Electrical stimulation therapeutic instrument control system for acupuncture for dredging brain and activating collaterals

Abstract

The invention discloses an electrical stimulation therapeutic apparatus control system for brain-dredging collateral-activating acupuncture, and the system comprises the following steps: collecting electrophysiological signals of a plurality of electrical stimulation channels, carrying out the preprocessing of the electrophysiological signals to generate a response matrix, screening out a target electrical stimulation channel which is kept continuously active in a plurality of periods through a K-WTA mechanism, and carrying out the screening of the target electrical stimulation channel, based on skin electricity, myoelectricity, nerve response delay and other data collected by a target channel, an atlas reflecting individual nerve response is constructed, whether a'gas gaining 'state is reached or not is judged, a response level is generated according to the strength of the'gas gaining' state, after the'gas gaining 'state is reached, channel connection weights are updated through a Hebbian learning rule in combination with the linkage frequency between channels, and the response level is generated. And establishing an electrical stimulation path memory map, generating electrical stimulation parameters corresponding to current intensity, frequency, waveform and pulse width according to the gas obtaining level and path weight change, and formulating an electrical stimulation output strategy. The pertinence and the stability of the treatment process can be improved.

Description

Technical Field

[0001] This invention relates to the field of neuromodulation and traditional Chinese medicine acupuncture treatment technology, and in particular to an electrical stimulation therapy device control system for acupuncture to promote brain circulation and improve blood circulation. Background Technology

[0002] With the development of TCM neuromodulation theory and its integration with electrostimulation therapy, acupuncture to unblock the brain and activate blood circulation has gradually become one of the important means of intervention for neurological diseases such as acute cerebral infarction. In existing technologies, most electrostimulation therapy devices use fixed frequency and current parameters to synchronously stimulate multiple channels, lacking the ability to dynamically identify and regulate individual differences in neural responses, making it difficult to achieve precise identification and control of personalized "deqi" states.

[0003] On the one hand, traditional electrostimulation systems lack fine-grained analysis of the response characteristics of electrostimulation channels, making it impossible to extract electrostimulation channels with stable activation characteristics based on neural response patterns over multiple cycles. This results in randomness in stimulation path selection, hindering precise current control and efficacy optimization. On the other hand, existing technologies generally neglect the interactive regulation mechanism between electrophysiological signals and physiological feedback signals. They cannot dynamically adjust stimulation parameters and channel weights based on multi-source feedback such as skin conductance, electromyography, and neural delay, making it difficult to effectively construct a memorable and plastically modifiable electrostimulation pathway model. The judgment of the "deqi" state remains based on subjective perception or a single physiological indicator, lacking an objective evaluation mechanism that integrates channel synchronicity and feedback trends. This limits the quantitative evaluation of electrostimulation efficacy and the ability to generate strategies.

[0004] Therefore, how to provide a control system for an electrical stimulation therapy device used in acupuncture to promote blood circulation and relieve brain stagnation is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose a control system for an electrostimulation therapy device used in acupuncture to promote brain circulation and improve blood circulation. This invention integrates multi-cycle sparse screening, individual neural response modeling, Hebbian rule-based reinforcement learning and feedback modulation mechanisms. The system realizes dynamic selection of target electrostimulation channel sets, construction of individual response maps, judgment of Qi intensity levels, generation of electrostimulation pathway memory maps, and parameter strategy optimization control. It has the advantages of personalized response, plasticity of stimulation pathways, and adaptive adjustment of output parameters.

[0006] An electrical stimulation therapy device control system for acupuncture to promote blood circulation and relieve brain stagnation, according to an embodiment of the present invention, includes the following steps:

[0007] The signal encoding module is used to acquire raw electrophysiological signals from multiple electrical stimulation channels, perform preprocessing, and generate a temporal response matrix for the electrical stimulation channels.

[0008] The target channel screening module is used to apply the K-WTA mechanism to perform multi-cycle sparse screening on the electrical stimulation channel temporal response matrix to dynamically select a set of target electrical stimulation channels with continuous activation characteristics.

[0009] The sensory feedback modeling module is used to combine the skin conductance changes and electromyographic response delays corresponding to each electrical stimulation channel in the target electrical stimulation channel set to construct a nonlinear response model between the electrical stimulation channel and the individual neural response.

