Acupoint electric body surface array signal analysis method and system based on spatial distribution characteristics

Abstract

The invention relates to the field of traditional Chinese medicine acupuncture science, and discloses an acupoint electric body surface array signal analysis method and system based on spatial distribution characteristics, and the method comprises the steps: obtaining a multi-channel bio-electricity signal of an acupoint to be analyzed through employing a matrix array type electrode patch, and synchronously collecting external modal data; performing noise reduction on the bio-electricity signal, extracting a multi-dimensional feature set, fusing the multi-dimensional feature set with external modal data, and mapping the multi-dimensional feature set into a visual area map according to a spatial position and time; then space-time conjoint analysis is carried out, and a space-time dynamic feature set is extracted; inputting the feature set into a machine learning algorithm, predicting acupoint electrical activity space distribution change modes at different time points in a classified manner, and automatically identifying an abnormal mode; and finally, according to the prediction and identification result, evaluating the spatial distribution mode of the acupoint electrical activity, and outputting a comprehensive analysis report. Measurement interference factors can be effectively eliminated, quality control is enhanced, and the reliability and repeatability of analysis results are remarkably improved, so that the acupoint bioelectricity characteristics are reflected more accurately.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traditional Chinese acupuncture, more particularly to an acupoint electrodermal array signal analysis method and system based on spatial distribution characteristics. BACKGROUND

[0002] As an important means to explore the electrical characteristics of meridians and acupoints and the correlation of viscera, acupoint electrodermal signal analysis has experienced evolution from traditional single-point measurement to modern array and intelligent analysis. Traditional techniques mostly rely on single-electrode or double-electrode contact measurement, and conduct positioning analysis by detecting parameters such as acupoint local skin resistance and potential difference. However, due to interference factors such as electrode contact pressure, skin state and environmental temperature and humidity, the signal stability is poor, the repeatability is low, and it is difficult to capture the dynamic change characteristics of acupoint electrical signals. At the same time, traditional methods mostly use linear filtering or time / frequency domain statistical feature extraction, which cannot effectively analyze the nonlinear and non-stationary characteristics of meridian and acupoint electrical signals. Influenced by factors such as neural conduction, metabolic activity and body surface structure, the analysis based on a single parameter is easy to lose key physiological information.

[0003] In recent years, with the concept of acupoint dynamic characteristics being proposed and the emergence of detection methods and means of body surface biophysics, as a certain condition under which the size, position and local biological active substances change, acupoints show many body surface dynamic forms such as heat sensitivity, pain sensitivity, electrical sensitivity and force sensitivity. Among them, the electrical sensitivity research based on the electrical characteristics of meridians and acupoints finds the body surface reaction site related to diseases and interventions through the specific changes of electrical signals such as potential and resistance, which can reflect the individual's current physiological and pathological state in real time and sensitively.

[0004] However, the electrical sensitivity research based on the electrical characteristics of meridians and acupoints in the prior art still has the following shortcomings: Firstly, the prior art mostly relies on single-electrode or double-electrode contact measurement, and the data dimension obtained is limited, which is difficult to fully reflect the acupoint electrical activity. Moreover, when obtaining bioelectric signals and other related data, there is often a lack of synchronization mechanism, resulting in inaccurate time correspondence between data. Secondly, the noise reduction processing of bioelectric signals in the prior art may be simple, which is difficult to effectively remove noise interference and affect signal quality. When extracting acupoint electrical activity characteristics, only part of the basic parameters may be focused on, and the multi-dimensional characteristics of the signal are ignored. Thirdly, the prior art rarely deeply fuses the bioelectric signal characteristics with external modal data, which is difficult to fully understand the mutual relationship between acupoint electrical activity and the overall state of the body and the environment. Moreover, the analysis of acupoint electrical activity may be static or simple time domain analysis, which is difficult to capture its dynamic change rule. Fourth, existing technologies may lack effective algorithms and models for classifying and predicting spatial distribution patterns of acupoint electrical activity, resulting in low accuracy. When identifying abnormal spatial distribution patterns, they may rely on manual judgment or simple rules, which can easily lead to missed or false detections. Furthermore, the assessment of acupoint electrical activity may be somewhat one-sided. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, the present invention provides a method and system for analyzing hole electric field array signals based on spatial distribution characteristics, in order to solve the problems existing in the background art.

