF DESCRIPTION OF DRAWINGS

[0084] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed in the embodiment description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0085] Figure 1 is a deep learning-based AI tongue health analysis and diagnosis method flowchart provided by an embodiment of the present application;

[0086] Figure 2 is a generation process schematic diagram of an AI tongue health analysis model provided by an embodiment of the present application;

[0087] Figure 3 is an integrated system structure block diagram of a diagnostic instrument provided by an embodiment of the present application;

[0088] Figure 4 is a structural schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0089] The technical solutions in the present application will be described below with reference to the drawings.

[0090] In the embodiments of the present application, the words such as "example", "for example" and the like are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific manner. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0091] In the embodiments of the present application, "image" and "picture" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent. "Of", "corresponding" and "corresponding" can be used interchangeably at times. It should be pointed out that when the distinction is not emphasized, the meanings expressed are consistent.

[0092] In the embodiments of the present application, sometimes the subscript such as W1 may be mistakenly used in the form of non-subscript such as W1. When the distinction is not emphasized, the meanings expressed are consistent.

[0093] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail with reference to the drawings and specific embodiments.

[0094] The embodiment of the present application provides an AI tongue health analysis and diagnosis method based on deep learning, which can be realized by an electronic device, which can be a terminal or a server. Figure 1 As shown in the flow chart of the AI tongue health analysis and diagnosis method based on deep learning, the processing flow of the method can include the following steps:

[0095] S1, collecting tongue sensing data of a patient and uploading and saving the tongue sensing data to a background server, wherein the tongue sensing data includes tongue pictures and tongue surface temperature and humidity;

[0096] S2, analyzing and calculating, by the background server, multi-modal tongue image parameters of the patient according to the tongue sensing data, including tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and calculating constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters;

[0097] S3, inputting the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient into a pre-deployed AI tongue health analysis model, identifying tongue image parameter features of the patient through the AI tongue health analysis model, and outputting tongue diagnosis results and rehabilitation measures matched with the tongue image parameter features;

[0098] S4, generating a tongue health analysis and diagnosis report of the patient according to the tongue image parameter features, the tongue diagnosis results and the rehabilitation measures.

[0099] Preferably, S2 further includes:

[0100] The background server pre-constructs an electronic medical record file of the patient in a HIS system;

[0101] When the tongue sensing data of the patient is received, the tongue sensing data is saved into the electronic medical record file of the patient.

[0102] Preferably, S3 further includes:

[0103] The background server saves the tongue image parameter features, the tongue diagnosis results and the rehabilitation measures of the patient into the electronic medical record file.

[0104] In the embodiment, the attached drawings can be combined Figure 3The diagnostic instrument and the background are implemented, the diagnostic instrument can collect the tongue sensing data of the patient and upload the background through wireless, the background analyzes the tongue sensing data, and through the AI tongue image health analysis model deployed in advance on the background, the tongue image parameter characteristics of the patient are identified, and the tongue diagnosis result and the rehabilitation measures matched with the tongue image parameter characteristics are output (according to the tongue image characteristics, the corresponding disease type, tongue image symptom degree and corresponding TCM rehabilitation suggestion (such as TCM drink suggestion or Chinese herbal medicine formula for treating weak spleen and stomach) are automatically identified and matched). The background can generate the tongue image health analysis diagnosis report of the patient according to the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measures, and push the report to the doctor PC end or the patient APP end, so as to assist the doctor in TCM diagnosis and quickly participate in tongue image analysis.

[0105] The specific application design is as follows:

[0106] Hardware design of diagnostic instrument

[0107] 1. Camera module:

[0108] High-resolution camera (such as 1080p or higher), supporting auto-focus and color calibration.

[0109] Equipped with ring-shaped LED fill light to ensure accurate capture of tongue image under different lighting conditions.

[0110] 2. Sensor module:

[0111] Optional infrared sensor or spectrum sensor for detecting physiological parameters such as tongue surface humidity and temperature.

[0112] 3. Processor and storage:

[0113] Built-in high-performance low-power processor (such as ARM Cortex series), supporting real-time AI calculation.

[0114] Built-in storage space, supporting local data storage and cloud synchronization.

[0115] 4. Display screen and interaction:

[0116] Small touch screen (such as 2-3 inches) for displaying tongue image and analysis results.

