Human body acupuncture scheme intelligent screening method and system based on brain-computer interface technology

Abstract

The invention relates to the technical field of traditional Chinese medicine acupuncture and moxibustion, in particular to a human body acupuncture and moxibustion scheme intelligent screening method and system based on the brain-computer interface technology. Inputting the nerve response characteristic parameters into an acupoint efficacy prediction model, outputting nerve regulation efficacy scores of the acupoints, and generating an initial acupuncture scheme; in the process of executing the initial acupuncture scheme, electroencephalogram signals are collected in real time, and a neural feedback intensity value is calculated; and when the feedback intensity is lower than a preset response baseline, triggering a dynamic weight optimization algorithm, updating the scoring model and generating an optimized acupuncture scheme. According to the invention, objective evaluation and dynamic optimization of the human body to different acupoint nerve responses are realized, and the method has the advantages of high intelligence, high real-time performance, excellent individualized adaptation capability and the like, and is suitable for intelligent decision system deployment for assisting acupuncture therapy.

Description

Technical Field

[0001] The present invention relates to the technical field of traditional Chinese medicine acupuncture, and in particular to a method and system for intelligent screening of human acupuncture schemes based on brain-computer interface technology. Background Art

[0002] Guided by traditional Chinese medicine theory, acupuncture, as an important therapy for harmonizing yin and yang, unblocking meridians, and stimulating the body's self-healing capacity, has been widely used as an adjunct treatment for a variety of conditions, including neuromodulation, pain control, insomnia and anxiety, and cerebrovascular rehabilitation. In recent years, with the rapid development of brain science and biosignal processing technology, a growing number of studies have attempted to correlate electroencephalogram (EEG) signals with acupoint stimulation responses, revealing the objective mechanisms of acupuncture's effects from a neurophysiological perspective and promoting its scientific and intelligent transformation.

[0003] However, existing technologies still face several key technical bottlenecks in the field of intelligent acupuncture screening: first, most methods rely on static sampling or clinical experience to recommend acupoints, lack systematic modeling and quantitative evaluation of EEG signal characteristics, and cannot accurately reflect the differences in the regulation of neural networks by different acupoints; second, there is a lack of a closed-loop feedback mechanism, and it is impossible to dynamically adjust the stimulation strategy based on the patient's neural response during treatment, resulting in a deviation between the actual therapeutic effect and the model output; third, current neural feedback analysis mostly uses a single indicator (such as power spectrum or ERP), and fails to integrate multi-source features to construct a high-dimensional response representation, which limits the sensitivity and generalization ability of the scoring mechanism. Summary of the Invention

[0004] The present invention provides a method and system for intelligent screening of human acupuncture plans based on brain-computer interface technology, and provides a new technical path that integrates EEG signal feature extraction, multidimensional scoring modeling and real-time optimization feedback, realizing an intelligent acupuncture plan screening method that is controllable throughout the entire process of "perception-scoring-feedback-optimization".

[0005] The intelligent screening method for human acupuncture programs based on brain-computer interface technology includes the following steps: S1: acquiring the original EEG signal of the patient when receiving stimulation of a preset acupoint group through an EEG acquisition device, and extracting neural response characteristic parameters from the original EEG signal; S2: Inputting the neural response characteristic parameters into a pre-trained acupoint efficacy prediction model, outputting a neural regulation efficacy score corresponding to each acupoint, screening acupoints with a score above a threshold based on the neural regulation efficacy score, and generating an initial acupuncture plan; S3: During the implementation of the initial acupuncture program, the patient's real-time neurofeedback intensity value is continuously monitored by an EEG acquisition device; S4: If the real-time neurofeedback intensity value is lower than the preset response baseline, the dynamic weight optimization algorithm is triggered to recalculate the acupoint neuromodulation efficacy score, and an optimized acupuncture plan is generated and executed based on the updated neuromodulation efficacy score.

[0006] Optionally, the S1 includes: S11: synchronously collecting, by the EEG acquisition device, multi-channel raw EEG signals of the patient when receiving stimulation of a preset acupoint group, performing bandpass filtering and power frequency noise elimination processing on the raw EEG signals to generate preprocessed EEG signals; S12: performing independent component analysis on the pre-processed EEG signal to remove interference from electrooculographic artifacts and myoelectric artifacts to generate a pure EEG signal.

