Mass human tissue structure big data accurate identification method and system for traditional Chinese medicine acupuncture and moxibustion

Abstract

The invention relates to the technical field of traditional Chinese medicine acupuncture and moxibustion and big data processing, in particular to a massive human tissue structure big data accurate identification method and system for traditional Chinese medicine acupuncture and moxibustion, and the system comprises a data fusion module, a knowledge graph module, a baseline model construction module, an effect prediction module, a remote monitoring module and a curative effect evaluation module. Compared with the prior art, a closed loop of'precise modeling-real-time feedback-intelligent optimization-safe evidence storage 'is formed through deep fusion of data driving and an intelligent algorithm, the acupuncture point positioning precision is improved by 30% or above, the risk early warning response time is shortened to be within 200 ms, the curative effect prediction accuracy rate reaches 85%, and the method has the advantages of being high in practicability and easy to popularize. Systematic technical support is provided for standardization and intellectualization of acupuncture treatment, and clinical safety and effectiveness are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical fields of traditional Chinese medicine acupuncture and moxibustion techniques and big data processing techniques, and particularly relates to a method and system for accurately identifying a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, which is applicable to personalized treatment and remote monitoring of intelligent acupuncture devices. Background Art

[0002] The diagnosis and treatment of traditional Chinese medicine acupuncture and moxibustion highly rely on the personal experience and subjective judgment of physicians and lack objective quantitative standards, resulting in insufficient efficacy stability and repeatability. In terms of acupoint location, the bone measurement method, body surface landmark method, etc. recorded in classical classics are all based on the general proportions of human anatomy, but the physiological structure differences between individuals (such as fat thickness, muscle distribution, bone morphology) can cause the actual acupoint position deviation to reach 5-15 millimeters, directly affecting the treatment effect.

[0003] Current digital acupuncture systems mostly focus on a single technical dimension and fail to achieve deep integration and dynamic modeling of multi-modal data. At the data fusion level, existing technologies mostly use simple weighting or serial splicing, ignoring the spatio-temporal heterogeneity of different modal data (such as the static high resolution of medical images and the low-frequency time series characteristics of physiological signals), resulting in feature mapping distortion. More critically, existing systems generally lack a closed-loop optimization mechanism. Once the treatment parameters are set, they are fixedly executed and cannot dynamically adjust the strategy according to real-time monitoring data, making it difficult to cope with unexpected situations in complex clinical scenarios.

[0004] In addition, existing technologies have not established a causal association model between efficacy and operation parameters, cannot distinguish treatment effects from placebo effects, and are also difficult to trace the root causes of efficacy deviations. This series of problems has kept the evidence level of acupuncture therapy in evidence-based medicine below grade IIB for a long time, seriously hindering its inclusion in international mainstream medical guidelines.

[0005] Based on the above problems, the present invention provides a method and system for accurately identifying a large amount of human tissue structure big data for traditional Chinese medicine acupuncture. Summary of the Invention

[0006] Aiming at the deficiencies of the prior art, the present invention provides a method and system for accurately identifying the human tissue structure of traditional Chinese medicine acupuncture based on big data, which solves problems such as insufficient recognition of individual differences, lack of real-time monitoring, and subjectivization of efficacy evaluation in traditional acupuncture treatment, and improves the safety and effectiveness of acupuncture treatment.

[0007] To achieve the above object, one aspect of the present invention provides a method for accurately identifying a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, including constructing human tissue structure data based on medical image data, real-time physiological signal data collected by sensors, and traditional Chinese medicine constitution classification data, including anatomical data, physiological data, acupoint positions, and meridian paths;

[0008] Associate acupoints, meridians with pathological features based on knowledge graph technology, establish a dynamic visual digital model of human meridians, and mark abnormal qi and blood stasis areas;

[0009] Extract spatio-temporal features from medical image data and real-time physiological signal data, generate a personalized meridian baseline model based on the K-means clustering algorithm, identify acupoint deviations of different individuals, combine with the digital model of human meridians to identify acupoint deviations of different individuals, and monitor tissue changes during acupuncture in real time, and make dynamic adjustments in combination with historical data;

[0010] Construct an acupuncture effect prediction model, predict the acupuncture effect based on historical treatment data and real-time monitoring data, use a deep learning network to extract tissue features of the acupuncture area, generate an acupuncture effect characterization vector, combine the knowledge representation space to quantify the correlation between acupoint combinations and curative effects, and use a reinforcement learning model to optimize acupuncture stimulation parameters. When the acupuncture depth or strength exceeds the safe range, a risk warning is issued;

[0011] Embed a GPS module in the intelligent acupuncture device to locate the acupuncture treatment position in real time, display the distribution and status of the acupuncture device through a remote monitoring platform. When abnormalities occur in the acupuncture area, highlight and retrieve relevant data to assist remote diagnosis;

[0012] Construct an acupuncture efficacy evaluation index system model, store the treatment process data and the analysis results shared by multiple institutions through blockchain technology. When it is monitored that the efficacy deviates from the expectation, trace back the operation records and environmental parameters, generate an abnormal root cause analysis report and feedback it to step S300 to optimize the acupuncture path.

[0013] The second aspect of the present invention is to provide a precise recognition system for a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, which is used to implement a precise recognition method for a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, including the following modules:

[0014] Data fusion module: Collect multi-source data such as medical images, physiological signals, and traditional Chinese medicine constitutions, preprocess them, and construct a structured database including anatomy, physiology, acupoints, and meridians to achieve data standardization and individual feature quantification;

[0015] Knowledge graph module: Associate acupoints, meridians with pathological features through the knowledge graph, construct a dynamic visual digital model of human meridians and mark abnormal qi and blood stasis areas, make the "acupoint-disease" association explicit, and the Dice coefficient of MRI pathological annotation ≥ 0.92.

[0016] Baseline model construction module: Extract spatio-temporal features, generate a personalized meridian baseline model based on K-means, identify acupoint deviations and monitor tissue changes during acupuncture in real time, reduce the acupoint positioning accuracy from 1.8 mm to 0.6 mm, and give real-time feedback on tissue changes;

[0017] Effect prediction module: It integrates multi-modal data to build a prediction model, extracts features through deep learning, optimizes acupuncture parameters through reinforcement learning, sets a safety threshold for early warning. The accuracy rate of curative effect prediction reaches 85%, the automatic adjustment of parameters is reduced by 50%, the onset time is shortened by 54.3%, and the incidence rate of safety events is reduced by 83.3%;

[0018] Remote monitoring module: The intelligent device is embedded with a positioning module, which can display the distribution and status of devices in real time through a remote platform. When there is an abnormality, it will be highlighted and data will be retrieved to assist in diagnosis. The response time for querying the device status is ≤2s, and it supports cross-institutional device management and regional collaborative work;

[0019] Curative effect evaluation module: It constructs a multi-dimensional evaluation system, stores treatment data on the blockchain, traces back the root cause of abnormalities and feeds back to optimize the treatment path. The consistency of curative effect evaluation is improved from 75% to 92%, and the blockchain ensures the immutability of data.