[0117] Support Bluetooth or Wi-Fi connection for data synchronization with the background.

[0118] 3. Battery and charging:

[0119] Built-in rechargeable lithium battery, supporting USB-C fast charging, with at least 24 hours of battery life.

[0120] The tongue image recognition model in the background is based on a CNN convolutional neural network (in this embodiment, a CNN-LSTM hybrid model is preferred, which can learn mixed features from multi-modal data). The training data includes a large number of tongue image pictures and corresponding health labels.

[0121] Support multi-dimensional analysis: tongue color (red, light, purple, etc.), tongue fur (thickness, color), cracks, tooth marks, etc.

[0122] Health assessment system:

[0123] Combining traditional Chinese medicine tongue diagnosis theory and modern medical data, personalized health recommendations are provided.

[0124] Support preliminary screening of common health problems (such as dampness, qi deficiency, blood stasis, etc.).

[0125] User interface:

[0126] Simple and intuitive UI design, supports multiple languages.

[0127] Display analysis results through the APP or device screen and provide health improvement suggestions.

[0128] The diagnostic instrument can be used by patients at home. After collecting data, the data is uploaded to the background. After the doctor analyzes the data, confirms it (the TCM doctor confirms that it is correct, inputs the report recommendation, forwards the instruction, and pushes it to the corresponding port; if there is an adjustment, it can be adjusted) and then issued by the background, thereby greatly improving the efficiency of traditional Chinese medicine tongue diagnosis.

[0129] It can also be used to arrange patients to use the diagnostic instrument when registering, write the collected data into the HIS system, and automatically traverse the patient's AI analysis and diagnosis results in the background. After the doctor analyzes the data, confirms it, and then issues it by the background.

[0130] In this embodiment, an "AI tongue image health analysis model" is deployed in the background.

[0131] First, the background server analyzes and calculates the patient's multi-modal tongue image parameters based on the tongue sensor data: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity, and tongue temperature deviation, and calculates the patient's constitution score, disease risk index, and health degree based on the multi-modal tongue image parameters; The specific calculation is as follows:

[0132] Preferably, the constitution score is calculated as follows:

[0133] ,

[0134] Weight allocation (based on clinical data statistics):

[0135] ,

[0136] ;

[0137] Body type determination rule:

[0138] ;

[0139] Wherein:

[0140] D, S, T, F, H, W represent tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity, tongue temperature deviation, wherein:

[0141] ,

[0142] Red = 1.0, light white = 0.2, purple = 0.8, red for heat syndrome, light white for deficiency cold, purple for blood stasis; Based on clinical data statistics, wi is the weight of the ith color;

[0143] S = a ⋅ width-length ratio + b ⋅ tooth mark depth (fat tongue > 0.7, thin tongue < 0.3), fat tongue is mainly wet, thin tongue is mainly yin deficiency;

[0144] , thin fur <0.3, thick fur >0.7;

[0145] F = ∑ crack length ⋅ depth weight, no crack = 0, deep crack >0.6;

[0146] H = (sensor humidity value - baseline value) / normal range, dry tongue <0.3;

[0147] W = (measured temperature - 36.5) / 2, low temperature <0, fever >0.5.

[0148] Preferably, the disease risk index is calculated as follows:

[0149] ,

[0150] Wherein, the index is statistically obtained by a logistic regression model:

[0151] k1 = 1.2 (red / purple tongue associated with inflammation), k2 = 0.8 (crack associated with chronic disease), k3 = 1.0 (high temperature associated with infection);

[0152] Risk level:

[0153] Low risk (<0.3): normal or sub-health;

[0154] Medium risk (0.3-0.6): suggest digestive / metabolic problems;

[0155] High risk (>0.6): cardiovascular / immune system warning.

[0156] Preferably, the health degree is calculated as follows:

[0157] ,

[0158] Wherein, the health degree is calculated by time series model:

[0159] Parameter change rate i: indicates the 24-hour fluctuation standard deviation of tongue color, humidity or temperature, i represents tongue color, humidity or temperature;

[0160] Threshold setting:

[0161] ΔC>0.3ΔC>0.3 (inflammation activity);

[0162] ΔH<−0.2ΔH<−0.2 (dehydration warning);

[0163] ΔW>0.4ΔW>0.4 (acute fever).