[0010] The Qi-obtaining identification module is used to analyze the temporal characteristics of multi-channel synchronous activation in the nonlinear response model, and to determine whether the current state of Qi-obtaining is entered based on preset criteria, and output the Qi-obtaining classification result.

[0011] The memory update module is used to update the channel connection weights based on the linkage frequency of channel pairs in the target electrical stimulation channel set after detecting the Qi state, and generate a stimulation pathway memory map by using the Hebbian learning rules.

[0012] The output control module is used to generate a set of stimulation parameters for the next stimulation cycle based on the stimulation pathway memory map and the results of the Qi acquisition grading.

[0013] The closed-loop scheduling module is used to generate a closed-loop electrical stimulation output strategy by combining the current set of stimulation parameters, the set of target electrical stimulation channels, and the evolution trend of the stimulation pathway memory map.

[0014] Optionally, modules can be integrated using the following methods:

[0015] Raw electrophysiological signals from multiple electrical stimulation channels were acquired and preprocessed to generate a temporal response matrix for each electrical stimulation channel.

[0016] Based on the K-WTA mechanism, a multi-cycle sparse screening process is performed on the temporal response matrix of electrical stimulation channels to select a set of target electrical stimulation channels that remain continuously activated in multiple stimulation cycles.

[0017] The changes in skin electrical signals, electromyographic response amplitude, and stimulation response delay corresponding to each electrical stimulation channel in the target electrical stimulation channel set were collected, and a nonlinear response model between the electrical stimulation channel and the individual neural response was constructed to form an individual response atlas.

[0018] Based on the activation synchronicity and feedback change trend between electrical stimulation channels in the individual response spectrum, it is determined whether the current state of obtaining qi has been reached, and the intensity level of obtaining qi is generated.

[0019] When the state of obtaining qi is determined, the linkage frequency of electrical stimulation channel pairs in the target electrical stimulation channel set is statistically analyzed, and the channel connection weights are updated based on the Hebbian rule to construct an electrical stimulation pathway memory map.

[0020] Based on the electrical stimulation pathway memory map and the intensity level of the obtained qi, a set of electrical stimulation parameters consisting of current intensity, stimulation frequency, waveform type and pulse width is generated;

[0021] An electrical stimulation output strategy is generated based on the set of electrical stimulation parameters, the set of target electrical stimulation channels, and the weight change trend in the electrical stimulation pathway memory map.

[0022] Optionally, the preprocessing includes bandpass filtering, Hilbert transform envelope extraction, amplitude normalization, median filtering for noise reduction, and time series alignment.

[0023] Optionally, the selection of the target electrical stimulation channel set includes:

[0024] Within each stimulation cycle, the response amplitudes of all electrical stimulation channels are collected, and the response vectors of the electrical stimulation channels are constructed. The response vectors of the electrical stimulation channels in multiple consecutive cycles are arranged in chronological order to form the temporal response matrix of the electrical stimulation channels.

[0025] The electrical stimulation channel response vector corresponding to each period in the electrical stimulation channel time-series response matrix is ​​input into the K-WTA mechanism, and the response amplitude is sorted in descending order. The top K electrical stimulation channels with the highest response amplitude are selected, and the remaining electrical stimulation channels are set to zero, resulting in a sparse electrical stimulation channel time-series response matrix.

[0026] With a sliding time window period of T, the temporal response matrix of the sparse electrical stimulation channels is cumulatively statistically analyzed, and the number of times each electrical stimulation channel is selected by the K-WTA mechanism in T consecutive periods is recorded, denoted as n. i , where i is the electrical stimulation channel number;

[0027] Based on statistical results, the activation ratio S of each electrical stimulation channel within the sliding time window is calculated. i =n i / T, set the activation ratio threshold θ∈[0,1], when the activation ratio S i When ≥θ, the corresponding electrical stimulation channel is marked as a stable activation channel;

[0028] All electrical stimulation channels that meet the activation ratio threshold are grouped into a target electrical stimulation channel set, and the target electrical stimulation channel set and the sparsed electrical stimulation channel temporal response matrix are output.

[0029] Optionally, forming the individual response profile includes:

[0030] The changes in skin electrical signal amplitude, electromyographic response amplitude, and neural response delay of each electrical stimulation channel in the target electrical stimulation channel set during the electrical stimulation cycle are collected. These three factors together constitute the individual neural response of that electrical stimulation channel.