[0007] Optionally, the S1 further includes: S13: extracting an event-related potential waveform from the clean EEG signal, and locking the P300 component amplitude within a time window of 0-800 ms after the stimulus is applied as a first neural response characteristic parameter; S14: performing wavelet transform decomposition on the clean EEG signal, and calculating the average power spectral density of the θ frequency band within a time window of 1-3 seconds after the stimulation is applied as a second neural response characteristic parameter; S15: constructing a brain functional connection network based on the pure EEG signal, using the phase lag index between the prefrontal lobe and limbic system channels as the connection strength, and calculating the network small-world attribute value as the third neural response characteristic parameter; S16: Integrate the first neural response characteristic parameter, the second neural response characteristic parameter, and the third neural response characteristic parameter to generate a neural response characteristic parameter set.

[0008] Optionally, the pre-trained acupoint efficacy prediction model includes: Input layer: used to receive the neural response feature parameter set; A feature fusion module: connected to the input layer, configured to perform weighted fusion on the first neural response feature parameter, the second neural response feature parameter, and the third neural response feature parameter in the neural response feature parameter set to generate a fused feature vector; Acupoint efficacy calculation module: including a fully connected neural network, used to map the fused feature vector to the intermediate value of the neural regulation efficacy corresponding to each acupoint; Normalization output layer: used to normalize the intermediate value of the neural regulation efficacy to generate a neural regulation efficacy score.

[0009] Optionally, the S2 includes: S21: inputting the generated neural response feature parameter set into the input layer of the pre-trained acupoint efficacy prediction model; S22: performing weighted fusion on the first neural response feature parameter, the second neural response feature parameter, and the third neural response feature parameter in the neural response feature parameter set using a feature fusion module of the acupoint efficacy prediction model to generate a fused feature vector; S23: inputting the fused feature vector into the acupoint efficacy calculation module of the acupoint efficacy prediction model, and generating the intermediate value of the neural regulation efficacy corresponding to each acupoint through calculation by a fully connected neural network; S24: normalizing the intermediate values ​​of the neuromodulation efficacy through the normalized output layer of the acupoint efficacy prediction model to generate a neuromodulation efficacy score for each acupoint.

[0010] Optionally, the S2 further includes: S25: comparing the neuromodulation efficacy score with a preset threshold, and screening out all acupuncture points having neuromodulation efficacy scores higher than the preset threshold; S26: Based on the screened acupoints and their corresponding neuromodulatory efficacy scores, an initial acupuncture plan including an acupoint list and stimulation parameters is generated.

[0011] Optionally, the S3 includes: S31, during the implementation of the initial acupuncture program, continuously monitoring the patient's real-time first neural response characteristic parameter, second neural response characteristic parameter, and third neural response characteristic parameter through an electroencephalogram acquisition device; S32, calculating a real-time neural feedback intensity value based on the patient's real-time first neural response characteristic parameter, second neural response characteristic parameter, and third neural response characteristic parameter , calculated as: ; in, is the real-time first neural response characteristic parameter, is the real-time second neural response characteristic parameter, is the real-time third neural response characteristic parameter, is the preset weight coefficient.

[0012] Optionally, the S4 includes: S41: If the real-time neurofeedback intensity value is lower than a preset response baseline, triggering a dynamic weight optimization algorithm; S42: Calculate the deviation between the real-time neurofeedback intensity value and the preset response baseline through a dynamic weight optimization algorithm , calculated as: ; in, is the preset response baseline value; S43: Adjust the weight coefficient of the feature fusion module according to the deviation. The update formula is: ; in, To optimize the weight coefficient, is the original weight coefficient, is the regulating factor.

[0013] Optionally, the S4 further includes: S44: updating the optimized weight coefficient to the feature fusion module of the acupoint efficacy prediction model to generate an optimized acupoint efficacy prediction model; S45: Inputting the neural response characteristic parameter set into the optimized acupoint efficacy prediction model, recalculating the neural regulation efficacy score corresponding to each acupoint, and generating an updated acupoint neural regulation efficacy score; S46: Based on the updated acupoint nerve regulation efficacy score, screening acupoints that are higher than a preset threshold value and generating an optimized acupuncture plan; S47: Terminate the current initial acupuncture plan and execute the optimized acupuncture plan.