[0164] Through the above analysis and calculation of the patient's multi-modal tongue image parameters: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, the multi-modal tongue image features of the patient can be fused to calculate the tongue image symptoms. Through quantitative analysis, quantitative evaluation and feasibility evaluation can be realized. The clinical tongue diagnosis features of the patient can be quantitatively recognized and analyzed, and the patient's constitution score, disease risk index and health degree can be calculated for model prediction. Through the fusion analysis of multi-modal tongue image features, the patient's state can be accurately identified and the corresponding TCM rehabilitation strategy can be recommended.

[0165] Secondly, the background analyzes the patient's information through the model, identifies its features and outputs the corresponding tongue image state features and corresponding TCM rehabilitation strategies. The background inputs the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient into the pre-deployed AI tongue image health analysis model, identifies the tongue image parameter features of the patient through the AI tongue image health analysis model, and outputs the tongue diagnosis results and rehabilitation measures matched with the tongue image parameter features.

[0166] The specific model application is as follows:

[0167] As shown in Figure 2 Preferably, the generation method of the AI tongue image health analysis model comprises:

[0168] Collecting tongue sensing data of a plurality of patients and preprocessing;

[0169] The tongue sensing data is subjected to feature engineering, and corresponding tongue image parameter features are extracted: according to the tongue sensing data, multi-modal tongue image parameters of the patient are analyzed and calculated: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and the constitution score, disease risk index and health degree of the patient are calculated according to the multi-modal tongue image parameters;

[0170] The tongue image parameter features are subjected to feature labeling, and all tongue diagnosis results and corresponding rehabilitation measures conforming to the tongue image parameter features are labeled;

[0171] The labeled features of each patient are counted to obtain a feature set, and the feature set is divided into a training set and a validation set according to a preset ratio;

[0172] The training set is input into a pre-deployed CNN-LSTM hybrid model for feature learning and training to generate an initial AI tongue image health analysis model;

[0173] The validation set is used to verify the recognition performance of the initial AI tongue image health analysis model:

[0174] If the verification is passed, the AI tongue image health analysis model is deployed and applied to a background server;

[0175] Otherwise, the above steps are repeated.

[0176] The specific model training is as follows:

[0177] The present scheme constructs a TCM tongue diagnosis intelligent system through a closed-loop process of multi-modal data acquisition, feature engineering, hybrid model training and intelligent diagnosis. The core architecture includes:

[0178] Front-end perception layer: tongue image acquisition equipment (high-resolution camera, temperature and humidity sensor); the diagnostic instrument shown in FIG. 1 can be referred to; Figure 3

[0179] Data calculation layer: multi-modal parameter calculation engine; calculated by the background.

[0180] AI model layer: CNN-LSTM hybrid model; the model architecture is known and will not be described here.

[0181] Application layer: diagnostic report generation and doctor-patient interaction platform.

[0182] Model construction steps:

[0183] 1. Data acquisition and preprocessing (preliminary feature engineering)

[0184] Data source (may be a hospital, arrange patients to collect data, can collect data through a diagnostic instrument or other equipment): ​

[0185] Tongue image: Collect the front and side images of the tongue using a medical-grade camera with a resolution of ≥ 12 million pixels using a standard D65 light source. Save the images in RGB format with a size of 1024 x 768.

[0186] Tongue surface temperature and humidity: Use a MEMS temperature and humidity sensor with an accuracy of ±0.2°C and ±3% RH to measure the steady-state value in the middle of the tongue (take the average value for 10 seconds).

[0187] Preprocessing method:

[0188] Image normalization:

[0189] Light compensation based on HSV color space (eliminate reflective areas);

[0190] Use U-Net to segment the tongue outline and crop the non-tongue background;

[0191] Histogram equalization to enhance texture details (such as cracks and moss).

[0192] Temperature and humidity calibration:

[0193] Dynamic compensation according to environmental temperature and humidity (sensor synchronous acquisition) to eliminate measurement deviation.

[0194] 2. Multimodal tongue parameter calculation (core of feature engineering)

[0195] Extract 6 key parameters through computer vision and biosensing algorithms: (Please refer to the calculation method of "physical fitness score, disease risk index, and health degree" above).

[0196] Tongue color analysis: Calculate the Lab* average of the main color area (the first 2 / 3) of the tongue in the LAB color space and map it to the TCM color card (pale white / pale red / purple).