[0031] The changes in skin electrical signal amplitude, electromyographic response amplitude, and neural response delay are organized according to the electrical stimulation channel number to construct an individual neural response vector at the electrical stimulation channel level.

[0032] The current intensity, stimulation frequency, and pulse width applied to each electrical stimulation channel in the current cycle are collected to form a corresponding electrical stimulation parameter vector, which is then combined with the individual neural response vector at the electrical stimulation channel level to generate a parameter comparison structure between the channel-level stimulation input and the individual neural response.

[0033] Based on preset weighting coefficients, the three indicators in the individual neural response are standardized and the scores of the indicators are calculated separately. The scores are then superimposed to generate a nonlinear response score for each electrical stimulation channel.

[0034] The numbers of all electrical stimulation channels in the target electrical stimulation channel set are paired with nonlinear response scores to form a nonlinear response matrix;

[0035] Based on the nonlinear response matrix, the relative magnitude, spatial distribution relationship and physiological location mapping between channels of the nonlinear response score values ​​are set. The score matrix is ​​transformed into a spatial response heatmap, and a three-dimensional channel mapping structure including the response amplitude layer, the neural delay layer and the physiological mapping layer is constructed.

[0036] The three-dimensional channel mapping structure is defined as an individual response map.

[0037] Optionally, the generated gas intensity level includes:

[0038] Extract the nonlinear response score values ​​corresponding to the target electrical stimulation channel set in the individual response map to form the nonlinear response score vector of the target electrical stimulation channel set;

[0039] Based on the nonlinear response score vector of the target electrical stimulation channel set, pairwise score correlation calculation is performed between all electrical stimulation channels to obtain the electrical stimulation channel score correlation coefficient matrix, and the mean of the electrical stimulation channel score correlation coefficient matrix is ​​calculated. The mean is defined as the activation synchronicity index.

[0040] The temporal changes of individual neural response vectors at the electrical stimulation channel level are extracted from individual response maps, and a feedback change trend vector is constructed for each electrical stimulation channel. The feedback change trend vector includes the change trend of skin electrodermal signal amplitude, the change trend of electromyographic response amplitude, and the change trend of neural response delay.

[0041] The feedback trend vectors of all electrical stimulation channels are statistically aggregated. If there are multiple electrical stimulation channels that simultaneously satisfy the trends of increased skin electromyography signal amplitude, enhanced electromyography response, and shortened neural response delay in multiple consecutive stimulation cycles, it is determined that there is an individual state of enhanced neural response feedback.

[0042] When the activation synchronicity index is higher than the activation synchronicity threshold and there is an enhanced state of individual neural response feedback, the target electrical stimulation channel set is determined to have reached the state of obtaining qi.

[0043] The activation synchronicity index and the duration of the enhanced state of individual neural response feedback are input into a preset Qi-obtaining level evaluation function to generate a Qi-obtaining intensity level.

[0044] Optionally, constructing a memory map of electrical stimulation pathways includes:

[0045] The input neuron activation signal generated by each electrical stimulation channel in the target electrical stimulation channel set during the current stimulation cycle is defined as the electrical stimulation input encoded signal x. i , where i is the electrical stimulation channel number;

[0046] Receive the output neuron activation signal y of each electrical stimulation channel in the target electrical stimulation channel set during the current cycle. j This constitutes the output response signal of the electrical stimulation channel, where j is the electrical stimulation channel number;

[0047] Collect the multi-source physiological feedback signal vector f = {f1, f2, ..., f} corresponding to each electrical stimulation channel. k The physiological feedback signal includes the amplitude of the skin electrodermal signal, the amplitude of the electromyographic response, and the subjective qi-delivery score, and the feedback modulation factor α(f)∈[0,1] is calculated;

[0048] The connection weights of the channel pairs are updated using the Hebbian learning rule, and the update formula is as follows:

[0049] Δw ij =η·α(f)·x i ·y j ;

[0050] Where η∈(0,1) is the preset learning rate, x i y j These are the current cycle activation values ​​of the input and output neurons, respectively.