[0014] The intelligent screening system for human acupuncture programs based on brain-computer interface technology is used to implement the above-mentioned intelligent screening method for human acupuncture programs based on brain-computer interface technology, and includes the following modules: EEG acquisition module: used to collect raw EEG signals during the stimulation of preset acupoint groups, complete bandpass filtering and power frequency noise elimination processing, and output pure EEG signals; Neural response feature extraction module: used to extract the P300 component amplitude, theta frequency band average power spectral density and prefrontal-limbic system phase lag index from the clean EEG signal to generate a neural response feature parameter set; Acupoint efficacy prediction module: used to map the neural response characteristic parameter set into the neural regulation efficacy score of each acupoint; An initial plan generating module is used to screen acupoints whose efficacy is higher than a preset threshold based on the scores, and generate an initial acupuncture plan including an acupoint list and stimulation parameters; Neurofeedback monitoring module: used to continuously collect real-time EEG signals and calculate real-time neurofeedback intensity values ​​during the implementation of the initial plan; Dynamic optimization module: used to trigger the dynamic weight optimization algorithm when the feedback intensity value is lower than the response baseline, update the feature fusion module weight, regenerate the optimized acupuncture plan and execute it instead.

[0015] Beneficial effects of the present invention: This method extracts three types of neural response characteristic parameters (P300 component, theta-band power spectral density, and small-world properties) from raw EEG signals through preprocessing, artifact removal, event-related potential extraction, wavelet transform analysis, and brain functional connectivity network construction. This system generates an objective and quantifiable set of neural response characteristic parameters, eliminating the subjectivity of traditional empirical acupoint selection methods. An acupoint efficacy prediction model constructed through feature fusion and a fully connected neural network accurately outputs a neural regulation efficacy score for each acupoint, ensuring that the generated initial acupuncture plan is consistent with the patient's current neurological state, thereby improving the physiological rationality and individual adaptability of acupoint selection.

[0016] This invention achieves real-time EEG signal acquisition, dynamic feature extraction, and neural feedback intensity value calculation, allowing timely capture of changes in a patient's neural response to the initial acupuncture regimen. Through quantitative analysis of feedback deviations and dynamic updating of feature fusion module weight coefficients, the scoring strategy and prediction model are automatically corrected when insufficient feedback occurs, enabling the acupuncture regimen to be adaptively optimized based on the patient's actual neural response. This closed-loop mechanism ensures the timeliness and controllability of the acupoint screening strategy throughout the intervention process, effectively improving the response matching and sustained efficacy of neural regulation.

[0017] This invention not only integrates traditional Chinese medicine acupoint knowledge with modern EEG analysis techniques, but also introduces adjustable parameters (such as weight coefficients, scaling factors, and adjustment factors) and a learning structure (neural network predictor) at the model level, enabling the system to possess excellent transferability and generalization performance across diverse disease states, stimulation patterns, and individual differences. Furthermore, its clear algorithmic flow and manageable computational burden facilitate its deployment in wearable EEG devices and clinical decision-making support systems, providing technical support for advancing the intelligent, data-based, and precise development of TCM acupuncture. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention; Figure 2 Schematic diagram of the system flow of an embodiment of the present invention. DETAILED DESCRIPTION

[0020] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. It is also noted that, to provide a more detailed description, the following embodiments are best and preferred embodiments, and those skilled in the art may employ alternative methods for implementing certain known technologies. Furthermore, the accompanying drawings are intended only to provide a more detailed description of the embodiments and are not intended to limit the present invention.

[0021] It should be noted that references in the specification to "one embodiment," "an embodiment," "exemplary embodiments," "some embodiments," etc. indicate that the described embodiments may include specific features, structures, or characteristics, but not necessarily every embodiment will include such specific features, structures, or characteristics. Furthermore, when specific features, structures, or characteristics are described in conjunction with an embodiment, it is within the knowledge of persons skilled in the relevant art to implement such features, structures, or characteristics in conjunction with other embodiments (whether or not explicitly described).

[0022] In general, terms can be understood, at least in part, from their use in context. For example, depending at least in part on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending at least in part on the context, allow for the presence of other factors that are not necessarily explicitly described.

[0023] like Figure 1 As shown, the intelligent screening method for human acupuncture programs based on brain-computer interface technology includes the following steps: S1: Obtain the original EEG signals of the patient when receiving the preset acupoint group stimulation through the EEG acquisition device, and extract the neural response characteristic parameters in the original EEG signals; S2: Input the neural response characteristic parameters into the pre-trained acupoint efficacy prediction model, output the neural regulation efficacy score corresponding to each acupoint, screen the acupoints with a score above the threshold based on the neural regulation efficacy score, and generate the initial acupuncture plan; S3: During the implementation of the initial acupuncture program, the patient's real-time neurofeedback intensity value is continuously monitored through the EEG acquisition device; S4: If the real-time neurofeedback intensity value is lower than the preset response baseline, the dynamic weight optimization algorithm is triggered to recalculate the acupoint neuromodulation efficacy score, and an optimized acupuncture plan is generated and executed based on the updated neuromodulation efficacy score.