[0197] Tongue shape classification: Detect the tongue contour curvature based on the improved Active Shape Model (ASM), match the ellipticity / thinness index, and divide the tongue into three categories: fat, thin, and tooth marks. Ellipticity threshold: >0.85 for thin tongue, <0.7 for fat tongue.

[0198] Moss thickness: Use ResNet-50 to extract tongue texture features and calculate the moss coverage rate (%) through the gray level co-occurrence matrix (GLCM). Classification: thin (<30%), medium (30-60%), thick (>60%). This place selects fat tongue >0.7 and thin tongue <0.3.

[0199] Crack index: Extract the crack topology based on Canny edge detection and calculate the total length and density (unit area crack pixel ratio).

[0200] Tongue surface humidity: The relative humidity deviation (ΔRH%) was calculated by comparing the humidity sensor measurement after environmental compensation with the healthy baseline (mean value of healthy population ± 2σ).

[0201] Tongue temperature deviation: The absolute temperature difference (ΔT ℃) was calculated by comparing the temperature sensor measurement after environmental compensation with the healthy baseline (36.5 ± 0.3 ℃).

[0202] 3. AI Tongue Health Analysis Model Training (CNN-LSTM Hybrid Architecture)

[0203] Model Structure Design:

[0204] Input Layer:

[0205] Image Branch: Tongue RGB image (224x224) → ResNet-50 to extract visual features (output 1024-dimensional vector);

[0206] Parameter Branch: 6-dimensional tongue parameters (after standardization) → fully connected layer mapped to 64-dimensional vector.

[0207] Feature Fusion:

[0208] Concatenate image features and parameter features (1088-dimensional), input LSTM layer (128 units) to capture temporal dependency relationships (such as dynamic correlation between tongue color and temperature and humidity).

[0209] The feature set can be divided into training set and validation set according to the ratio of 8:2.

[0210] Output Layer:

[0211] Softmax classifier outputs 8 tongue diagnosis results (deficiency of Yin / Yang / Heat, etc.);

[0212] Regression head outputs the priority score of rehabilitation measures (0-1, such as traditional Chinese medicine prescription A=0.92, acupuncture B=0.75).

[0213] Training Strategy:

[0214] Data Augmentation: Randomly rotate tongue image (±15°), color jitter (ΔHSV≤10%), Gaussian noise (σ=0.01) to expand data set;

[0215] Loss Function:

[0216] Classification Loss: Focal Loss (α=0.25, γ=2) to solve class imbalance;

[0217] Regression Loss: Smooth L1 Loss to constrain rehabilitation measure score.

[0218] Optimizer: AdamW (lr=1e-4, weight decay=1e-5), Early Stopping Strategy (patience=10).

[0219] 4. Model Validation and Deployment‌

[0220] Validation Metrics‌:

[0221] Classification Performance: Calculate Precision, Recall, and require F1-score ≥ 0.85 (admin can complete validation according to F1 score, etc.).

[0222] Regression Performance: Pearson Correlation Coefficient between Rehabilitation Measures Score and Expert Annotation ≥ 0.9.

[0223] Deployment Scheme‌:

[0224] Edge-Cloud Collaboration‌:

[0225] Front-end device runs a lightweight model (MobileNetV3 extracts tongue feature);

[0226] Background server performs LSTM inference and report generation, response time ≤ 1.5 seconds.

[0227] A / B Testing‌:

[0228] Compare AI diagnosis with 3 deputy chief physician diagnoses for consistency (Kappa coefficient ≥ 0.7).

[0229] Three, Clinical Application Examples‌

[0230] Case 1: Auxiliary Diagnosis of Diabetes with Heatiness Syndrome‌

[0231] Input Data‌:

[0232] Tongue image: red tongue (a* value = 35.2), yellow and thick fur (coverage rate 68%);

[0233] Temperature and humidity: tongue temperature 37.1℃ (ΔT = +0.6℃), humidity deviation ΔRH = +12%.

[0234] Model Output‌:

[0235] Tongue diagnosis result: heatiness accumulation (confidence 92%);

[0236] Rehabilitation measures: heat-clearing and dampness-removing prescription (score 0.94), diet taboo list (score 0.88).

[0237] Clinical Verification: Consistent with physician diagnosis results, patient's tongue fur thickness decreased to 45% after executing the recommendations (two-week follow-up).