[0051] The linkage frequency of all channel pairs in the target electrical stimulation channel set within M consecutive stimulation cycles is statistically analyzed. The linkage frequency refers to the sum of the cycle counts in which the channel pairs simultaneously satisfy the K-WTA activation and gas-generating state conditions within the same cycle.

[0052] Whenever a pair of channels is selected as an active channel by the K-WTA mechanism in the same cycle, and both are above the current cycle's gas intensity level threshold, it is considered a linkage event.

[0053] The cumulative connection weights, linkage frequencies, and spatial topological relationships of each channel pair are integrated to form a directed graph structure. Nodes represent electrical stimulation channels, edges represent memory connections, edge weights are connection weights, and edge frequencies are linkage frequencies.

[0054] The above directed graph structure is defined as an electrical stimulation pathway memory map, which represents the long-term linkage memory and stimulation effect coupling between channels in the target electrical stimulation channel set.

[0055] Optionally, the generated set of electrical stimulation parameters includes:

[0056] Based on the connection weights of electrical stimulation channel pairs in the electrical stimulation pathway memory map and their changing trends over multiple consecutive stimulation cycles, the pathway activation intensity of the electrical stimulation channel pairs is calculated. The pathway activation intensity and the intensity level of the obtained qi are substituted into the electrical stimulation parameter adjustment function as input variables. The electrical stimulation parameter adjustment function is used to output the optimal combination of current intensity, stimulation frequency, waveform type and pulse width for each target electrical stimulation channel. Based on the number of pathways and the sum of connection weights in which each target electrical stimulation channel participates in the electrical stimulation pathway memory map, the output optimal combination is amplitude-weighted to form an electrical stimulation parameter set.

[0057] Optionally, the strategy for generating electrical stimulation output includes:

[0058] Extract the numerical change sequence of the connection weight of each electrical stimulation channel pair in the electrical stimulation pathway memory map over multiple consecutive stimulation cycles, calculate the mean difference, range amplitude, and direction of variation of the numerical change sequence, and construct the feature vector of the trend of connection weight change of electrical stimulation channel pair.

[0059] Based on the weighted superposition of the characteristic vectors of the changing trends of the connection weights of all electrical stimulation channels involved in the electrical stimulation pathway memory map for each target electrical stimulation channel, the connection stability index and response modulation tendency index of the target electrical stimulation channel are obtained.

[0060] The connection stability index and response adjustment tendency index are combined with the set of electrical stimulation parameters and the intensity level of the target electrical stimulation channel, and input to the electrical stimulation output strategy generation function. The electrical stimulation output strategy generation function dynamically generates the output activation status and activation sequence of different target electrical stimulation channels based on the channel stability, the intensity of the qi level and the gradient of parameter changes.

[0061] Based on the generated output activation status and activation sequence, an electrical stimulation output strategy is generated.

[0062] The beneficial effects of this invention are:

[0063] (1) By constructing a temporal response matrix of electrical stimulation channels and introducing a multi-period sparse screening mechanism, this invention can dynamically screen out a set of target electrical stimulation channels that are continuously activated in multiple stimulation cycles, which significantly improves the stability of stimulation localization and the accuracy of channel selection. The introduced K-WTA mechanism and activation ratio screening strategy ensure that the channel response has stable periodicity and spatial distribution consistency.

[0064] (2) This invention establishes a nonlinear response model by collecting multidimensional individual neural response indicators such as skin electrical signals, electromyographic response and neural delay, and further forms an individual response map, realizing the quantitative assessment and visualization of individual differences in electrical stimulation channels. The map constructs a channel response hierarchical structure through a three-dimensional heat map, which significantly improves the scientific nature of individualized parameter regulation.

[0065] (3) Based on the individual response map, this invention extracts the activation synchronicity and feedback change trend, establishes an objective judgment mechanism for the Qi state, combines the dynamic change trend of weight in the electrical stimulation pathway memory map, introduces an improved Hebbian rule to construct a physiological feedback-driven connection regulation algorithm, so that the electrical stimulation pathway has the ability to learn linkage frequency and memory plasticity, and improves the sensitivity of Qi state recognition and the rationality of strategy generation.