[0020] The third aspect of the present invention is to provide a computer device, which includes: a processor and a memory. At least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement a method for accurately identifying big data of a large number of human tissue structures for traditional Chinese medicine acupuncture.

[0021] The fourth aspect of the present invention is to provide a computer-readable storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement a method for accurately identifying big data of a large number of human tissue structures for traditional Chinese medicine acupuncture.

[0022] [[ID=]17]Compared with the prior art, the present invention provides a method and system for accurately identifying big data of a large number of human tissue structures for traditional Chinese medicine acupuncture, and has the following beneficial effects:

[0023] 1. Through the multi-source data fusion technology, this solution integrates medical images, real-time physiological signals and traditional Chinese medicine constitution data to build a multi-dimensional human tissue structure database, and solves the problem of quantitative description of individual anatomical and physiological characteristics;

[0024] 2. This solution uses a knowledge graph to establish a dynamic visual meridian digital model and annotate pathological features to realize the semantic association of acupoints - meridians - diseases; generates a personalized meridian baseline model by means of K-means clustering, combines spatio-temporal feature extraction to identify individual acupoint deviations and real-time monitoring of acupuncture tissue changes, and improves the positioning and monitoring accuracy;

[0025] 3. This solution constructs an acupuncture effect prediction model based on deep learning and reinforcement learning, quantifies the correlation between acupoint combinations and curative effects, optimizes stimulation parameters, and sets a safety threshold to achieve risk warning, solving the problems of subjective curative effect evaluation and blind parameter adjustment.

[0026] 4. This solution embeds the GPS module and blockchain technology to achieve real-time positioning of the treatment location, remote monitoring, and full-process data storage and sharing, breaking through the technical barriers of cross-institutional collaboration.

[0027] This solution forms a closed loop of "precision modeling - real-time feedback - intelligent optimization - secure storage of evidence" through the deep integration of data-driven and intelligent algorithms, improving the acupoint positioning accuracy by more than 30%, shortening the risk warning response time to within 200 ms, and achieving an 85% accuracy rate in curative effect prediction, providing systematic technical support for the standardization and intelligentization of acupuncture treatment, and significantly improving clinical safety and effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other accompanying drawings can be obtained based on these drawings without creative efforts.

[0029] Figure 1 It is a flowchart of the big data precise identification method for traditional Chinese medicine acupuncture of the present invention;

[0030] Figure 2 It is a schematic diagram of the connection of the system architecture modules of the present invention;

[0031] Figure 3 It is a schematic diagram of the construction and deviation identification of the personalized meridian baseline model of the present invention;

[0032] Figure 4 It is an architecture diagram of the acupuncture effect prediction model and reinforcement learning optimization of the present invention;

[0033] Figure 5 It is a schematic diagram of the interface of the remote monitoring platform of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0034] To make the purpose, technical solutions, and advantages of the present application clearer, the following will further describe the embodiments of the present application in detail in conjunction with the accompanying drawings.

[0035] Aiming at the problems existing in traditional Chinese medicine acupuncture, such as insufficient accuracy of acupoint positioning affected by individual differences, lack of real-time safety monitoring and dynamic parameter adjustment during acupuncture, curative effect evaluation relying on subjective experience, and difficulty in cross-institutional data sharing.

[0036] This solution proposes a precise identification method for big data of the human tissue structure for traditional Chinese medicine acupuncture.

[0037] Through multi-source data fusion technology, integrate medical images, real-time physiological signals, and traditional Chinese medicine constitution data to construct a multi-dimensional human tissue structure database;

[0038] Use a knowledge graph to establish a dynamic visual digital meridian model and label pathological features, generate a personalized meridian baseline model with the help of K-means clustering, and combine spatio-temporal feature extraction to identify individual acupoint deviations and monitor the changes of acupuncture tissues in real time;

[0039] Build an acupuncture effect prediction model based on deep learning and reinforcement learning, quantify the correlation between acupoint combinations and curative effects, and optimize stimulation parameters. At the same time, set a safety threshold to achieve risk warning;

[0040] By embedding a GPS module and blockchain technology, realize real-time positioning of the treatment location, remote monitoring, and full-process data storage, sharing and verification.

[0041] This solution forms a closed loop of "precise modeling - real-time feedback - intelligent optimization - safe evidence storage" through the deep integration of data-driven and intelligent algorithms, improves the acupoint positioning accuracy by more than 30%, shortens the risk warning response time to within 200ms, and the curative effect prediction accuracy reaches 85%, providing systematic technical support for the standardization and intelligence of acupuncture treatment, and significantly improving clinical safety and effectiveness.

[0042] Example 1, as Figure 1 shown, gives an illustrative description of the precise identification method for big data of the human tissue structure for traditional Chinese medicine acupuncture provided by the embodiments of this application.

[0043] S100: Based on medical image data, real-time physiological signal data collected by sensors, and traditional Chinese medicine constitution classification data, construct human tissue structure data, including anatomical data, physiological data, acupoint positions, and meridian paths. Use a 3T MRI device to obtain human tomographic images, and perform DICOM format parsing on CT data to obtain medical image data. Collect real-time physiological signals by deploying multi-channel physiological sensors, develop a standardized constitution differentiation questionnaire, and use the analytic hierarchy process (AHP) to construct a body mass quantification model to collect traditional Chinese medicine constitution classification data;

[0044] S200: Based on knowledge graph technology, associate acupoints, meridians, and pathological features, and establish a dynamic visual digital human meridian model. The digital human meridian model contains three major categories of entities and seven types of relationships, and label abnormal qi and blood stasis areas;

[0045] S300: Extract spatio-temporal features from medical image data and real-time physiological signal data. Based on the K-means clustering algorithm (using the elbow method to determine the optimal number of clusters K), generate a personalized meridian baseline model, identify acupoint deviations of different individuals, extract the three-dimensional coordinates of acupoints from MRI / CT images (accuracy ±0.3 mm), establish a deep learning-based acupoint localization model, combine with the digital model of the human meridian, identify acupoint deviations of different individuals, monitor tissue changes during acupuncture in real time, and perform dynamic adjustment in combination with historical data;