[0238] Case 2: Sub-health state warning

[0239] Input data:

[0240] Pale white tongue color (L=82), crack index 0.15 (mild);

[0241] Tongue temperature 35.9℃ (ΔT=-0.6℃).

[0242] Model output:

[0243] Constitution score: 62 (yang deficiency tendency);

[0244] Warning prompt: Disease risk index 0.48 (suggesting strengthening of warming and tonifying regulation).

[0245] Technical advantages:

[0246] Multi-modal data fusion: For the first time, tongue surface temperature and humidity sensor data are combined with visual features to model, breaking through the limitations of traditional tongue diagnosis relying only on morphology; LSTM is used to capture the dynamic relationship between tongue color and temperature (such as red tongue with low temperature abnormalities in yin deficiency syndrome).

[0247] Quantitative innovation in traditional Chinese medicine: A mathematical mapping model of tongue parameters and constitution / disease is established.

[0248] Test data shows that the system improves the efficiency of tongue diagnosis by 3 times (single diagnosis ≤2 minutes), and the consistency of diagnosis is improved to 89%.

[0249] Through the above technical solutions, the system realizes the standardization, intelligentization, and replicability of traditional Chinese medicine tongue diagnosis, providing an efficient tool for "preventing disease" health management.

[0250] Reference Figure 3 On the other hand, an AI tongue health diagnosis analysis diagnostic instrument based on deep learning is provided, which is used to realize the above-mentioned AI tongue health analysis diagnosis method based on deep learning, and the AI tongue health diagnosis analysis diagnostic instrument comprises:

[0251] A camera module for capturing tongue pictures of patients and feeding back to a processor;

[0252] A temperature and humidity sensor for collecting tongue surface temperature and humidity of patients and feeding back to a processor;

[0253] A processor for preprocessing the tongue pictures and tongue surface temperature and humidity of patients, and for local storage and uploading;

[0254] a memory for caching the tongue picture and the tongue surface temperature and humidity of the patient;

[0255] a touch screen for inputting control parameters;

[0256] a wireless module for uploading the tongue picture and the tongue surface temperature and humidity of the patient to a background server;

[0257] a power supply for power supply;

[0258] The camera module, temperature and humidity sensor, memory, touch screen, wireless module and power supply are electrically connected with the processor respectively.

[0259] The AI tongue health diagnosis analysis diagnostic instrument is in communication connection with the background server through the wireless module, uploads the tongue picture and the tongue surface temperature and humidity of the patient to the background server, analyzes and calculates the multi-modal tongue image parameters of the patient, i.e. tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, according to the tongue sensing data, calculates the constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters, inputs the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient into the pre-deployed AI tongue health analysis model, identifies the tongue image parameter characteristics of the patient through the AI tongue health analysis model, and outputs the tongue diagnosis result and rehabilitation measures matched with the tongue image parameter characteristics; generates the tongue image health analysis diagnosis report of the patient according to the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measures.

[0260] The carrier or size of the diagnostic instrument and other hardware systems are not limited in the embodiment. The working principle of the diagnostic instrument is understood in combination with the previous method steps, which will not be described here.

[0261] The hardware design and application requirements of the diagnostic instrument are as follows:

[0262] 1. Core functions

[0263] Tongue image collection: capture the color, shape, texture, cracks, and fur of the tongue through a high-resolution camera and light source system.

[0264] AI analysis: use deep learning algorithms to analyze tongue images and identify health-related features (such as tongue color, fur thickness, and crack distribution).

[0265] Health assessment: provide health status assessment (such as constitution type and potential health problems) based on tongue image features, combined with traditional Chinese medicine theory and modern medical data.

[0266] Data storage and tracking: record each tongue diagnosis result to support health trend analysis.

[0267] Portable design: The device is small and lightweight, suitable for carrying around and easy to operate.

[0268] 2. Portable design

[0269] Size and weight:

[0270] The device is controlled in the size of a palm (such as 10cm x 5cm x 2cm), and the weight does not exceed 200g.

[0271] Material:

[0272] The shell is made of lightweight and durable materials (such as ABS plastic or aluminum alloy), and is waterproof and drop-proof.

[0273] Folding or modular design:

[0274] The camera module can be folded or detached, making it easy to store and carry.

[0275] 3. Use scenarios

[0276] Home health monitoring: Users can perform daily tongue diagnosis at home to understand their health status.