[0006] This invention provides the following technical solution: a method for analyzing hole electric field surface array signals based on spatial distribution characteristics, comprising: S1: Acquire multi-channel bioelectrical signals sensed by the matrix array electrode patches placed on the acupoints to be analyzed, and simultaneously acquire external modal data; S2: The multi-channel bioelectric signal is denoised to obtain a denoised multi-channel signal; S3: Process the denoised multi-channel signal to extract a multi-dimensional feature set characterizing the electrical activity of acupoints; S4: The multi-dimensional feature set is fused with the external modal data, and the fused features are mapped into a visualization area map corresponding to the matrix array electrode patch based on their spatial location and time information. S5: Perform spatial-temporal joint analysis on the visualized area map to extract a set of spatiotemporal dynamic features that quantitatively describe the dynamic pattern of acupoint electrical activity; S6: Using the spatiotemporal dynamic feature set as input, a machine learning algorithm is used to classify and predict the spatial distribution change patterns of acupoint electrical activity at different time points, and to automatically identify abnormal spatial distribution patterns that are significantly different from normal patterns. S7: Based on the prediction results and abnormal pattern recognition results, evaluate the spatial distribution pattern of acupoint electrical activity and output a comprehensive analysis report.

[0007] Preferably, S1 acquires multi-channel bioelectric signals of the acupoint to be analyzed through a matrix array electrode patch, wherein the matrix array electrode patch includes multiple silver chloride point electrodes with an adjacent spacing of 3-7 mm, and adjacent silver chloride point electrodes are respectively configured as acquisition electrodes and output electrodes. The acquired multi-channel bioelectrical signals can be obtained using the following acquisition modes, depending on the specific analytical purpose: Interference response acquisition mode: After intervention is performed on the preset distal acupoint of the acupoint to be analyzed, the bioelectrical signals on the acupoint to be analyzed are acquired in real time at equal time intervals; Resting state acquisition mode: continuously acquiring bioelectric signals on the acupoint to be analyzed in a resting state without external intervention; Disease-related acquisition mode: acquiring bioelectric signals placed on the acupoint to be analyzed corresponding to the specific disease of the user; In the acquisition of multi-channel bioelectric signals, external modal data is synchronously acquired, including electrocardiogram, electroencephalogram, acupoint pressure value, electromyogram, respiratory signal, skin temperature and environmental parameters.

[0008] Preferably, the specific operation of S2 for noise reduction processing of the multi-channel bioelectric signals includes: The multi-channel bioelectric signals are filtered, and the filtered signals are subjected to artifact removal processing, which includes identifying and removing physiological artifact components by using a blind source separation algorithm. Finally, the signals subjected to artifact removal processing are subjected to smoothing processing to obtain multi-channel signals with reduced noise.

[0009] Preferably, the S3 extracted multi-dimensional feature set representing acupoint electrical activity includes: The multi-channel signals with reduced noise are subjected to time-frequency analysis to extract time-frequency features representing acupoint electrical activity intensity and rhythm; The multi-channel signals with reduced noise are subjected to nonlinear analysis to extract nonlinear features representing acupoint electrical activity complexity and regularity; Based on the spatial correlation between each channel signal in the multi-channel signals with reduced noise, a functional network is constructed to extract spatial network features; The time-frequency features, nonlinear features and spatial network features are combined to form the multi-dimensional feature set.

[0010] Preferably, S4 performs spatiotemporal alignment of the multi-dimensional feature set with the external modal data, and generates a unified fusion feature vector by using a fusion algorithm; According to the known physical coordinates of each electrode in the matrix array electrode patch, each fusion feature vector is mapped to the corresponding spatial coordinate point to form a feature space distribution matrix; The feature space distribution matrix is subjected to interpolation and smoothing processing, and a visualized area map corresponding to the layout of the matrix array electrode patch is generated according to a preset rendering rule.