[0046] S400: Build an acupuncture effect prediction model. Based on historical treatment data and real-time monitoring data, predict the acupuncture effect. Use a deep learning network to extract tissue features of the acupuncture area, generate an acupuncture effect characterization vector, combine with the knowledge representation space to quantify the correlation between acupoint combinations and curative effects, and use a reinforcement learning model to optimize acupuncture stimulation parameters. When the acupuncture depth or force exceeds the safe range, issue a risk warning;

[0047] S500: Embed a GPS module in the intelligent acupuncture device, use a Beidou + GPS dual-mode positioning module to real-time locate the acupuncture treatment position, build an electronic map based on MapBox, real-time display the distribution of acupuncture devices, display the distribution and status of acupuncture devices through a remote monitoring platform. When abnormalities occur in the acupuncture area, highlight and retrieve relevant data, use a cumulative sum control chart (CUSUM) to monitor physiological signal changes, and assist in remote diagnosis;

[0048] S600: Build an acupuncture efficacy evaluation index system model. Use blockchain technology to record treatment process data and share analysis results among multiple institutions. When the monitored efficacy deviates from the expectation, trace back the operation records and environmental parameters, generate an abnormal root cause analysis report and feedback it to step S300 to optimize the acupuncture path.

[0049] Such as Figure 2 For step S100, the step S100 is to build human tissue structure data, including:

[0050] S110:

[0051] S1101: Medical image data acquisition:

[0052] Use a 3T MRI device to obtain human tomographic images (slice thickness 1 mm, resolution 512×512), covering common acupuncture sites such as the head, neck, trunk, and limbs, including T1-weighted images (showing anatomical structures), T2-weighted images (showing soft tissue features), and MR angiography (showing blood flow distribution);

[0053] Parse the CT data in DICOM format, extract the bone coordinates (such as the spinous process and the rib orientation), the thickness of the muscle fascia layer (segmented by the Hounsfield value threshold, and the Hounsfield value range of muscle tissue is [-20, 70]), and the distribution of the fat layer (the Hounsfield value range of adipose tissue is [-190, -30]). The tissue density quantization formula is

[0054]

[0055] where H i is the voxel Hounsfield value, and n is the number of voxels in the region of interest;

[0056] S1102: Deploy multi-channel physiological sensors:

[0057] The bioelectric sensor (sampling rate 1kHz) collects electromyogram (EMG) and skin conductance (SC, reflecting sympathetic nerve activity);

[0058] The pressure sensor (accuracy 0.1N) monitors the tissue pressure change during acupuncture;

[0059] The infrared sensor (resolution 0.1°C) collects the local body surface temperature;

[0060] The pulse oximeter (sampling rate 50Hz) obtains the heart rate (HR) and blood oxygen saturation (SpO2);

[0061] Signal preprocessing: Use a 50Hz notch filter to remove power frequency interference, smooth the electromyogram signal through a Savitzky-Golay filter (window length 5s), and extract the time-domain features (mean, standard deviation) and frequency-domain features (power spectral density) of the signal based on wavelet transform (db4 wavelet, decomposition level 3);

[0062] S1103: Develop a standardized physical constitution differentiation questionnaire, including 9 physical constitutions in the national standard of "Classification and Determination of Traditional Chinese Medicine Physical Constitution" (such as qi deficiency, blood stasis, phlegm-dampness, etc.), and set 10-15 differentiation items for each physical constitution dimension (such as the blood stasis constitution includes "purple and dull tongue" and "coarse and swollen sublingual collaterals");

[0063] Use the analytic hierarchy process (AHP) to construct a physical constitution quantification model:

[0064] Establish a judgment matrix to calculate the weights of each item. For example, the weights of "tongue image", "pulse condition", and "symptoms" are 0.4, 0.3, and 0.3 respectively;

[0065] Physical constitution score formula: where ω i is the item weight, and x i is the item score, with a range of 0-5 points.

[0066] S120: Based on the human anatomical coordinate system, map the MRI / CT image coordinates (Cartesian coordinate system, with the origin at the glabella), the surface acupoint positioning (converting the bone measurement method into actual distances, such as "12 cun from the cubital crease to the wrist crease" converted at 1 cun = 3.3 cm), and the sensor position (with the sternal angle as the positioning origin) to the world coordinate system (accuracy ±0.5 mm).

[0067] S1201: Perform Z-score standardization on numerical data (such as tissue density, EMG amplitude):

[0068]

[0069] where μ is the feature mean and σ is the standard deviation;

[0070] Perform one-hot encoding on categorical data (such as constitution type, pathological features) to generate a feature vector with a dimension of 1×1024 (including 89 basic features such as anatomy, physiology, and constitution);

[0071] S1202: Use a relational database (MySQL) to store three types of data:

[0072] Anatomy table: Includes the 3D coordinates of organs, the depth of tissue layers (such as the skin layer is 0.2 - 0.5 cm, and the muscle layer is 1 - 3 cm), and the position of the bone barrier (such as the safe acupuncture depth in the subclavian fossa is ≤2 cm);

[0073] Physiology table: Stores the time series of real-time physiological signals (timestamp accuracy 1 ms) and acupoint sensitivity parameters (such as the pain threshold of Hegu acupoint is 0.8 - 1.2 N);

[0074] Constitution table: Records the constitution score and key indicators for syndrome differentiation (such as the baseline blood viscosity of patients with blood stasis constitution is 15% - 20% higher than normal).

[0075] S130:

[0076] S1301: Use the IsolationForest algorithm to identify sensor outliers (such as a sudden increase in the pressure sensor reading exceeding 5 N and lasting for 100 ms, which is determined as a tissue damage warning signal), and perform data cleaning in combination with expert rules (such as automatically re-acquiring images with an MRI slice thickness deviation > 2 mm);

[0077] S1302: Establish a data version management system. When the patient receives a new acupuncture treatment, automatically synchronize the latest physiological signals (such as a decrease in the EMG signal amplitude by more than 30% after treatment is marked as an effective stimulus) and image review data (such as the volume reduction rate of the lesion area) to the database to ensure the timeliness of the model input data.