[0277] Auxiliary diagnosis in TCM clinics: Provide objective tongue analysis data for TCM doctors to assist in diagnosis.

[0278] Health management agencies: Used for health screening and long-term health management.

[0279] 4. Technical challenges and solutions

[0280] Tongue image standardization:

[0281] Through standardized light source and shooting angle, reduce environmental light interference.

[0282] AI model accuracy:

[0283] Use large-scale labeled data set to train model, and combine expert feedback to continuously optimize.

[0284] User operation simplicity:

[0285] Design one-key operation process, users only need to align the tongue and press the button to complete the detection.

[0286] 5. Market prospects

[0287] Target users:

[0288] Ordinary users who care about health, TCM enthusiasts, chronic disease patients, health management agencies, etc.

[0289] Competitive advantage:

[0290] Combining AI technology and TCM theory, scientific and personalized health assessment is provided.

[0291] Portable design, suitable for daily use.

[0292] The specific diagnostic instrument appearance, size and other structural design are not considered in the scope of the application, and the application system and functions of the application can be integrated.

[0293] Through the above design, the portable AI tongue diagnosis instrument can become an important tool for modern health management, and promote the intelligent development of TCM diagnosis technology.

[0294] Figure 4 is a structural schematic diagram of an electronic device provided by an embodiment of the application, as Figure 4 shown, the electronic device 410 can include a first processor 2001.

[0295] Optionally, the electronic device 410 can also include a memory 2002 and a transceiver 2003.

[0296] Among them, the first processor 2001 and the memory 2002 and the transceiver 2003, such as can be connected through the communication bus.

[0297] The following will be combined Figure 4 The specific components of the electronic device 410 will be introduced:

[0298] Among them, the first processor 2001 is the control center of the electronic device 410, which can be a processor, or a plurality of processing elements. For example, the first processor 2001 is one or more central processing units (CPU), which can also be an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the application, such as one or more microprocessors (digital signal processor, DSP), or one or more field programmable gate arrays (FPGA).

[0299] Optionally, the first processor 2001 can execute the various functions of the electronic device 410 by running or executing the software program stored in the memory 2002 and calling the data stored in the memory 2002.

[0300] In a specific implementation, as an embodiment, the first processor 2001 can include one or more CPUs, such as Figure 4 CPU0 and CPU1 shown in the figure.

[0301] In a particular implementation, as one embodiment, the electronic device 410 can also include multiple processors, such as the first processor 2001 and the second processor 2004 shown in FIG. 20. Figure 4 Each of these processors can be a single-CPU or a multi-CPU. The processor here can refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0302] The memory 2002 is configured to store a software program for implementing the solution of the present application, and the first processor 2001 is configured to control the execution of the software program. The implementation can refer to the above method embodiments, and thus details are not repeated here.

[0303] Optionally, the memory 2002 can be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disk storage, a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to this. The memory 2002 can be integrated with the first processor 2001 or exist independently and be coupled to the first processor 2001 through an interface circuit (not shown in FIG. 20) of the electronic device 410. The embodiments of the present application do not make a specific limitation in this regard. Figure 4

[0304] The transceiver 2003 is configured to communicate with a network device or a terminal device.

[0305] Optionally, the transceiver 2003 can include a receiver and a transmitter (not shown separately in FIG. 20). The receiver is configured to implement the receiving function, and the transmitter is configured to implement the transmitting function. Figure 4

[0306] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently and be coupled to the first processor 2001 through an interface circuit (not shown in FIG. 20) of the electronic device 410.​​Figure 4 The first processor 2001 is coupled with the electronic device 410 shown in FIG. 4, and embodiments of the present application do not make specific limitations thereto.

[0307] It should be noted that, ​ The structure of the electronic device 410 shown in FIG. 4 does not constitute a limitation on the router, and the actual knowledge structure recognition device can include more or fewer components than those shown, or combine certain components, or different component arrangements.

[0308] In addition, the technical effects of the electronic device 410 can refer to the technical effects of the AI tongue health analysis and diagnosis method based on deep learning described above. Here, it will not be described again.