[0011] Preferably, S5 stacks a plurality of visualized area maps arranged in time sequence to construct a spatiotemporal data cube; By analyzing the spatiotemporal data cube, spatial distribution evolution features, spatiotemporal coupling features and spatiotemporal pattern recognition features are extracted therefrom; The extracted spatial distribution evolution features, spatio-temporal coupling features and spatio-temporal pattern recognition features are integrated to form a spatio-temporal dynamic feature set for quantitatively describing the dynamic pattern of the acupoint electrical activity.

[0012] Preferably, the S6 inputs the spatio-temporal dynamic feature set into a trained machine learning model, the machine learning model learns the mapping relationship between the features and the preset physiological or pathological labels, and outputs the classification result of the current acupoint electrical activity spatial distribution pattern and the prediction result of the future trend; By constructing a feature distribution baseline under the normal mode, using an unsupervised anomaly detection algorithm, comparing the spatio-temporal dynamic feature set with the feature distribution baseline, obtaining the deviation degree, when the deviation degree exceeds the preset deviation threshold, the corresponding spatio-temporal dynamic feature is marked as an abnormal spatial distribution pattern.

[0013] Preferably, the S7 dynamically associates and integrates the prediction result and the abnormal pattern recognition result with the spatio-temporal dynamic feature set and the visual area map, scores the spatial distribution pattern of the acupoint electrical activity in multiple dimensions through a preset evaluation rule, and automatically generates a comprehensive analysis report according to the multi-dimensional score result.

[0014] To achieve the above purpose, the application provides the following technical scheme: an acupoint electrical body surface array signal analysis system based on spatial distribution features, which implements the acupoint electrical body surface array signal analysis method based on spatial distribution features, comprising: A signal acquisition module: acquiring multi-channel bioelectric signals sensed by a matrix array electrode patch placed on the acupoint to be analyzed, and synchronously acquiring external modality data; A signal noise reduction processing module: performing noise reduction processing on the multi-channel bioelectric signals to obtain noise-reduced multi-channel signals; A feature extraction module: processing the noise-reduced multi-channel signals to extract a multi-dimensional feature set representing acupoint electrical activity; An area map generation module: fusing the multi-dimensional feature set with the external modality data, and mapping the fused features to a visual area map corresponding to the matrix array electrode patch according to their spatial position and time information; A joint analysis module: performing spatial-temporal joint analysis on the visual area map to extract a spatio-temporal dynamic feature set quantitatively describing the dynamic pattern of acupoint electrical activity; A prediction and recognition module: using a machine learning algorithm to classify and predict the spatial distribution change pattern of acupoint electrical activity at different time points by taking the spatio-temporal dynamic feature set as input, and automatically identifying abnormal spatial distribution patterns that are significantly different from the normal pattern; Result output module: according to the prediction result and the abnormal pattern recognition result, the spatial distribution pattern of acupoint electrical activity is evaluated, and a comprehensive analysis report is output.

[0015] Technical effects and advantages of the present application: By using the matrix array electrode patch to acquire multi-channel bioelectric signals, and synchronously acquiring external modal data, the data sources are greatly enriched, the association information of acupoint electrical activity and external environment can be more comprehensively and accurately captured, and multi-channel bioelectric signals and external modal data are synchronously acquired, ensuring the consistency of data in time, providing a reliable foundation for subsequent joint analysis.

[0016] Through special noise reduction processing of multi-channel bioelectric signals, the non-linear and non-stationary characteristics of acupoint surface signals can be better adapted, the signal-to-noise ratio can be improved, more key physiological information can be retained, and a multi-dimensional feature set representing acupoint electrical activity can be extracted from the denoised multi-channel signals, covering time domain, frequency domain, spatial domain and other aspects, which can more comprehensively describe the characteristics of acupoint electrical activity. By fusing the multi-dimensional feature set with external modal data, and mapping the spatial position and time information into a visual area graph, the organic fusion and intuitive display of multi-source data are realized, and then the spatial-time joint analysis is performed on the visual area graph to extract a spatio-temporal dynamic feature set quantitatively describing the dynamic pattern of acupoint electrical activity, which can more accurately reveal the changes of acupoint electrical activity in different time and space.