[0078] The S200 is used for knowledge graph construction and digital model establishment. The multi-dimensional data output by the S100 serves as the basic entity data for the knowledge graph construction of the S200. Specifically:

[0079] S210:

[0080] S2101: Define three major categories of entities:

[0081] Anatomical entities: acupoints (361 meridian acupoints + Ashi points), meridians (14 main meridians + eight extra meridians), tissue layers (skin, fascia, blood vessels, nerves);

[0082] Pathological entities: types of qi and blood stasis (qi stagnation, blood stasis, phlegm-dampness), disease syndromes (such as wind-cold-dampness arthralgia, hyperactivity of liver yang), abnormal physiological indicators (such as blood flow velocity < 0.5 m / s marked as blood stasis);

[0083] Therapeutic entities: acupuncture parameters (depth, frequency, lifting and thrusting amplitude), efficacy indicators (VAS score, hemorheology parameters);

[0084] S2101: Construct seven types of relationships:

[0085] Acupoint - Meridian: attribution relationship (such as "Hegu acupoint belongs to the Large Intestine Meridian of Hand-Yangming");

[0086] Meridian - Pathology: conduction relationship (such as "blockage of the Spleen Meridian of Foot-Taiyin leads to internal retention of water-dampness");

[0087] Tissue - Safety: threshold relationship (such as "the safe threshold of acupuncture depth in the lung area = 2.5 cm + 0.3 × the thickness of the fat layer");

[0088] Acupoint - Efficacy: association relationship (through historical data statistics, the combination of "Neiguan acupoint + Zusanli acupoint" has a 40% improvement rate in the efficacy of treating stomach fullness and discomfort).

[0089] S220:

[0090] S2201: Parse classical books such as "Compendium of Acupuncture and Moxibustion" and "Diagnostics of Traditional Chinese Medicine" through natural language processing (NLP), and use named entity recognition (NER) technology to extract acupoint aliases (such as "Quchi acupoint is also known as Guichen") and descriptions of meridian courses (such as "the Heart Meridian of Hand-Shaoyin starts from Jiquan and ends at Shaochong");

[0091] Extract "acupoint - disease" association pairs from electronic medical records (such as "the effective rate of Sanyinjiao acupoint in treating dysmenorrhea is 85%"), and combine the physiological data of the S100 to annotate pathological features (such as the area with abnormal electromyogram signals corresponding to the blocked sites of meridians);

[0092] S2202: Using the Neo4j graph database, node attributes include coordinates (e.g., 3D coordinates of Hegu acupoint: X=10cm, Y=8cm, Z=1.5cm), anatomical layer (middle of the muscle layer), and sensitive nerve (superficial branch of the radial nerve); edge attributes record relationship weights (e.g., "Hegu-headache" association weight = 0.78, OR value calculated based on logistic regression).

[0093] S230:

[0094] S2301: Develop a visualization engine based on VTK (Visualization Toolkit) to transparently overlay meridian pathways (e.g., the Governor Vessel along the midline of the spine) with anatomical structures (vertebral bodies, blood vessels), supporting zooming (with an accuracy of 0.1mm), rotation, and layered viewing (e.g., displaying the path of acupuncture points in the fascia layer separately);

[0095] S2302: Identify abnormal areas using image analysis algorithms (e.g., a U-Net-based semantic segmentation model with a Dice coefficient ≥ 0.92):

[0096] Blood stasis syndrome: MRI shows local vascular stenosis (diameter <2mm) and weakened blood flow signals, which is marked as "Qi and blood stagnation area of the hand Taiyin lung meridian" on the meridian model;

[0097] Phlegm-dampness syndrome: CT shows fat layer thickness > 2.5 cm (average value of the same site + 1.5σ), and the corresponding acupoint depth safety threshold is automatically adjusted (e.g., Fenglong acupoint depth = standard value - 0.5 cm);

[0098] S2303: When S300 detects abnormal real-time physiological signals (such as a sudden increase in the electromyographic signal >20μV for 200ms during acupuncture), it automatically marks the "nerve stimulation risk point" in the map and associates it with the safety depth threshold adjustment rules of the adjacent acupoints (such as reducing the needle insertion depth by 0.3cm).

[0099] The knowledge graph of step S200 provides standard acupoint coordinates and meridian direction references for acupoint deviation identification in step S300, and provides prior knowledge of "acupoint-pathology-efficacy" for the efficacy prediction model in step S400, forming an intermediate layer support of "knowledge modeling-visualization-dynamic labeling".

[0100] Specifically, the S300 is used for personalized meridian baseline model construction and real-time monitoring, including:

[0101] S310:

[0102] S3101: Extract 3D coordinates of acupoints from MRI / CT images (accuracy ±0.3mm):

[0103] (x acupoint ,yacupoint , z acupoint ) = f segmentation (MRI_volume)

[0104] where f segmentation is a deep learning-based acupoint location model (ResNet+U-Net architecture, location error ≤ 0.5 mm);

[0105] Calculate tissue layer parameters: skin layer depth (epidermis + dermis thickness, area with MRIT2-weighted image gray value > 150), muscle layer thickness (distance between fascia layers, CT value range 20 - 70), bone barrier distance (e.g., depth warning is triggered when the minimum distance from an acupoint on the nape to the atlas is < 1 cm);

[0106] S3102: Process physiological signals (such as electromyogram, heart rate) with a sliding window (window length 5 s, step size 1 s) and extract:

[0107] Time domain features: mean (μ), standard deviation (σ), peak-to-peak value (PP);

[0108] Frequency domain features: main peak frequency of power spectral density (PSD) (e.g., the power ratio of the 50 - 100 Hz frequency band in the electromyogram signal reflects muscle tension);

[0109] Time-frequency features: Generate a time-frequency diagram through short-time Fourier transform (STFT), and use CNN to extract time-frequency domain texture features (e.g., when acupuncture is effective, the energy in the 10 - 30 Hz frequency band is significantly enhanced).

[0110] S320:

[0111] S3201: Input feature matrix where n is the number of samples, d = 89 dimensions, including anatomical, physiological, and constitutional features, and the elbow method is used to determine the optimal number of clusters K (usually K = 5 - 8, corresponding to the baseline models of different constitutional types);

[0112] Distance metric formula: Euclidean distance combined with Mahalanobis distance to eliminate the influence of feature dimensions:

[0113]

[0114] where S is the feature covariance matrix;

[0115] S3202: Generate a mean vector μ for each cluster k as the baseline feature, including:

[0116] Standard acupoint coordinates: (Considering constitutional differences, e.g., the baseline value of acupoint depth for phlegm-dampness constitution is 0.5 cm deeper than that of the normal group);

[0117] Physiological signal baseline range: The heart rate baseline is Values outside the range are marked as abnormal.