[0024] S1 includes: The EEG acquisition device includes: Multi-lead electrode array, using silver / silver chloride electrodes arranged in the international 10-20 system, covering the frontal, parietal, temporal, and occipital brain regions, with ≥16 channels, used to contact the patient's scalp to collect raw EEG signals; Bioelectric signal amplifier, connected to a multi-lead electrode array, with a common mode rejection ratio ≥110dB, input impedance ≥100MΩ, and an adjustable gain range of 5000-20000 times, used to amplify raw EEG signals; Analog-to-digital conversion module, with a resolution of ≥24 bits and a sampling rate of 250-1000 Hz, is used to convert analog EEG signals into digital raw EEG signals; Stimulus synchronization marker unit, used to inject event markers into the EEG signal stream when acupuncture stimulation is applied, to achieve precise time alignment of the stimulation event with the original EEG signal; Hardware preprocessor with built-in 0.5-45Hz bandpass filter and 50 / 60Hz power frequency notch filter, used for initial noise reduction of digital raw EEG signals.

[0025] S11, Raw EEG Signal Acquisition and Preprocessing: A multi-lead electrode array is placed on the patient's scalp, evenly spaced using the international 10-20 system. Silver / silver chloride electrodes are used, covering the frontal lobe (e.g., Fp1, Fp2), parietal lobe (e.g., P3, P4), temporal lobe (e.g., T7, T8), and occipital lobe (e.g., O1, O2). The array should have at least 16 channels to ensure coverage of the main neural functional areas for subsequent feature extraction.

[0026] The multi-lead electrode array is connected to a bioelectric signal amplifier with a common-mode rejection ratio ≥110dB, an input impedance ≥100MΩ, and adjustable gain from 5,000 to 20,000x. This amplifier amplifies the weak raw EEG signals and transmits them to an analog-to-digital conversion module for sampling and digitization. The module's sampling rate is set at 250Hz and its resolution is 24-bit, enabling high-fidelity reproduction of the signal waveform.

[0027] At the moment of acupuncture or electroacupuncture stimulation, the stimulation synchronization marking unit injects event markers into the EEG acquisition system to achieve temporal alignment of the stimulation event and the EEG signal for subsequent event-related potential (ERP) calculation.

[0028] The collected digital raw EEG signals are then preliminarily filtered by the hardware preprocessor built into the system. The preprocessor integrates a 0.5–45Hz bandpass filter and a 50 / 60Hz power frequency notch filter to output the preprocessed EEG signals. , used for subsequent analysis.

[0029] S12, Independent Component Analysis (ICA) artifact removal: preprocessing EEG signals Apply the Independent Component Analysis (ICA) algorithm to separate the source signal components with statistical independence. The goals of ICA are: ; in, represents the observed EEG mixed signal matrix, is the mixing matrix, For independent source signals.

[0030] Identify the electrooculographic artifacts (such as blink EOG) and electromyographic artifacts (high-frequency EMG) manually or automatically, and remove the inverse transform at the source signal level to obtain a pure EEG signal. .

[0031] S13, event-related potential (ERP) P300 extraction: After each acupoint stimulation event (Event Marker), the EEG data within the 0-800ms time window was intercepted and averaged across all trials to obtain a typical event-related potential (ERP) waveform.

[0032] Lock the P300 component, which is defined as the positive peak that appears around 300ms after the stimulus, and record its maximum amplitude as the first neural response characteristic parameter ,Right now: ; For example, when a patient receives stimulation at the Zusanli acupoint, the ERP waveform has a main peak at 420ms with an amplitude of 9.2μV. The first neural response characteristic parameter corresponding to the stimulation is .

[0033] S14, wavelet transform and Band power spectrum density extraction: Daubechies-based wavelet (db4) is used to perform discrete wavelet decomposition on the pure EEG signal to extract the power spectrum density within 1-3 seconds after stimulation. The energy of the frequency band (4-8Hz) is used to calculate the average power spectral density (PSD) as the second neural response characteristic parameter .

[0034] The calculation formula is as follows: ; in, , express Frequency band at time The power density value.

[0035] For example, wavelet analysis of the EEG signal after stimulation of Baihui acupoint can yield the following results within 1-3 seconds: The average PSD of the frequency band is , then the second neural response characteristic parameter corresponding to the stimulus is .