[0017] By using machine learning algorithm for classification and prediction with the spatio-temporal dynamic feature set as input, the complex patterns in the data can be automatically learned, the accuracy of classification and prediction can be improved, the abnormal spatial distribution pattern significantly different from the normal pattern can be automatically identified, and more sensitive and reliable basis for early diagnosis and monitoring of diseases can be provided, according to the prediction result and the abnormal pattern recognition result, the spatial distribution pattern of acupoint electrical activity is evaluated, and a comprehensive analysis report is output, which provides comprehensive and accurate information for researchers and clinicians, and helps to make more reasonable decisions. BRIEF DESCRIPTION OF DRAWINGS

[0018] Figure 1 The method steps of the present application.

[0019] Figure 2 The system structure block diagram of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the present application will be clearly and completely described below with reference to the drawings in the present application. In addition, the forms of each structure described in the following embodiments are only examples, and the spatial distribution feature-based acupoint surface array signal analysis method and system involved in the present application are not limited to each structure described in the following embodiments. All other embodiments obtained by those skilled in the art without creative labor belong to the scope of protection of the present application.

[0021] As Figure 1 The present embodiment provides a spatial distribution feature-based acupoint surface array signal analysis method, comprising: S1: acquiring multi-channel bioelectric signals sensed by a matrix array electrode patch placed on an acupoint to be analyzed, and synchronously acquiring external modality data.

[0022] In the present embodiment, the S1 acquires multi-channel bioelectric signals of the acupoint to be analyzed by a matrix array electrode patch, wherein the matrix array electrode patch comprises a plurality of silver chloride dot electrodes with a neighboring spacing of 3-7 mm, and adjacent silver chloride dot electrodes are respectively configured as acquisition electrodes and output electrodes. The acquired multi-channel bioelectric signals can be acquired by bioelectric signal acquisition modes including the following modes according to different analysis purposes: Dry response acquisition mode: after intervention on a preset distal acupoint of the acupoint to be analyzed, bioelectric signals on the acupoint to be analyzed are acquired in real time at equal time intervals; Resting state acquisition mode: bioelectric signals on the acupoint to be analyzed are continuously acquired in a resting state without external intervention; Disease correlation acquisition mode: bioelectric signals placed on the acupoint to be analyzed corresponding to a specific disease of a user are acquired. When acquiring multi-channel bioelectric signals, external modality data are synchronously acquired, and the external modality data include electrocardiogram, electroencephalogram, acupoint pressure value, electromyogram, respiratory signal, skin temperature and environmental parameters.

[0023] S2: performing noise reduction processing on the multi-channel bioelectric signals to obtain multi-channel signals after noise reduction.

[0024] In the present embodiment, the specific operation of the S2 for performing noise reduction processing on the multi-channel bioelectric signals comprises: The multi-channel bioelectric signals are subjected to filtering processing, and then the signals after filtering processing are subjected to artifact removal processing, the artifact removal processing comprises identifying and removing physiological artifact components by using a blind source separation algorithm, and finally, the signals after artifact removal processing are subjected to smoothing processing to obtain multi-channel signals after noise reduction.

[0025] S3: Process the denoised multi-channel signal to extract a multi-dimensional feature set characterizing the electrical activity of acupoints.

[0026] In this embodiment, the S3 step of extracting a multi-dimensional feature set characterizing the electrical activity of acupoints includes: Time-frequency analysis was performed on the denoised multi-channel signal to extract time-frequency features characterizing the intensity and rhythm of acupoint electrical activity; Nonlinear analysis was performed on the denoised multichannel signal to extract nonlinear features characterizing the complexity and regularity of acupoint electrical activity; Based on the spatial correlation between the signals of each channel in the denoised multi-channel signal, a functional network is constructed to extract spatial network features. The time-frequency features, nonlinear features, and spatial network features are combined to form the multi-dimensional feature set.

[0027] It should be specifically noted that time-frequency analysis employs short-time Fourier transform or wavelet transform to process the denoised multi-channel signal to obtain time-frequency characteristics such as power spectral density, energy proportion of a specific frequency band, and dominant frequency of each channel signal in different frequency bands; nonlinear analysis quantifies its complexity and regularity by calculating the sample entropy, approximate entropy, or Lempel-Ziv complexity of each channel signal; and spatial network characteristics are constructed based on the Pearson correlation coefficient or phase-locked value between each channel signal to build a functional connectivity matrix, and then calculate the network topology properties.