[0309] It should be understood that the first processor 2001 in embodiments of the present application can be a central processing unit (CPU), and the processor can also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), ready programmable gate arrays (FPGAs) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0310] It should also be understood that the memory in the embodiments of the present application can be volatile or nonvolatile memory, or can include both volatile and nonvolatile memory. The nonvolatile memory can be read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrically EPROM (EEPROM), or flash memory. The volatile memory can be random access memory (RAM) used as external cache. By way of example, and not limitation, many forms of random access memory (RAM) are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchlink DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0311] The above-described embodiments can be implemented, in whole or in part, by software, hardware (such as a circuit), firmware or any combination thereof. When implemented by software, the above-described embodiments can be implemented, in whole or in part, in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable AI tongue health diagnostic analysis diagnostic instrument. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD), or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0312] It should be understood that the term "and / or" herein merely describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B, which can represent three cases of A alone, A and B together, and B alone, where A and B can be singular or plural. In addition, the character " / " herein generally represents that the associated objects before and after are an "or" relationship, but can also represent an "and / or" relationship, which can be understood in the context before and after.

[0313] In the present application, "at least one" means one or more, and "multiple" means two or more. "At least one of the following" or the like means any combination of the items, including any combination of single or multiple items. For example, at least one of a, b, or c can mean a, b, c, a-b, a-c, b-c, or a-b-c, where a, b, and c can be single or multiple.

[0314] It should be understood that in various embodiments of the present application, the size of the sequence number of the above-described processes does not mean the order of execution, and the execution order of the processes should be determined by their functions and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0315] Those skilled in the art can clearly understand that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0316] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described devices, AI tongue health diagnosis analysis diagnostic instrument and units can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0317] In several embodiments provided by the present application, it should be understood that the disclosed devices, AI tongue health diagnosis analysis diagnostic instrument and methods can be implemented by other means. For example, the above-described AI tongue health diagnosis analysis diagnostic instrument embodiments are only illustrative, for example, the division of the units is only a logical functional division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units or components shown or discussed can be through some interface, indirect coupling or communication connection between the AI tongue health diagnosis analysis diagnostic instrument or units can be electrical, mechanical or other forms.

[0318] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the present embodiment.

[0319] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0320] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the present application that essentially contribute to the prior art or the parts of the technical solutions can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0321] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A deep learning-based AI tongue image health analysis and diagnosis method, characterized in that, The method comprises: S1, collecting tongue sensing data of a patient and uploading and saving to a background server, the tongue sensing data comprising tongue pictures and tongue surface temperature and humidity; S2, the background server analyzes and calculates the multi-modal tongue image parameters of the patient according to the tongue sensing data: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and calculates the constitution score, disease risk index and health degree of the patient according to the multi-modal tongue image parameters; S3, inputting the multi-modal tongue image parameters, constitution score, disease risk index and health degree of the patient into a pre-deployed AI tongue image health analysis model, identifying the tongue image parameter characteristics of the patient through the AI tongue image health analysis model, and outputting the tongue diagnosis result and rehabilitation measures matched with the tongue image parameter characteristics; S4, generating a tongue image health analysis diagnosis report of the patient according to the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measures. 2.The deep learning-based AI tongue health analysis and diagnosis method of claim 1, wherein In S2, further comprising: The background server pre-constructs the electronic medical record file of the patient in the HIS system; When the tongue sensing data of the patient is received, the tongue sensing data is saved to the electronic medical record file of the patient. 3.The deep learning-based AI tongue health analysis and diagnosis method of claim 1, wherein, In S3, further comprising: The background server saves the tongue image parameter characteristics, the tongue diagnosis result and the rehabilitation measures of the patient to the electronic medical record file. 4.The deep learning-based AI tongue health analysis and diagnosis method of claim 3, wherein, The generation method of the AI tongue image health analysis model comprises: Collecting tongue sensing data of a plurality of patients and preprocessing; Feature engineering is performed on the tongue sensing data to extract corresponding tongue image parameter characteristics: according to the tongue sensing data, the multi-modal tongue image parameters of the patient are analyzed and calculated: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and the constitution score, disease risk index and health degree of the patient are calculated according to the multi-modal tongue image parameters; The tongue image parameter characteristics are labeled, and all tongue diagnosis results and corresponding rehabilitation measures that match the tongue image parameter characteristics are labeled; The labeled features of each patient are counted to obtain a feature set, and the feature set is divided into a training set and a validation set according to a preset proportion; The training set is input into a pre-deployed CNN-LSTM hybrid model for feature learning training to generate an initial AI tongue image health analysis model; The identification performance of the initial AI tongue image health analysis model is verified by using the validation set: If the verification is passed, the AI tongue image health analysis model is deployed and applied to the background server; Otherwise, repeat the above steps. 5.The deep learning-based AI tongue health analysis and diagnosis method of claim 1, wherein The constitution score is calculated as follows: ， Weight distribution (based on clinical data statistics): ， ； Constitution type determination rule: ； Wherein: C, S, T, F, H and W represent tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, wherein: ， Red = 1.0, light white = 0.2, purple = 0.8, red is hot, light white is cold, and purple is blood stasis; based on clinical data statistics, wi is the weight of the ith color; S = α ⋅ width-length ratio + β ⋅ tooth mark depth (fat tongue > 0.7, thin tongue < 0.3), fat tongue is mainly wet, and thin tongue is mainly yin deficiency; , thin moss <0.3, thick moss >0.7; F = ∑ crack length ⋅ depth weight, no crack = 0, deep crack > 0.6; H = (sensor humidity value - baseline value) / normal range, dry tongue <0.3; W = (measured temperature - 36.5) / 2, low temperature <0, fever >0.