[0036] S15, brain functional connectivity network construction and small-world property calculation: Based on pure EEG signals, prefrontal cortex (such as Fp1, Fp2) and limbic system (such as T7, T8, Cz) channel pairs are selected to calculate the phase lag index (PLI) between them for constructing the neural connectivity matrix.

[0037] Constructing an undirected weighted graph , the nodes are brain area channels, the edge weights are the PLI values ​​between the corresponding channels, and the small-world property coefficients are further calculated , calculated as: ; in, is the network clustering coefficient, is the characteristic path length, The small-worldness of the random network corresponds to the index that reflects the integration and separation efficiency of the neural network. This value is used as the third neural response characteristic parameter. .

[0038] S16, feature parameter set generation: the above three parameters Integrate according to a unified structure to generate a neural response feature parameter set for subsequent modeling .

[0039] S2 includes: The pre-trained acupoint efficacy prediction model includes an input layer, a feature fusion module, an acupoint efficacy calculation module, and a normalized output layer.

[0040] The input layer is used to receive a neural response feature parameter set; Feature fusion module: connected to the input layer, used to perform weighted fusion on the first neural response feature parameter, the second neural response feature parameter, and the third neural response feature parameter in the neural response feature parameter set to generate a fused feature vector; The acupoint efficacy calculation module includes a fully connected neural network, which is used to map the fused feature vector to the intermediate value of the neural regulation efficacy corresponding to each acupoint; The normalized output layer is used to normalize the intermediate values ​​of the neuromodulation efficacy to generate a neuromodulation efficacy score.

[0041] S21, neural response feature parameter set input: the neural response feature parameter set Input to the input layer of the pre-trained acupoint efficacy prediction model.

[0042] In this embodiment, each parameter is normalized to the interval [0, 1]. The normalization processing example is: , similarly we can get ; S22, feature weighted fusion to generate fused feature vector: The feature fusion module performs weighted fusion operation on the three types of input features to generate a fused feature vector The fusion formula is: ; in, is the weight coefficient set in the feature fusion module, satisfy .

[0043] S23, the acupoint effectiveness calculation module generates an intermediate value: the fusion feature vector The data is input to the acupoint efficacy calculation module, and forward propagated through the trained fully connected neural network to output the intermediate value of the neural regulation efficacy corresponding to each acupoint. Assume that the neural network has a three-layer structure, namely the input layer (1 node), the hidden layer (8 nodes, ReLU activation), and the output layer (corresponding to the number of acupuncture points), then: ; in is the total number of acupoints to be evaluated.

[0044] S24, the normalized output layer generates a neural regulation effectiveness score: the median value of the neural regulation effectiveness The Sigmoid normalization function is applied for scoring, and the calculation formula is as follows: ; in, For the The neuroregulatory efficacy score of each acupoint, is the scaling factor.

[0045] In the present invention, This value controls the steepness of the sigmoid normalization function, enhancing the model's ability to distinguish intermediate values ​​of the neuromodulatory efficacy of different acupoints when outputting scores. Cross-validation of hundreds of training data sets revealed that a scaling factor of 1.2 improved the model's score distribution stability and neural response stratification within the score range [50, 100]. This prevented scores from concentrating in the intermediate range (e.g., 60-70) and leading to ambiguity in recognition, thereby improving the model's ability to distinguish highly effective acupoints.

[0046] Example: If the median value of a certain acupoint ,but: ; S25, Acupoint screening: Score the neuromodulatory efficacy of all acupoints With preset threshold Compare and filter out all that meet: .

[0047] In the present invention, the preset threshold is set to 70, which is used to screen out acupoints with significant neuromodulatory efficacy when generating the initial acupuncture plan. According to the results of the early clinical pilot test, when the acupoint neuromodulatory efficacy score is higher than 70 points, the patient's EEG response indicators (such as P300 component, The power spectral density (PSD) increased statistically significantly after stimulation (p < 0.01). Therefore, setting 70 as the functional screening threshold ensures the effectiveness of the plan while preventing low-responsive acupoints from entering the final treatment plan, thereby improving the accuracy of neuromodulation.

[0048] S26, Initial Acupuncture Plan Generation: Based on the selected acupoint set and its corresponding neuromodulation efficacy score, and in combination with clinical acupuncture principles (such as treating the upper disease by removing the lower acupoints and cross-selection of acupoints), the stimulation parameters (stimulation method, current intensity, duration, etc.) of each acupoint are set, and a complete initial acupuncture plan is output, which includes: Acupoint name list, score value of each acupoint, and stimulation parameters corresponding to each acupoint (e.g., frequency 5 Hz, intensity 2 mA, duration 10 min).