[0028] S4: The multi-dimensional feature set is fused with the external modal data, and the fused features are mapped into a visualization area map corresponding to the matrix array electrode patch based on their spatial location and time information.

[0029] In this embodiment, step S4 aligns the multi-dimensional feature set with the external modal data in a spatiotemporal manner and generates a unified fusion feature vector using a fusion algorithm; Based on the known physical coordinates of each electrode in the matrix array electrode patch, each fused feature vector is mapped to the corresponding spatial coordinate point to form a feature space distribution matrix. The feature space distribution matrix is ​​interpolated and smoothed, and a visualization area map corresponding to the matrix array electrode patch layout is generated according to preset rendering rules.

[0030] It should be specifically explained that the fusion algorithm employs a late-stage fusion strategy, which involves concatenating the multi-dimensional feature set with the aligned pressure and temperature feature vectors to generate a unified fused feature vector. Subsequently, based on the known two-dimensional physical coordinates of each electrode in the matrix array electrode patch, each fused feature vector is mapped to the feature value of the corresponding coordinate point, forming a feature space distribution matrix. Next, this matrix is ​​encrypted using a bilinear interpolation algorithm and smoothed using Gaussian filtering. Finally, according to the preset rendering rules, the magnitude of the feature values ​​is linearly mapped onto a color spectrum from blue to red to generate a visual region map.

[0031] S5: Perform spatial-temporal joint analysis on the visualized area map to extract a set of spatiotemporal dynamic features that quantitatively describe the dynamic pattern of acupoint electrical activity.

[0032] In this embodiment, step S5 stacks multiple visualization area maps arranged in time series to construct a spatiotemporal data cube; By analyzing the spatiotemporal data cube, spatial distribution evolution characteristics, spatiotemporal coupling characteristics, and spatiotemporal pattern recognition characteristics are extracted from it. The extracted spatial distribution evolution features, spatiotemporal coupling features, and spatiotemporal pattern recognition features are integrated to form a spatiotemporal dynamic feature set that quantitatively describes the dynamic pattern of acupoint electroactivity.

[0033] It should be specifically explained that the spatial distribution evolution characteristics are obtained by analyzing the spatiotemporal data cube, calculating the centroid position of the visualized area map at each time point, connecting them to form the trajectory of the centroid changing over time, obtaining the centroid migration trajectory, calculating the rate of change of the area of ​​the high-energy region of acupoint electrical activity over time, obtaining the energy diffusion rate, quantifying the intensity of the change of the geometry of the visualized area map over time, obtaining the morphological fluctuation index, and using the centroid migration trajectory, energy diffusion rate, and morphological fluctuation index as spatial distribution evolution characteristics. The spatiotemporal coupling characteristics are obtained by analyzing the spatiotemporal data cube, calculating the cross-correlation function between signals at different spatial locations to determine the time delay of signal propagation, calculating the coherence of signals in a specific frequency band to quantify the synchronicity of activities, and analyzing the Granger causality between them to infer the predictive or driving effect of activities in one region on another region. Thus, the obtained time delay correlation, coherence and causality are used as spatiotemporal coupling characteristics. Spatiotemporal pattern recognition features are obtained by analyzing the spatiotemporal data cube and using template matching or convolutional neural networks to identify and count preset or self-learned recurring spatiotemporal dynamic patterns in the spatiotemporal data cube. Instantaneous spatial activation events with specific spatial morphology that appear and disappear quickly in a short period of time are identified, as well as persistent functional connection units that maintain stable spatial structure and synchronicity over a longer time scale. The obtained statistical features of recurring patterns, transient activation event features, and persistent functional connection features are used as spatiotemporal pattern recognition features.

[0034] S6: Using the spatiotemporal dynamic feature set as input, a machine learning algorithm is used to classify and predict the spatial distribution change patterns of acupoint electrical activity at different time points, and to automatically identify abnormal spatial distribution patterns that are significantly different from the normal patterns.