5. 6.The deep learning-based AI tongue health analysis and diagnosis method of claim 5, wherein, The disease risk index is calculated as follows: ， Where the index is statistically analyzed by a logistic regression model: k1=1.2 (red / purple tongue associated with inflammation), k2=0.8 (cracks associated with chronic disease), k3=1.0 (high temperature associated with infection); Risk level: Low risk (<0.3): normal or sub-health; Medium risk (0.3-0.6): suggests digestive / metabolic problems; High risk (>0.6): cardiovascular / immune system warning. 7.The deep learning-based AI tongue health analysis and diagnosis method of claim 5, wherein, The health degree is calculated as follows: ， Where the health degree is statistically analyzed by a time series model: Parameter change rate i: indicates the 24-hour fluctuation standard deviation of tongue color, humidity or temperature, i represents tongue color, humidity or temperature; Threshold setting: ΔC>0.3ΔC>0.3 (inflammatory activity); ΔH<−0.2ΔH<−0.2 (dehydration warning); ΔW>0.4ΔW>0.4 (acute fever).

8. An AI tongue image health diagnosis analysis diagnosis instrument based on deep learning, the AI tongue image health diagnosis analysis diagnosis instrument based on deep learning is used to realize the AI tongue image health diagnosis analysis diagnosis method based on deep learning as any one of claims 1-7, characterized in that, The AI tongue health diagnosis analysis diagnostic instrument comprises: A camera module for collecting a patient's tongue picture and feeding back to the processor; A temperature and humidity sensor for collecting the patient's tongue surface temperature and humidity and feeding back to the processor; A processor for preprocessing the patient's tongue picture and tongue surface temperature and humidity, and for local storage and uploading; A memory for buffering the patient's tongue picture and tongue surface temperature and humidity; A touch screen for inputting control parameters; A wireless module for uploading the patient's tongue picture and tongue surface temperature and humidity to a background server; A power supply for power supply; The camera module, temperature and humidity sensor, memory, touch screen, wireless module and power supply are respectively electrically connected with the processor; The AI tongue health diagnosis analysis diagnostic instrument is connected with the background server through the wireless module, uploads the patient's tongue picture and tongue surface temperature and humidity to the background server; according to the tongue sensor data, the background server analyzes and calculates the patient's multi-modal tongue parameters: tongue color, tongue shape, tongue fur thickness, crack index, tongue surface humidity and tongue temperature deviation, and calculates the patient's constitution score, disease risk index and health degree; input the patient's multi-modal tongue parameters, constitution score, disease risk index and health degree into the pre-deployed AI tongue health analysis model, identify the patient's tongue parameter characteristics through the AI tongue health analysis model, and output the tongue diagnosis result and rehabilitation measures matched with the tongue parameter characteristics; according to the tongue parameter characteristics, the tongue diagnosis result and the rehabilitation measures, generate the patient's tongue health analysis diagnosis report.

9. An electronic device, comprising: The electronic device comprises: A processor; A memory having computer readable instructions stored thereon, the computer readable instructions being executed by the processor to implement the method of any one of claims 1 to 7.

10. A computer readable storage medium, characterized in that, The computer readable storage medium stores program code, which can be called and executed by the processor to implement the method of any one of claims 1 to 7.

Images