[0066] (4) This invention integrates the intensity level of Qi, the weight trend of the electrical stimulation pathway memory map and the stimulation parameters of the target channel to construct a parameter adjustment function, generate a dynamically adaptive set of electrical stimulation parameters and an electrical stimulation output strategy, and achieve precise control over the output sequence, start-stop rhythm and parameter update rhythm, thereby improving the clinical adaptability and efficacy consistency during the electrical stimulation process. Attached Figure Description

[0067] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0068] Figure 1 This is a schematic diagram of the control system of an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation, as proposed in this invention. Detailed Implementation

[0069] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0070] refer to Figure 1 A control system for an electrical stimulation therapy device used in acupuncture to promote blood circulation and improve brain function includes the following steps:

[0071] The signal encoding module is used to acquire raw electrophysiological signals from multiple electrical stimulation channels, perform preprocessing, and generate a temporal response matrix for the electrical stimulation channels.

[0072] The target channel screening module is used to apply the K-WTA mechanism to perform multi-cycle sparse screening on the electrical stimulation channel temporal response matrix to dynamically select a set of target electrical stimulation channels with continuous activation characteristics.

[0073] The sensory feedback modeling module is used to combine the skin conductance changes and electromyographic response delays corresponding to each electrical stimulation channel in the target electrical stimulation channel set to construct a nonlinear response model between the electrical stimulation channel and the individual neural response.

[0074] The Qi-obtaining identification module is used to analyze the temporal characteristics of multi-channel synchronous activation in the nonlinear response model, and to determine whether the current state of Qi-obtaining is entered based on preset criteria, and output the Qi-obtaining classification result.

[0075] The memory update module is used to update the channel connection weights based on the linkage frequency of channel pairs in the target electrical stimulation channel set after detecting the Qi state, and generate a stimulation pathway memory map by using the Hebbian learning rules.

[0076] The output control module is used to generate a set of stimulation parameters for the next stimulation cycle based on the stimulation pathway memory map and the results of the Qi acquisition grading.

[0077] The closed-loop scheduling module is used to generate a closed-loop electrical stimulation output strategy by combining the current set of stimulation parameters, the set of target electrical stimulation channels, and the evolution trend of the stimulation pathway memory map.

[0078] In this embodiment, the modules are interconnected using the following method:

[0079] Raw electrophysiological signals from multiple electrical stimulation channels were acquired and preprocessed to generate a temporal response matrix for each electrical stimulation channel.

[0080] Based on the K-WTA mechanism, a multi-cycle sparse screening process is performed on the temporal response matrix of electrical stimulation channels to select a set of target electrical stimulation channels that remain continuously activated in multiple stimulation cycles.

[0081] The changes in skin electrical signals, electromyographic response amplitude, and stimulation response delay corresponding to each electrical stimulation channel in the target electrical stimulation channel set were collected, and a nonlinear response model between the electrical stimulation channel and the individual neural response was constructed to form an individual response atlas.

[0082] Based on the activation synchronicity and feedback change trend between electrical stimulation channels in the individual response spectrum, it is determined whether the current state of obtaining qi has been reached, and the intensity level of obtaining qi is generated.

[0083] When the state of obtaining qi is determined, the linkage frequency of electrical stimulation channel pairs in the target electrical stimulation channel set is statistically analyzed, and the channel connection weights are updated based on the Hebbian rule to construct an electrical stimulation pathway memory map.

[0084] Based on the electrical stimulation pathway memory map and the intensity level of the obtained qi, a set of electrical stimulation parameters consisting of current intensity, stimulation frequency, waveform type and pulse width is generated;

[0085] An electrical stimulation output strategy is generated based on the set of electrical stimulation parameters, the set of target electrical stimulation channels, and the weight change trend in the electrical stimulation pathway memory map.

[0086] In this embodiment, the preprocessing includes bandpass filtering, Hilbert transform envelope extraction, amplitude normalization, median filtering for noise reduction, and time series alignment.

[0087] In this embodiment, the selection of the target electrical stimulation channel set includes:

[0088] Within each stimulation cycle, the response amplitudes of all electrical stimulation channels are collected, and the response vectors of the electrical stimulation channels are constructed. The response vectors of the electrical stimulation channels in multiple consecutive cycles are arranged in chronological order to form the temporal response matrix of the electrical stimulation channels.