[0118] S330:

[0119] S3301: Euclidean distance deviation between the actual acupoint coordinates and the baseline coordinates:

[0120]

[0121] When Δd > 1 cm, it is determined as a significant deviation, triggering the acupoint repositioning process (automatically retrieving and re-segmenting MRI images);

[0122] S3302: Deploy a multi-modal sensor fusion algorithm (Extended Kalman Filter EKF) to estimate the tissue state in real time

[0123]

[0124] where u t is the acupuncture depth control amount, z t is the sensor measurement value (pressure, impedance), ω t , v t are the process and measurement noises;

[0125] Dynamic adjustment strategy: When a sudden drop in tissue impedance is detected (> 30% of the baseline value, indicating entry into the vascular layer), automatically trigger a 0.2 cm retraction of the needle and reduce the stimulation intensity (frequency drops from 10 Hz to 2 Hz).

[0126] The baseline model of S300 provides individual feature inputs for the acupuncture effect prediction of S400, and the real-time monitoring data is fed back to S200 to update the pathological annotation, forming a closed-loop control of "feature extraction - model generation - deviation identification - dynamic adjustment".

[0127] The step S400 is for acupuncture effect prediction and parameter optimization, including:

[0128] S410:

[0129] S4101: The input data includes:

[0130] Historical treatment data: treatment efficacy level (1 - 5 points), acupuncture parameters (depth, frequency, duration), patient baseline characteristics (age, physical fitness score);

[0131] Real-time monitoring data: changes in CT values in the acupuncture area (reflecting tissue edema degree), changes in the power spectrum of electromyography signals (reflecting nerve activation level);

[0132] Adopt dual-channel fusion features:

[0133] Image channel: 3D-CNN extracts MRI tomography features (receptive field 3×3×3, output dimension 512);

[0134] Signal channel: LSTM processes physiological signal time series (128 units in the hidden layer, capturing signal fluctuations 5 minutes before and after acupuncture);

[0135] S4102: The fused features generate a 128-dimensional treatment effect representation vector v through a fully connected layer effect , calculation formula: v effect =σ(W fusion [v image ; v signal +b fusion ), where [;] represents vector concatenation, and σ is the ReLU activation function.

[0136] S420:

[0137] S4201: Represent the acupoint combination using One-Hot encoding (e.g., "Hegu + Neiguan" is encoded as a 1×362 vector, with the corresponding acupoint position being 1 and the rest being 0), and generate acupoint semantic vectors through knowledge graph embedding technology (TransE)

[0138] S4202: Establish an "acupoint combination - treatment effect" correlation matrix (C is the number of acupoint combinations, E is the treatment effect dimension), calculate the correlation degree r through historical data training ce :

[0139]

[0140] Among them, count(c,e) is the number of occurrences of combination c in treatment effect e

[0141] Calculate the matching degree between the real-time acupoint combination and the target treatment effect by combining cosine similarity:

[0142]

[0143] S430:

[0144] S4301: State S t : Current acupuncture parameters (depth d_t, frequency f_t, force F_t), physiological signal features (EMG amplitude EMG_t, tissue impedance Z_t), acupoint deviation Δd_t;

[0145] Action A t : Parameter adjustment amount (Δd ∈ {-0.5, -0.3, 0, +0.3, +0.5}mm, Δf ∈ {-5, 0, +5}Hz, ΔF ∈ {-1, 0, +1}N);

[0146] Reward R t : Reward for improved efficacy (e.g., +10 for each 1-point decrease in VAS score) - penalty for risk ( - 5 for each 0.1 cm of depth exceeding the limit);

[0147] S4302: Use the Experience Replay mechanism to store transfer samples (S t , A t , R t , S t+1 ), update the target network every 100 steps, and the loss function is:

[0148] L(θ) = E (s,a,r,s')～D [(r + γmax a' Qs', a'; θ - ) - Qs, a; θ)) 2

[0149] where γ = 0.99 is the discount factor, and θ - are the target network parameters;

[0150] S4303: Anatomical threshold: The safe upper limit of the acupuncture depth in the pulmonary body surface projection area, d safe = 2.5 cm - 0.2 × age (years) / 50;

[0151] Physiological threshold: The mutation rate of electromyogram signal is determined as the risk of nerve injury;

[0152] Early warning response: When the real-time depth d t ≥ d safe × 1.1, trigger a level-three early warning (acoustic and light alarm → automatically stop the needle insertion → send a remote help signal).

[0153] The prediction result of S400 guides S300 to adjust the acupuncture path, and the risk early warning data is fed back to the S500 remote monitoring platform to form an intelligent decision-making layer of "prediction - optimization - control".

[0154] The step S500 is used for building a remote positioning and monitoring platform, including:

[0155] S510:

[0156] S5101: Adopt a Beidou + GPS dual-mode positioning module (positioning accuracy ≤ 5 m), collect the coordinates of the treatment position (longitude λ, latitude φ, altitude h) in real time, and obtain the device attitude angles (pitch angle, roll angle, yaw angle) in combination with an electronic compass to ensure that the acupuncture direction is perpendicular to the anatomical layers;

[0157] ​S5102: Support 4G / 5G (Sub-6GHz band, upload rate ≥ 10Mbps) and Bluetooth 5.0 (for short-range device interconnection). The data transmission protocol uses HL7 FHIR (Healthcare Information Exchange Standard) to ensure compatibility with the hospital information system (HIS).

[0158] S520:

[0159] S5201: Build an electronic map based on MapBox to display the distribution of acupuncture devices in real time (icon colors distinguish status: green = normal, yellow = warning, red = failure). Click on the device icon to view detailed information:

[0160] Real-time data: Acupuncture depth (accuracy 0.1mm), current acupoint name, physiological signal curve (trend in the past 10 minutes);

[0161] Historical records: Parameter logs of the last 5 treatments, efficacy evaluation reports, details of abnormal events;

[0162] S5202: Abnormal detection algorithm: Use the Cumulative Sum Control Chart (CUSUM) to monitor changes in physiological signals. When the statistic S t = max(0, S t-1 + x t - μ0) / σ0) exceeds the threshold H = 5, it is determined as an abnormal event;

[0163] Remote diagnosis support: Automatically retrieve real-time image slices (last MRI / CT data) of abnormal devices, S300 baseline model parameters, and S400 prediction results, and form a diagnostic package containing "location - status - data" to be pushed to physicians;

[0164] S530:

[0165] S5301: Encrypt the transmitted data using the AES-256 encryption algorithm, and the key is dynamically negotiated through the Elliptic Curve Diffie-Hellman (ECDH) protocol to ensure that the data cannot be tampered with during transmission;

[0166] S5302: Role-based access control (RBAC) system. Physician accounts are divided into three levels of permissions:

[0167] Junior: View the device status of the patients they belong to;

[0168] Intermediate: Adjust acupuncture parameters and receive warning information;

[0169] Senior: Manage the device cluster and export statistical reports.