[0035] In this embodiment, step S6 inputs the spatiotemporal dynamic feature set into the trained machine learning model. The machine learning model learns the mapping relationship between features and preset physiological or pathological labels, and outputs the classification result and prediction result of the current spatial distribution pattern of acupoint electrical activity. By constructing a feature distribution baseline under normal conditions, an unsupervised anomaly detection algorithm is used to compare the spatiotemporal dynamic feature set with the feature distribution baseline to obtain the degree of deviation. When the degree of deviation exceeds a preset deviation threshold, the corresponding spatiotemporal dynamic feature is marked as an abnormal spatial distribution pattern.

[0036] It should be specifically noted that the machine learning model employs a Long Short-Term Memory (LSTM) network, with its input being a spatiotemporal dynamic feature set. During the training phase, the spatiotemporal dynamic feature set, labeled by clinical experts with tags such as "healthy," "Qi deficiency," and "blood stasis," is first divided into a training set and a validation set according to a preset ratio. Then, the LSTM model is iteratively trained using the training set, with the cross-entropy loss function as the optimization objective. The Adam optimizer is used to update the network weights, and the optimal training epochs are determined by monitoring the accuracy on the validation set, thereby obtaining a well-trained model. This model can classify the current spatial distribution pattern (e.g., determine it as a Qi deficiency pattern) and predict the pattern evolution trend within the next few minutes. Meanwhile, the unsupervised anomaly detection algorithm specifically employs an isolated forest model. First, it collects a large number of spatiotemporal dynamic feature sets from healthy subjects, calculates their mean and covariance matrices to construct a baseline for the feature distribution of normal patterns, and then inputs the spatiotemporal dynamic feature set to be tested into the trained isolated forest model. The model outputs an anomaly score as the degree of deviation. When this degree of deviation exceeds a preset deviation threshold, anomaly identification information is generated. This anomaly identification information is dynamically marked on the generated visualization map in the form of highlighting, flashing, or covering with a specific color, thereby enabling users to intuitively locate acupoints with abnormal electrical activity. The preset deviation threshold is determined through cross-validation.

[0037] S7: Based on the prediction results and abnormal pattern recognition results, evaluate the spatial distribution pattern of acupoint electrical activity and output a comprehensive analysis report.

[0038] In this embodiment, step S7 dynamically associates and integrates the prediction results and abnormal pattern recognition results with the spatiotemporal dynamic feature set and the visualized region map, performs multi-dimensional scoring on the spatial distribution pattern of acupoint electrical activity through preset evaluation rules, and automatically generates a comprehensive analysis report based on the multi-dimensional scoring results.

[0039] It should be specifically explained that the specific process for achieving quantitative assessment and report generation of acupoint status includes: First, the anomaly scores from the anomaly pattern recognition results are overlaid and rendered as a heatmap on a visualization area map, and the predicted future trends are marked on the map with dynamic arrows or trajectory lines to achieve dynamic correlation. The preset evaluation rules are as follows: calculate the "abnormality degree" dimension score based on the anomaly scores, calculate the trend risk dimension score based on the deviation between the predicted results and the current pattern, and calculate the stability dimension score based on the volatility of the centroid migration trajectory. After normalizing the scores of each dimension to 0-100, they are weighted and summed according to preset weights to obtain a comprehensive score. Finally, a comprehensive analysis report is automatically generated based on the scoring results. The specific contents include: the comprehensive score and the sub-scores of each dimension displayed in numerical and star ratings, a dynamic visualization area map marked with anomaly areas and evolution trends, and preliminary intervention suggestions based on a preset knowledge base.

[0040] like Figure 2 The embodiment shown provides an implementation system for a hole electric field surface array signal analysis method based on spatial distribution characteristics. The system includes a signal acquisition module, a signal denoising module, a feature extraction module, a region map generation module, a joint analysis module, a prediction and recognition module, and a result output module. The signal acquisition module is connected to the signal denoising module, the signal denoising module is connected to the feature extraction module, the feature extraction module is connected to the region map generation module, the region map generation module is connected to the joint analysis module, the joint analysis module is connected to the prediction and recognition module, and the prediction and recognition module is connected to the result output module.