[0089] The electrical stimulation channel response vector corresponding to each period in the electrical stimulation channel time-series response matrix is ​​input into the K-WTA mechanism, and the response amplitude is sorted in descending order. The top K electrical stimulation channels with the highest response amplitude are selected, and the remaining electrical stimulation channels are set to zero, resulting in a sparse electrical stimulation channel time-series response matrix.

[0090] With a sliding time window period of T, the temporal response matrix of the sparse electrical stimulation channels is cumulatively statistically analyzed, and the number of times each electrical stimulation channel is selected by the K-WTA mechanism in T consecutive periods is recorded, denoted as n. i , where i is the electrical stimulation channel number;

[0091] Based on statistical results, the activation ratio S of each electrical stimulation channel within the sliding time window is calculated. i =n i / T, set the activation ratio threshold θ∈[0,1], when the activation ratio S i When ≥θ, the corresponding electrical stimulation channel is marked as a stable activation channel;

[0092] All electrical stimulation channels that meet the activation ratio threshold are combined into a target electrical stimulation channel set, and the target electrical stimulation channel set and the sparsed electrical stimulation channel temporal response matrix are output.

[0093] The electrical stimulation channel that meets the activation ratio threshold is defined as follows: within a preset sliding time window T, if the ratio S of the total number of cycles selected by the K-WTA mechanism for a certain electrical stimulation channel to T is [value missing]. i Satisfying the activation ratio threshold S i If the corresponding electrical stimulation channel is selected as the activation channel and there are at least two non-continuous cycles in all cycles, and the corresponding response amplitude is ranked in the top 50% of the top K among all electrical stimulation channels, then the electrical stimulation channel is considered to have stable response characteristics in multiple cycles. All electrical stimulation channels that simultaneously meet the above activation ratio threshold condition and response ranking stability condition constitute the target electrical stimulation channel set.

[0094] In this embodiment, forming an individual response profile includes:

[0095] The changes in skin electrical signal amplitude, electromyographic response amplitude, and neural response delay of each electrical stimulation channel in the target electrical stimulation channel set during the electrical stimulation cycle are collected. These three factors together constitute the individual neural response of the electrical stimulation channel. The change in skin electrical signal amplitude is the potential change amplitude of the skin area corresponding to the electrical stimulation channel. The electromyographic response amplitude is the electromyographic activation intensity generated by the muscle group corresponding to the electrical stimulation channel. The neural response delay is the time interval from the start of electrical stimulation to the first occurrence of any of the aforementioned physiological feedbacks.

[0096] The changes in the amplitude of the skin electrical signal, the amplitude of the electromyographic response, and the delay of the neural response are organized according to the electrical stimulation channel number to construct an electrical stimulation channel-level individual neural response vector, with each electrical stimulation channel corresponding to an electrical stimulation channel-level individual neural response vector;

[0097] The current intensity, stimulation frequency, and pulse width applied to each electrical stimulation channel in the current cycle are collected to form a corresponding electrical stimulation parameter vector, which is then combined with the individual neural response vector at the electrical stimulation channel level to generate a parameter comparison structure between the channel-level stimulation input and the individual neural response.

[0098] Based on preset weighting coefficients, the three indicators in the individual neural response are standardized and the scores of the indicators are calculated separately. After being superimposed, a nonlinear response score value for each electrical stimulation channel is generated. The nonlinear response score is used to quantify the intensity of the individual neural response of the electrical stimulation channel under the current stimulation parameters.

[0099] The numbers of all electrical stimulation channels in the target electrical stimulation channel set are paired with nonlinear response scores to form a nonlinear response matrix;

[0100] Based on the nonlinear response matrix, the relative magnitude, spatial distribution relationship and physiological location mapping between channels of the nonlinear response score values ​​are set. The score matrix is ​​transformed into a spatial response heatmap, and a three-dimensional channel mapping structure including the response amplitude layer, the neural delay layer and the physiological mapping layer is constructed.

[0101] The three-dimensional channel mapping structure is defined as an individual response map. The individual response map is based on multi-channel stimulation response and reflects the intensity, distribution and temporal characteristics of individual neural responses under specific parameter combinations of different electrical stimulation channels.