[0170] The positioning data in step S500 above provides treatment environment parameters for the efficacy evaluation in S600 (such as differences in equipment usage in different institutions), and the abnormal event records serve as important inputs for the root cause analysis in S600, forming a remote interaction layer of "hardware integration - platform display - security control".

[0171] Step S600 is used for efficacy evaluation and blockchain evidence storage, including:

[0172] S610:

[0173] S6101: Objective indicators (weight 60%):

[0174] Imaging indicators: Lesion volume reduction rate

[0175] Physiological indicators: Changes in hemorheological parameters (decrease rate of whole blood viscosity, change in erythrocyte aggregation index);

[0176] Safety indicators: Incidence rate of acupuncture abnormal events (such as number of hematoma and nerve stimulation events / number of treatments);

[0177] Subjective indicators (weight 40%):

[0178] Patient report: VAS pain score, TCM symptom integral (score of symptoms such as "fatigue" and "insomnia" from 0 to 5);

[0179] Physician evaluation: Level of needling sensation (weak / medium / strong, corresponding to 1 - 3 points);

[0180] S6102: Construct a judgment matrix to calculate the index weights. For example, the importance of "imaging indicators" relative to "physiological indicators" in objective indicators is 3 (scale 1 - 9), and finally obtain the index weight vector W = [ω1, ω2, …, ω n , comprehensive scoring formula: (s i is the standardized score of each index);

[0181] S620:

[0182] S6201: Use Hyperledger Fabric to build a consortium chain. The nodes include medical institutions, scientific research institutions, and regulatory institutions. Each block contains:

[0183] Treatment data: Patient ID (protected by hash processing for privacy), acupuncture timestamp (accuracy 1ms), output data of each step from S100 - S500 (baseline model parameters, prediction results, warning records);

[0184] Consensus mechanism: Adopt the PBFT (Practical Byzantine Fault Tolerance) algorithm to ensure the consistency of data on the chain, and the block generation time ≤ 10s;

[0185] S6202: Define the evidence storage contract RecordContract:

[0186] function recordTreatmentData(bytes32 patientHash, bytes dataHash, uint timestamp) public {

[0187] treatments[patientHash][timestamp] = dataHash;

[0188] emit DataRecorded(patientHash, timestamp);

[0189] }

[0190] Data sharing contract: After being authorized by the patient, it allows cross - institutional query of the desensitized treatment efficacy data (such as deleting the name and ID number, and retaining statistical features such as age and constitution type);

[0191] S630:

[0192] S6301: Build a root cause analysis model using the decision tree algorithm. The input variables include:

[0193] Operation records: The number of times of acupuncture depth adjustment, the history of parameter over - limit (such as being marked as "improper operation" when the depth exceeds the limit more than 3 times);

[0194] Environmental parameters: Room temperature during treatment (a room temperature below 18°C may cause an increase in muscle tension), the patient's emotional stress value (evaluated through heart rate variability HRV, RMSSD < 20ms indicates a high - stress state);

[0195] Root cause determination rule: If "acupoint deviation > 1cm" and "the re - positioning process is not triggered", then the root cause is "the baseline model has not been updated";

[0196] S6302: When the treatment efficacy deviates from the expectation (S effect <predicted value × 0.8), automatically trigger the back - tracking process:

[0197] Obtain the complete operation log from the blockchain;

[0198] Locate the problem through the root cause model (such as "the number of clusters K in the baseline model in step S300 is incorrect, resulting in acupoint depth deviation");

[0199] Generate optimization instructions (such as "adjust K = 5 to K = 6 and recalculate the baseline coordinates") and feedback them to S300 to update the personalized model;

[0200] The efficacy evaluation results of step S600 above are used as the training labels for the S400 prediction model, and the blockchain-certified data provides real-world evidence for the S200 knowledge graph, forming a quality control layer of "evaluation - certification - traceability - optimization".

[0201] Experimental Example 1:

[0202] Experimental Purpose

[0203] Verify the acupoint location accuracy of the personalized baseline model generated by K-means clustering in step S300 for patients with different constitutions, and the dynamic adjustment effect on the acupuncture depth.

[0204] Experimental Subjects

[0205] Select 20 patients with blood stasis constitution (BMI 24 - 28, fat layer thickness 1.5 - 2.5 cm), and randomly divide them into two groups:

[0206] Experimental Group (10 people): Adopt the data fusion and baseline model of this scheme S100 - S300;

[0207] Control Group (10 people): Use the traditional bone measurement method for location and manually adjust the acupuncture depth.

[0208] Step 1: Experimental Group: Obtain the three-dimensional coordinates of the Hegu acupoint through MRI, combine the electromyogram signal (average amplitude 18 μV) and the constitution score (blood stasis constitution score 82 points), and generate a baseline model through K-means clustering (baseline value of acupoint depth 1.8 cm, standard deviation 0.2 cm);

[0209] Control Group: Locate the Hegu acupoint by the bone measurement method (standard depth 1.5 cm) without individual difference correction.

[0210] Step 2: Experimental Group: Real-time monitor the tissue pressure signal. When the depth reaches 1.9 cm (exceeding the baseline + 0.1 cm), the system automatically prompts "Approaching the bottom of the fat layer, it is recommended to reduce the needle insertion speed".

[0211] Control Group: The doctor controls the depth based on experience, with an average depth of 1.6 ± 0.3 cm (in some patients, it reaches 2.0 cm, exceeding the safety threshold of 1.8 cm).

[0212] Step 3: Location Accuracy: Calculate the acupoint coordinate deviation through MRI reexamination;

[0213] Safety Incidents: Record the number of times of depth exceeding the limit (> 1.8 cm);

[0214] Qi Sensation: Subjective score of the patient (1 - 5 points, 5 points being the strongest).

[0215] Experimental Data Record Table:

[0216]

[0217] Experimental conclusion: The personalized baseline model significantly improves the accuracy of acupoint location, reduces the risk of depth exceeding the limit, enhances the qi sensation of acupuncture, and verifies the effectiveness of the individual difference quantification and dynamic adjustment mechanism in step S300.

[0218] Experimental example 2:

[0219] Experimental purpose

[0220] Verify the influence of the acupuncture parameter optimization algorithm combining deep learning and reinforcement learning in step S400 on the accuracy rate of curative effect prediction and safety events.