[0049] For example, if the acupoints finally screened out are: Hegu (score 88.3), Baihui (score 84.1), Zusanli (score 76.5), then the initial acupuncture plan is: S3 includes: S31, real-time data acquisition: During the implementation of the initial acupuncture program, the patient's first real-time neural response characteristic parameters are acquired through the EEG acquisition device through the S1 acquisition step. , the second neural response characteristic parameter and the third neural response characteristic parameter ; S32, calculating the real-time neural feedback strength value: the real-time first neural response characteristic parameter , the second neural response characteristic parameter and the third neural response characteristic parameter Calculate the real-time neurofeedback intensity value by weight , calculated as: ; in, are preset weight coefficients, which are 0.4, 0.3, and 0.3 respectively.

[0050] S4 includes: S41, feedback deviation detection and dynamic weight optimization algorithm triggering: During the execution of the initial acupuncture program, the system calculates the real-time neural feedback intensity value. , and the preset response baseline value When the judgment result meets the following conditions, it is considered that the feedback effect is insufficient and the dynamic weight optimization algorithm is triggered: ; In this embodiment, the response baseline value It is derived from the mean statistics of multiple standard acupoint stimulations before treatment, and is generally set to the 80% percentile of the patient's own nerve response, for example, If the current stimulus feedback value is , then the trigger condition is met.

[0051] S42, feedback deviation calculation: After the system starts the optimization algorithm, it first calculates the deviation of the current real-time feedback from the baseline , defined as follows: ; Example: ; This deviation measures the degree to which the current stimulus is insufficiently responsive to the nervous system and serves as the basis for adjusting the fusion weight.

[0052] S43, dynamic adjustment of feature fusion module weight coefficient: according to the calculated deviation , the original weight coefficient of the feature fusion module Adjust to form the optimized weight coefficient , the adjustment rules are as follows: ; in, represents the original weight coefficient, It is the adjustment factor, which controls the adjustment range.

[0053] In the present invention, it is set to 0.05, If the value is too large, even a slight deviation will cause a drastic change in the weight, resulting in an unstable scoring system; if the value is too small, the model adjustment range will be insufficient when the neural feedback decreases significantly, making it difficult to reflect dynamic adaptability.

[0054] The results of the preliminary validation experiment on 30 patients showed that When set to 0.05, the model's weight changes under different deviations (5% to 20%) remained within a stable range of ±10%. Furthermore, the optimized acupuncture regimen showed a clear trend of improved neural response, with an average feedback improvement rate of 12.6%. Therefore, the present invention sets the value to 0.05 to balance dynamic sensitivity and model stability.

[0055] The weight vector is then normalized to ensure that the sum is 1: ; Finally, a normalized optimization weight set is formed .

[0056] S44, optimize the acupoint efficacy prediction model generation: the above optimized fusion weight parameters The feature fusion module of the acupoint efficacy prediction model used in S2 is updated to generate an optimized acupoint efficacy prediction model.

[0057] The optimized fusion feature vector calculation formula becomes: ; in, is the neural response feature parameter set extracted from S1.

[0058] S45, recalculate the optimization score: re-input the neural response characteristic parameter set output by S16 into the optimized acupoint efficacy prediction model, repeat the S23 and S24 processes in step S2, and regenerate the updated neural regulation efficacy score for each acupoint , the re-scoring still uses the Siqmoid normalization function: ; in is the intermediate value output by the optimization model.

[0059] S46, Optimizing Acupuncture Plan Generation: Based on the Updated Neuromodulatory Efficacy Score ,The system screens all acupoints again, and the conditions are: , In this embodiment, the threshold is still set to 70. The screening results are used to construct a new optimized acupuncture program, which may include some new acupoints or delete acupoints with low response in the original program.

[0060] The stimulation parameters can also be updated in the optimization plan, and the frequency, intensity and other parameters can be automatically adjusted according to the score value. For example, the higher the score, the stimulation frequency and duration can be appropriately shortened to avoid over-excitement.

[0061] S47, plan switching and execution: the system immediately terminates the currently executed initial acupuncture plan, and starts and executes a new optimized acupuncture plan.