[0041] The signal acquisition module acquires multi-channel bioelectric signals sensed by the matrix array electrode patches placed on the acupoints to be analyzed, and simultaneously acquires external modal data; The signal noise reduction processing module performs noise reduction processing on the multi-channel bioelectric signal to obtain a noise-reduced multi-channel signal. The feature extraction module processes the denoised multi-channel signal to extract a multi-dimensional feature set characterizing the electrical activity of acupoints; The region map generation module fuses the multi-dimensional feature set with the external modal data, and maps the fused features into a visual region map corresponding to the matrix array electrode patch based on its spatial location and temporal information. The joint analysis module performs spatial-temporal joint analysis on the visualized area map to extract a set of spatiotemporal dynamic features that quantitatively describe the dynamic patterns of acupoint electrical activity. The prediction and recognition module takes the spatiotemporal dynamic feature set as input, uses machine learning algorithms to classify and predict the spatial distribution change patterns of acupoint electrical activity at different time points, and automatically identifies abnormal spatial distribution patterns that are significantly different from normal patterns. The result output module evaluates the spatial distribution pattern of acupoint electrical activity based on the prediction results and abnormal pattern recognition results, and outputs a comprehensive analysis report.

[0042] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0043] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for analyzing hole electric field surface array signals based on spatial distribution characteristics, characterized in that, include: S1: Acquire multi-channel bioelectrical signals sensed by the matrix array electrode patches placed on the acupoints to be analyzed, and simultaneously acquire external modal data; S2: The multi-channel bioelectric signal is denoised to obtain a denoised multi-channel signal; S3: Process the denoised multi-channel signal to extract a multi-dimensional feature set characterizing the electrical activity of acupoints; S4: The multi-dimensional feature set is fused with the external modal data, and the fused features are mapped into a visualization area map corresponding to the matrix array electrode patch based on their spatial location and time information. S5: Perform spatial-temporal joint analysis on the visualized area map to extract a set of spatiotemporal dynamic features that quantitatively describe the dynamic pattern of acupoint electrical activity; S6: Using the spatiotemporal dynamic feature set as input, a machine learning algorithm is used to classify and predict the spatial distribution change patterns of acupoint electrical activity at different time points, and to automatically identify abnormal spatial distribution patterns that are significantly different from normal patterns. S7: Based on the prediction results and abnormal pattern recognition results, evaluate the spatial distribution pattern of acupoint electrical activity and output a comprehensive analysis report.

2. The method for analyzing hole electric field surface array signals based on spatial distribution characteristics according to claim 1, characterized in that, The S1 acquires multi-channel bioelectric signals from the acupoints to be analyzed through a matrix array electrode patch, wherein the matrix array electrode patch includes multiple silver chloride point electrodes with an adjacent spacing of 3-7 mm, and adjacent silver chloride point electrodes are respectively configured as acquisition electrodes and output electrodes. The acquired multi-channel bioelectrical signals can be obtained using the following acquisition modes, depending on the specific analytical purpose: Interference response acquisition mode: After intervention is performed on the preset distal acupoint of the acupoint to be analyzed, the bioelectrical signals on the acupoint to be analyzed are acquired in real time at equal time intervals; Resting state acquisition mode: Bioelectric signals on the acupoints to be analyzed are continuously acquired in a resting state without external intervention; Disease-related data acquisition mode: Acquires bioelectrical signals placed on acupoints corresponding to a specific disease of the user; When acquiring multi-channel bioelectrical signals, external modal data is acquired simultaneously, including electrocardiogram, electroencephalogram, electromyogram, respiratory signal, skin temperature, and environmental parameters.

3. The method for analyzing hole electric field surface array signals based on spatial distribution characteristics according to claim 2, characterized in that, The specific operations of S2 for noise reduction processing of the multi-channel bioelectric signals include: The multi-channel bioelectric signal is filtered, and then the filtered signal is subjected to artifact removal processing. The artifact removal processing includes using a blind source separation algorithm to identify and remove physiological artifact components. Finally, the signal after artifact removal processing is smoothed to obtain a noise-reduced multi-channel signal.