[0221] Experimental subjects

[0222] Thirty patients with chronic low back pain (VAS score ≥ 6 points) were selected and divided into three groups:

[0223] Group A (10 people): This protocol (deep learning + reinforcement learning);

[0224] Group B (10 people): Only deep learning prediction (without parameter optimization);

[0225] Group C (10 people): Traditional empirical acupoint selection (without model assistance).

[0226] Step 1:

[0227] Group A: Extract MRI image features through 3D-CNN and optimize the acupuncture frequency in combination with the DQN algorithm (initial 2 Hz, target VAS score reduction ≥ 3 points);

[0228] Group B: Only use deep learning to predict the curative effect, and the parameters are manually adjusted by the physician;

[0229] Group C: Adopt the fixed acupoint combination of "Shenshu + Dachangshu", frequency 10 Hz.

[0230] Step 2:

[0231] Group A: When the curative effect compliance rate of the current parameters (depth 2.0 cm, frequency 8 Hz) predicted by the system is < 60%, automatically adjust to a depth of 1.8 cm and a frequency of 5 Hz;

[0232] Group B / Group C: Rely on the physician's experience for adjustment, and adjust the parameters 3 - 5 times per course of treatment on average.

[0233] Step 3:

[0234] Accuracy rate of curative effect prediction: The degree of coincidence between the actual curative effect and the model prediction;

[0235] Safety events: Record the number of abnormal nerve stimulations (sudden increase in EMG signal > 20 μV);

[0236] Onset time: The time when the VAS score first decreases by ≥2 points.

[0237] Experimental data record form

[0238]

[0239] Experimental conclusion

[0240] The parameter optimization model combining deep learning and reinforcement learning significantly improves the accuracy of efficacy prediction, reduces safety risks, shortens the onset time, and verifies the effectiveness of the "prediction - optimization - control" closed-loop in step S400.

[0241] Experimental example 3

[0242] Experimental purpose

[0243] Verify the application effect of remote monitoring and blockchain evidence storage in multi-institutional data sharing and efficacy evaluation in steps S500 - S600.

[0244] Experimental environment

[0245] Institution A: A top - three traditional Chinese medicine hospital (deploying 10 intelligent acupuncture devices);

[0246] Institution B: Community health service center (deploying 5 intelligent acupuncture devices);

[0247] Data platform: An alliance chain based on Hyperledger Fabric, containing 10 nodes.

[0248] Step 1:

[0249] Experimental group: Treatment data is uploaded to the blockchain in real - time (including the output of S100 - S500, with patient ID hashed), and the cross - institutional query response time ≤2s;

[0250] Control group: Shared through traditional Excel spreadsheets, with inconsistent data formats and a query response time of 20 - 30 minutes.

[0251] Step 2:

[0252] Experimental group: When the depth of a certain device in Institution B reaches 3.0 cm (the lung safety threshold is 2.5 cm), the platform triggers an alarm within 50 ms and automatically retrieves similar cases from Institution A (fat layer thickness 2.0 cm, safety depth 2.3 cm);

[0253] Control group: Manual phone communication, with an abnormal response time of 5 - 10 minutes.

[0254] Step 3:

[0255] Experimental group: Calculate the comprehensive efficacy score based on blockchain data (weight of objective indicators is 60%), and the evaluation takes 1 minute;

[0256] Control group: Manually summarize data, the evaluation takes 30 minutes, and the consistency error is 15% - 20%.

[0257] Experimental data record form

[0258]

[0259] Experimental conclusion

[0260] The remote monitoring platform and blockchain technology have realized the real-time sharing and secure storage of cross-institutional data, greatly improving the efficiency of exception handling and the objectivity of efficacy evaluation, and verifying the value of steps S500 - S600 in standardized diagnosis and treatment.

[0261] Example 2: A precise identification system for a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, used to implement a method for precise identification of a large amount of human tissue structure big data for traditional Chinese medicine acupuncture, including the following modules;

[0262] Data fusion module: Collect multi-source data such as medical images, physiological signals, and traditional Chinese medicine constitutions, and construct a structured database containing anatomy, physiology, acupoints, and meridians after preprocessing to achieve data standardization and individual feature quantification;

[0263] Knowledge graph module: Associate acupoints, meridians with pathological features through a knowledge graph, construct a dynamic and visual digital model of the human meridian system and mark abnormal qi and blood stasis areas, make explicit the "acupoint - disease" association, and the Dice coefficient of MRI pathological annotation is ≥0.92.

[0264] Baseline model construction module: Extract spatio-temporal features, generate a personalized meridian baseline model based on K-means, identify acupoint deviations and monitor tissue changes during acupuncture in real time, the acupoint positioning accuracy drops from 1.8 mm to 0.6 mm, and real-time feedback on tissue changes is provided;

[0265] Effect prediction module: Integrate multi-modal data to construct a prediction model, extract features through deep learning, optimize acupuncture parameters through reinforcement learning, set a safety threshold warning, the accuracy of efficacy prediction reaches 85%, the automatic adjustment of parameters is reduced by 50%, the onset time is shortened by 54.3%, and the incidence of safety events is reduced by 83.3%;

[0266] Remote monitoring module: An intelligent device is embedded with a positioning module, and the device distribution and status are displayed in real time through a remote platform. When abnormal, it is highlighted and data is retrieved to assist in diagnosis. The response time for querying the device status is ≤2 s, supporting cross-institutional device management and regional collaboration;

[0267] Therapeutic effect evaluation module: Build a multi-dimensional evaluation system, store treatment data on the blockchain, trace back the root cause of anomalies and feedback to optimize the treatment path. The consistency of therapeutic effect evaluation has been improved from 75% to 92%. The blockchain ensures the immutability of data.

[0268] Embodiment 3: A computer device, the computer device includes: a processor and a memory. At least one instruction, at least one program, a code set or an instruction set is stored in the memory. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement a method for accurately identifying big data of the human body tissue structure for traditional Chinese medicine acupuncture.

[0269] Embodiment 4: A computer-readable storage medium, characterized in that at least one instruction, at least one program, a code set or an instruction set is stored in the storage medium. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by a processor to implement a method for accurately identifying big data of the human body tissue structure for traditional Chinese medicine acupuncture.