[0062] like Figure 2 As shown, the intelligent screening system for human acupuncture programs based on brain-computer interface technology is used to implement the above-mentioned intelligent screening method for human acupuncture programs based on brain-computer interface technology, and includes the following modules: EEG acquisition module: used to collect raw EEG signals during the stimulation of preset acupoint groups, complete bandpass filtering and power frequency noise elimination processing, and output pure EEG signals; Neural response feature extraction module: used to extract the P300 component amplitude, theta frequency band average power spectral density and prefrontal-limbic system phase lag index from the clean EEG signal to generate a neural response feature parameter set; Acupoint efficacy prediction module: used to map the neural response characteristic parameter set into the neural regulation efficacy score of each acupoint; Initial plan generation module: used to screen acupoints with efficacy higher than a preset threshold based on the scores, and generate an initial acupuncture plan including acupoint list and stimulation parameters; Neurofeedback monitoring module: used to continuously collect real-time EEG signals and calculate real-time neurofeedback intensity values ​​during the implementation of the initial plan; Dynamic optimization module: used to trigger the dynamic weight optimization algorithm when the feedback intensity value is lower than the response baseline, update the feature fusion module weight, regenerate the optimized acupuncture plan and execute it instead.

[0063] The present invention encompasses any alternatives, modifications, equivalents, and solutions that fall within the spirit and scope of the present invention. To provide a thorough understanding of the present invention, specific details are described in detail below in connection with the preferred embodiments of the present invention, but those skilled in the art will be able to fully understand the present invention without these detailed descriptions. Furthermore, to avoid unnecessary confusion regarding the essence of the present invention, well-known methods, processes, procedures, components, and circuits have not been described in detail.

[0064] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. An intelligent screening method for human acupuncture programs based on brain-computer interface technology, characterized in that: The following steps are involved: S1: acquiring the original EEG signal of the patient when receiving stimulation of a preset acupoint group through an EEG acquisition device, and extracting neural response characteristic parameters from the original EEG signal; S2: Inputting the neural response characteristic parameters into a pre-trained acupoint efficacy prediction model, outputting a neural regulation efficacy score corresponding to each acupoint, screening acupoints with a score above a threshold based on the neural regulation efficacy score, and generating an initial acupuncture plan; S3: During the implementation of the initial acupuncture program, the patient's real-time neurofeedback intensity value is continuously monitored by an EEG acquisition device; S4: If the real-time neurofeedback intensity value is lower than the preset response baseline, the dynamic weight optimization algorithm is triggered to recalculate the acupoint neuromodulation efficacy score, and an optimized acupuncture plan is generated and executed based on the updated neuromodulation efficacy score.

2. The method for intelligent screening of human acupuncture programs based on brain-computer interface technology according to claim 1, characterized in that: Said S1 comprises: S11: synchronously collecting, by the EEG acquisition device, multi-channel raw EEG signals of the patient when receiving stimulation of a preset acupoint group, performing bandpass filtering and power frequency noise elimination processing on the raw EEG signals to generate preprocessed EEG signals; S12: performing independent component analysis on the pre-processed EEG signal to remove interference from electrooculographic artifacts and myoelectric artifacts to generate a pure EEG signal.

3. The intelligent screening method for human acupuncture programs based on brain-computer interface technology according to claim 2 is characterized in that: Said S1 further comprises: S13: extracting an event-related potential waveform from the clean EEG signal, and locking the P300 component amplitude within a time window of 0-800 ms after the stimulus is applied as a first neural response characteristic parameter; S14: performing wavelet transform decomposition on the clean EEG signal, and calculating the average power spectral density of the θ frequency band within a time window of 1-3 seconds after the stimulation is applied as a second neural response characteristic parameter; S15: constructing a brain functional connection network based on the pure EEG signal, using the phase lag index between the prefrontal lobe and limbic system channels as the connection strength, and calculating the network small-world attribute value as the third neural response characteristic parameter; S16: Integrate the first neural response characteristic parameter, the second neural response characteristic parameter, and the third neural response characteristic parameter to generate a neural response characteristic parameter set.

4. The intelligent screening method for human acupuncture programs based on brain-computer interface technology according to claim 3 is characterized in that: The pre-trained acupoint efficacy prediction model includes: Input layer: used to receive the neural response feature parameter set; A feature fusion module: connected to the input layer, configured to perform weighted fusion on the first neural response feature parameter, the second neural response feature parameter, and the third neural response feature parameter in the neural response feature parameter set to generate a fused feature vector; Acupoint efficacy calculation module: including a fully connected neural network, used to map the fused feature vector to the intermediate value of the neural regulation efficacy corresponding to each acupoint; Normalization output layer: used to normalize the intermediate value of the neural regulation efficacy to generate a neural regulation efficacy score.