[0102] The preset weight coefficients are determined through parameter response correlation analysis based on multiple sets of clinically collected samples. Statistical modeling is performed on the changes in skin electrical signal amplitude, electromyographic response amplitude, and nerve response delay under different electrical stimulation parameters during the Tongnao Huoluo acupuncture treatment. Pearson correlation coefficients and time window gradients between various physiological feedback indicators and treatment effects are calculated. Based on their sensitivity to the "deqi" state, they are standardized and normalized. Combined with expert experience scoring, hierarchical analysis is performed for comprehensive weighting, ultimately yielding a set of preset weight coefficients for the nonlinear response scoring function.

[0103] In this embodiment, the generated gas intensity level includes:

[0104] Extract the nonlinear response score values ​​corresponding to the target electrical stimulation channel set in the individual response map to form the nonlinear response score vector of the target electrical stimulation channel set;

[0105] Based on the nonlinear response score vector of the target electrical stimulation channel set, pairwise score correlation calculation is performed between all electrical stimulation channels to obtain the electrical stimulation channel score correlation coefficient matrix, and the mean of the electrical stimulation channel score correlation coefficient matrix is ​​calculated. The mean is defined as the activation synchronicity index.

[0106] The temporal changes of individual neural response vectors at the electrical stimulation channel level are extracted from individual response maps, and a feedback change trend vector is constructed for each electrical stimulation channel. The feedback change trend vector includes the change trend of skin electrodermal signal amplitude, the change trend of electromyographic response amplitude, and the change trend of neural response delay.

[0107] The feedback trend vectors of all electrical stimulation channels are statistically aggregated. If there are multiple electrical stimulation channels that simultaneously satisfy the trends of increased skin electromyography signal amplitude, enhanced electromyography response, and shortened neural response delay in multiple consecutive stimulation cycles, it is determined that there is an individual state of enhanced neural response feedback.

[0108] When the activation synchronicity index is higher than the activation synchronicity threshold and there is an enhanced state of individual neural response feedback, the target electrical stimulation channel set is determined to have reached the state of obtaining qi.

[0109] The activation synchronicity index and the duration of the enhanced state of individual neural response feedback are input into a preset Qi-de-qi level evaluation function to generate a Qi-de-qi intensity level. The Qi-de-qi intensity level is used to characterize the individualized Qi-de-qi degree of the target electrical stimulation channel set under the current electrical stimulation parameter combination.

[0110] The activation synchronicity threshold is a dynamic discrimination criterion used to judge the consistency of activation state within the target electrical stimulation channel set. Specifically, it is constructed as a historical distribution reference value based on the average correlation coefficient of the channel response scores in multiple past stimulation cycles of the target electrical stimulation channel set. The activation synchronicity threshold is dynamically calculated using the quantile method. The average correlation coefficient of all channels response scores in the P consecutive cycles before the target cycle constitutes a historical score sequence. The historical score sequence is sorted from smallest to largest, and the value corresponding to its Q percentile is taken as the current activation synchronicity threshold, where P is the window length, Q is the quantile setting value, and P and Q are both adjustable system parameters.

[0111] The preset Qi-attainment level evaluation function is a multi-parameter piecewise scoring function used to generate Qi-attainment intensity levels. Its inputs are the activation synchronicity index and the duration of the individual neural response feedback enhancement state, and its output is the Qi-attainment intensity level. The Qi-attainment level evaluation function establishes a two-dimensional scoring coordinate system based on the two input indices. The scoring interval for the activation synchronicity index is set to [0.0, 1.0], divided into 5 level segments, corresponding to very weak, weak, medium, strong, and very strong synchronicity levels, respectively. The scoring interval for the duration of the feedback enhancement state is set to [1 cycle, T cycle], where T is the sliding window length, divided into 3 level segments, corresponding to short-term, medium-term, and long-term response duration levels, respectively. Finally, the above two scoring dimensions are fused using a weighted decision mechanism. Each dimension is assigned weight factors w1 and w2, with w1 = 0.6 and w2 = 0.4 by default. The results are mapped to five levels of Qi-attainment intensity level labels through a two-dimensional matrix, and the output level values ​​are L1 to L5, used to quantify the Qi-attainment intensity of the target electrical stimulation channel set in the current cycle.