[0270] Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for accurately identifying big data of a large number of human tissue structures for traditional Chinese medicine acupuncture, characterized in that Including: S100: Construct human body tissue structure data based on medical image data, real-time physiological signal data collected by sensors, and traditional Chinese medicine constitution classification data, including anatomical data, physiological data, acupoint positions, and meridian paths; S200: Associate acupoints, meridians with pathological features based on knowledge graph technology, establish a dynamic visual digital model of human meridians, and mark abnormal qi and blood stasis areas; S300: Extract spatio-temporal features from medical image data and real-time physiological signal data, generate a personalized meridian baseline model based on the K-means clustering algorithm, identify acupoint deviations of different individuals, combine with the digital model of human meridians to identify acupoint deviations of different individuals, monitor tissue changes during acupuncture in real time, and make dynamic adjustments in combination with historical data; S400: Construct an acupuncture effect prediction model, predict the acupuncture effect based on historical treatment data and real-time monitoring data, use a deep learning network to extract tissue features of the acupuncture area, generate an acupuncture effect characterization vector, combine with the knowledge representation space to quantify the correlation between acupoint combinations and curative effects, use a reinforcement learning model to optimize acupuncture stimulation parameters, and issue a risk warning when the acupuncture depth or strength exceeds the safe range; S500: Embed a GPS module in the intelligent acupuncture device to real-time locate the acupuncture treatment position, display the distribution and status of the acupuncture device through a remote monitoring platform, highlight and retrieve relevant data when abnormalities occur in the acupuncture area to assist remote diagnosis; S600: Construct an acupuncture efficacy evaluation index system model, store the treatment process data and multi-institutional shared analysis results through blockchain technology, when the monitored efficacy deviates from the expectation, trace back the operation records and environmental parameters, generate an abnormal root cause analysis report and feedback it to step S300 to optimize the acupuncture path.

2. The method for precise identification of big data of massive human tissue structures for traditional Chinese medicine acupuncture according to claim 1, characterized in that: The medical image data includes MRI tomographic images, CT data, and MR angiography data, where the slice thickness of the MRI images ≤ 1 mm. When analyzing the CT data, muscle tissue and adipose tissue are segmented by the Hounsfield value threshold, and the tissue density is quantified by the formula Quantify tissue density.

3. The method for accurately identifying big data of the human tissue structure for traditional Chinese medicine acupuncture according to claim 1, characterized in that: In the dynamic visualization digital model of the human meridian constructed by the knowledge graph technology, three types of entities are defined: anatomical entities, pathological entities, and treatment entities. Among them, anatomical entities include 361 acupoints and 14 main meridians. Pathological entities are used to label the qi and blood stasis areas through MRI semantic segmentation. Treatment entities are associated with the acupuncture depth safety threshold formula d safe = 2.5 cm + 0.3 × fat layer thickness.

4. The method for accurately identifying big data of a vast number of human tissue structures for traditional Chinese medicine acupuncture according to claim 1, wherein: When the K-means clustering algorithm generates a personalized meridian baseline model, the elbow method is used to determine the optimal number of clusters K, and the Mahalanobis distance formula is used to measure the sample differences, where S is the feature covariance matrix, and the generated baseline model includes the three-dimensional coordinates of acupoints and the baseline range of physiological signals.

5. The method for accurately identifying big data of a massive human tissue structure for traditional Chinese medicine acupuncture according to claim 1, wherein: The acupuncture effect prediction model fuses images and signal features through a dual-channel approach. In the image channel, 3D-CNN is used to extract MRI features, and in the signal channel, LSTM is used to process the physiological signal time series to generate a 128-dimensional efficacy characterization vector v effect = σ(W fusion [v image ; v signal + b fusion ).

6. The method for accurately identifying a large amount of big data of human tissue structures for traditional Chinese medicine acupuncture according to claim 1, wherein: The intelligent acupuncture device is embedded with a Beidou+GPS dual-mode positioning module and transmits data through a 4G / 5G communication module. The data format follows the HL7 FHIR standard; the remote monitoring platform builds a GIS visualization system based on MapBox, monitors abnormal events through cumulative sum and control charts, and the statistic S t = max(0, S t-1 +(x t - μ0) / σ0) triggers an alarm when it exceeds the threshold H = 5.

7. The method for accurately identifying a large amount of big data of human tissue structures for traditional Chinese medicine acupuncture according to claim 1, wherein: The blockchain technology adopts the Hyperledger Fabric consortium chain architecture, the nodes include medical institutions, scientific research institutions, and regulatory institutions, the consensus mechanism is PBFT, the stored evidence data includes the patient ID hash value, acupuncture timestamp, and the output data of each step, and the smart contract defines the rules for data storage and sharing.

8. A big data precise recognition system for massive human tissue structures in traditional Chinese medicine acupuncture, which is used to implement the method described in any one of claims 1-7, characterized in that, Including the following modules: Data fusion module: Collect multi-source data such as medical images, physiological signals, and traditional Chinese medicine constitutions, preprocess them, and construct a structured database containing anatomy, physiology, acupoints, and meridians to achieve data standardization and individual feature quantification; Knowledge graph module: Associate acupoints, meridians with pathological features through the knowledge graph, construct a dynamic visual digital model of human meridians and mark abnormal qi and blood stasis areas, make the "acupoint-disease" association explicit, and the Dice coefficient of MRI pathological annotation ≥ 0.

92. Baseline model construction module: Extract spatio-temporal features, generate a personalized meridian baseline model based on K-means, identify acupoint deviations and monitor tissue changes during acupuncture in real time, reduce the acupoint positioning accuracy from 1.8mm to 0.6mm, and give real-time feedback on tissue changes; Effect prediction module: Integrate multi-modal data to build a prediction model, extract features through deep learning, optimize acupuncture parameters through reinforcement learning, set a safety threshold for early warning. The accuracy of curative effect prediction reaches 85%, the automatic adjustment of parameters is reduced by 50%, the onset time is shortened by 54.3%, and the incidence of safety events is reduced by 83.3%; Remote monitoring module: An intelligent device is embedded with a positioning module, which can display the distribution and status of devices in real time through a remote platform. When abnormal, it will be highlighted and relevant data will be retrieved to assist in diagnosis. The response time for querying the device status is ≤2s, supporting cross-institutional device management and regional collaboration; Curative effect evaluation module: Build a multi-dimensional evaluation system, use blockchain to store treatment data, trace back the root cause of anomalies and feedback to optimize the treatment path. The consistency of curative effect evaluation is improved from 75% to 92%, and the blockchain ensures the immutability of data.

9. A computer device, characterized in that, The computer device includes: a processor and a memory. The memory stores at least one instruction, at least one program, a code set or an instruction set. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the method for accurately identifying big data of the human body tissue structure for traditional Chinese medicine acupuncture as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The storage medium stores at least one instruction, at least one program, a code set or an instruction set. The at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the method for accurately identifying big data of the human body tissue structure for traditional Chinese medicine acupuncture as described in any one of claims 1 to 7.

Images