5. The intelligent screening method for human acupuncture programs based on brain-computer interface technology according to claim 4 is characterized in that: The S2 includes: S21: inputting the generated neural response feature parameter set into the input layer of the pre-trained acupoint efficacy prediction model; S22: performing weighted fusion on the first neural response feature parameter, the second neural response feature parameter, and the third neural response feature parameter in the neural response feature parameter set using a feature fusion module of the acupoint efficacy prediction model to generate a fused feature vector; S23: inputting the fused feature vector into the acupoint efficacy calculation module of the acupoint efficacy prediction model, and generating the intermediate value of the neural regulation efficacy corresponding to each acupoint through calculation by a fully connected neural network; S24: normalizing the intermediate values ​​of the neuromodulation efficacy through the normalized output layer of the acupoint efficacy prediction model to generate a neuromodulation efficacy score for each acupoint.

6. The method for intelligent screening of human acupuncture programs based on brain-computer interface technology according to claim 5, characterized in that: Said S2 further comprises: S25: comparing the neuromodulation efficacy score with a preset threshold, and screening out all acupuncture points having neuromodulation efficacy scores higher than the preset threshold; S26: Based on the screened acupoints and their corresponding neuromodulatory efficacy scores, an initial acupuncture plan including an acupoint list and stimulation parameters is generated.

7. The intelligent screening method for human acupuncture programs based on brain-computer interface technology according to claim 6 is characterized in that: The S3 includes: S31, during the implementation of the initial acupuncture program, continuously monitoring the patient's real-time first neural response characteristic parameter, second neural response characteristic parameter, and third neural response characteristic parameter through an electroencephalogram acquisition device; S32, calculating a real-time neural feedback intensity value based on the patient's real-time first neural response characteristic parameter, second neural response characteristic parameter, and third neural response characteristic parameter , calculated as: ； in, is the real-time first neural response characteristic parameter, is the real-time second neural response characteristic parameter, is the real-time third neural response characteristic parameter, is the preset weight coefficient.

8. The method for intelligent screening of human acupuncture programs based on brain-computer interface technology according to claim 7, characterized in that: The S4 includes: S41: If the real-time neurofeedback intensity value is lower than a preset response baseline, triggering a dynamic weight optimization algorithm; S42: Calculate the deviation between the real-time neurofeedback intensity value and the preset response baseline through a dynamic weight optimization algorithm , calculated as: ； in, is the preset response baseline value; S43: Adjust the weight coefficient of the feature fusion module according to the deviation. The update formula is: ； in, To optimize the weight coefficient, is the original weight coefficient, is the regulating factor.

9. The method for intelligent screening of acupuncture regimens for human body based on brain-computer interface technology according to claim 8, characterized in that: Said S4 further comprises: S44: updating the optimized weight coefficient to the feature fusion module of the acupoint efficacy prediction model to generate an optimized acupoint efficacy prediction model; S45: Inputting the neural response characteristic parameter set into the optimized acupoint efficacy prediction model, recalculating the neural regulation efficacy score corresponding to each acupoint, and generating an updated acupoint neural regulation efficacy score; S46: Based on the updated acupoint nerve regulation efficacy score, screening acupoints that are higher than a preset threshold value, and generating an optimized acupuncture plan; S47: Terminate the current initial acupuncture plan and execute the optimized acupuncture plan.

10. A system for intelligent screening of human acupuncture regimens based on brain-computer interface technology, for implementing the method for intelligent screening of human acupuncture regimens based on brain-computer interface technology as claimed in any one of claims 1 to 9, characterized in that: Includes the following modules: EEG acquisition module: used to collect raw EEG signals during the stimulation of preset acupoint groups, complete bandpass filtering and power frequency noise elimination processing, and output pure EEG signals; Neural response feature extraction module: used to extract the P300 component amplitude, theta frequency band average power spectral density and prefrontal-limbic system phase lag index from the clean EEG signal to generate a neural response feature parameter set; Acupoint efficacy prediction module: used to map the neural response characteristic parameter set into the neural regulation efficacy score of each acupoint; An initial plan generating module is used to screen acupoints whose efficacy is higher than a preset threshold based on the scores, and generate an initial acupuncture plan including an acupoint list and stimulation parameters; Neurofeedback monitoring module: used to continuously collect real-time EEG signals and calculate real-time neurofeedback intensity values ​​during the implementation of the initial plan; Dynamic optimization module: used to trigger the dynamic weight optimization algorithm when the feedback intensity value is lower than the response baseline, update the feature fusion module weight, regenerate the optimized acupuncture plan and execute it instead.

Images