4. The method for analyzing hole electric field surface array signals based on spatial distribution characteristics according to claim 3, characterized in that, The S3 extracts a multi-dimensional feature set characterizing the electrical activity of acupoints, including: Time-frequency analysis was performed on the denoised multi-channel signal to extract time-frequency features characterizing the intensity and rhythm of acupoint electrical activity; Nonlinear analysis was performed on the denoised multichannel signal to extract nonlinear features characterizing the complexity and regularity of acupoint electrical activity; Based on the spatial correlation between the signals of each channel in the denoised multi-channel signal, a functional network is constructed to extract spatial network features. The time-frequency features, nonlinear features, and spatial network features are combined to form the multi-dimensional feature set.

5. The method for analyzing hole electric field surface array signals based on spatial distribution characteristics according to claim 4, characterized in that, S4 aligns the multi-dimensional feature set with the external modal data in a spatiotemporal manner and generates a unified fusion feature vector using a fusion algorithm. Based on the known physical coordinates of each electrode in the matrix array electrode patch, each fused feature vector is mapped to the corresponding spatial coordinate point to form a feature space distribution matrix. The feature space distribution matrix is ​​interpolated and smoothed, and a visualization area map corresponding to the matrix array electrode patch layout is generated according to preset rendering rules.

6. The method for analyzing hole electric field surface array signals based on spatial distribution characteristics according to claim 5, characterized in that, S5 stacks multiple visualization area maps arranged in time series to construct a spatiotemporal data cube; By analyzing the spatiotemporal data cube, spatial distribution evolution characteristics, spatiotemporal coupling characteristics, and spatiotemporal pattern recognition characteristics are extracted from it. The extracted spatial distribution evolution features, spatiotemporal coupling features, and spatiotemporal pattern recognition features are integrated to form a spatiotemporal dynamic feature set that quantitatively describes the dynamic pattern of acupoint electroactivity.

7. The method for analyzing hole electric field array signals based on spatial distribution characteristics according to claim 6, characterized in that, S6 inputs the spatiotemporal dynamic feature set into the trained machine learning model. The machine learning model learns the mapping relationship between features and preset physiological or pathological labels, and outputs the classification result of the current spatial distribution pattern of acupoint electrical activity and the prediction result of future trends. By constructing a feature distribution baseline under normal conditions, an unsupervised anomaly detection algorithm is used to compare the spatiotemporal dynamic feature set with the feature distribution baseline to obtain the degree of deviation. When the degree of deviation exceeds a preset deviation threshold, the corresponding spatiotemporal dynamic feature is marked as an abnormal spatial distribution pattern.

8. The method for analyzing hole electric field array signals based on spatial distribution characteristics according to claim 7, characterized in that, S7 dynamically correlates and integrates the prediction results and abnormal pattern recognition results with the spatiotemporal dynamic feature set and the visualized region map, performs multi-dimensional scoring on the spatial distribution pattern of acupoint electrical activity through preset evaluation rules, and automatically generates a comprehensive analysis report based on the multi-dimensional scoring results.

9. A signal analysis system for a hole electric field array based on spatial distribution characteristics, implementing the signal analysis method for a hole electric field array based on spatial distribution characteristics as described in any one of claims 1-8, characterized in that, include: Signal acquisition module: acquires multi-channel bioelectric signals sensed by the matrix array electrode patches placed on the acupoints to be analyzed, and simultaneously acquires external modal data; Signal noise reduction processing module: performs noise reduction processing on the multi-channel bioelectric signals to obtain noise-reduced multi-channel signals; Feature extraction module: processes the denoised multi-channel signal to extract a multi-dimensional feature set characterizing the electrical activity of acupoints; Region map generation module: It fuses the multi-dimensional feature set with the external modal data, and maps the fused features into a visual region map corresponding to the matrix array electrode patch based on its spatial location and temporal information; Joint analysis module: Performs spatial-temporal joint analysis on the visualized area map to extract a set of spatiotemporal dynamic features that quantitatively describe the dynamic patterns of acupoint electrical activity; Prediction and Recognition Module: Taking the spatiotemporal dynamic feature set as input, using machine learning algorithms, classifying and predicting the spatial distribution change patterns of acupoint electrical activity at different time points, and automatically identifying abnormal spatial distribution patterns that are significantly different from normal patterns; Results output module: Based on the prediction results and abnormal pattern recognition results, evaluate the spatial distribution pattern of acupoint electrical activity and output a comprehensive analysis report.

Images