[0112] In this embodiment, constructing an electrical stimulation pathway memory map includes:

[0113] The input neuron activation signal generated by each electrical stimulation channel in the target electrical stimulation channel set during the current stimulation cycle is defined as the electrical stimulation input encoded signal x. i , where i is the electrical stimulation channel number;

[0114] Receive the output neuron activation signal y of each electrical stimulation channel in the target electrical stimulation channel set during the current cycle. j This constitutes the output response signal of the electrical stimulation channel, where j is the electrical stimulation channel number;

[0115] Collect the multi-source physiological feedback signal vector f = {f1, f2, ..., f} corresponding to each electrical stimulation channel. k The physiological feedback signal includes the amplitude of the skin electrodermal signal, the amplitude of the electromyographic response, and the subjective qi-delivery score, and the feedback modulation factor α(f)∈[0,1] is calculated as follows:

[0116]

[0117] The sigmoid function is defined as follows: w k The normalized weighting coefficients for the physiological feedback channel are set based on the inverse of the historical statistical average, f. k This represents the sampled value of the k-th feedback indicator in the current period;

[0118] This formula originates from the activation function mechanism in classic neural networks, belonging to the category of nonlinear transformations in mathematics. The sigmoid function is often used to simulate the nonlinear response of neurons to weighted inputs, and is specifically defined as follows: Multiple physiological feedback signals f k (e.g., skin conductance, electromyography, subjective scores, etc.) are weighted by a coefficient w. k Linear combination yields the total strength index of the feedback signal. The feedback modulation factor α(f) is then normalized using the sigmoid function to fall within the interval [0,1], and is used to dynamically regulate the enhancement amplitude in the Hebbian learning rule. Unlike traditional neural networks that use sigmoid in the output or hidden layers, this application innovatively uses it to construct a physiological feedback modulation factor, endowing the Hebbian synaptic update mechanism with physiologically interpretable modulation capabilities, thereby constituting a neural learning mechanism with individual adaptive modulation capabilities;

[0119] The feedback modulation factor α(f) serves as a weight to adjust the amplitude of Hebbian synaptic enhancement, and is used to determine the degree of enhancement or inhibition when updating the weights;

[0120] The connection weights of the channel pairs are updated using the Hebbian learning rule, and the update formula is as follows:

[0121] Δw ij =η·α(f)·x i ·y j ;

[0122] Where η∈(0,1) is the preset learning rate, x i y j These are the current cycle activation values ​​of the input and output neurons, respectively.

[0123] This formula is an improved derivation based on the classic Hebbian learning rule. The Hebbian rule originates from the principle of synaptic plasticity in neuroscience, and its basic form is Δw. ij =η·x i ·y j , used to describe the input neuron activation signal x i With the activation signal y of the output neuron j Simultaneously, when active, the connection weight w is adjusted. ij The enhancement effect is achieved by η, where η is the learning rate control factor. Based on this, this application introduces a feedback modulation factor α(f) to quantify the regulatory effect of multi-source physiological feedback signals on synaptic update intensity, thereby constructing a more individualized and adaptive synaptic learning mechanism. The introduction of the feedback modulation factor α(f) is based on psychological and neurofeedback regulation models. Its mathematical configuration uses the sigmoid function to compress and map the weighted sum of multi-dimensional physiological feedback, ensuring its value range is within [0,1], thus serving as a differentiable dynamic adjustment factor to control whether Hebbian learning is amplified or inhibited. Therefore, the original formula is extended to a form with feedback regulation weights, reflecting an innovative modeling of the behavior shaping process of Hebbian learning driven by physiological feedback. This derivation process maintains the core linkage product structure of the original formula while achieving controllable adjustment of the learning modulation sensitivity;

[0124] The linkage frequency of all channel pairs in the target electrical stimulation channel set within M consecutive stimulation cycles is statistically analyzed. The linkage frequency refers to the sum of the cycle counts in which the channel pairs simultaneously satisfy the K-WTA activation and gas-generating state conditions within the same cycle.

[0125] Whenever a pair of channels is selected as an active channel by the K-WTA mechanism in the same cycle, and both are above the current cycle's gas intensity level threshold, it is considered a linkage event.

[0126] The cumulative connection weights, linkage frequencies, and spatial topological relationships of each channel pair are integrated to form a directed graph structure. Nodes represent electrical stimulation channels, edges represent memory connections, edge weights are connection weights, and edge frequencies are linkage frequencies.

[0127] The above directed graph structure is defined as an electrical stimulation pathway memory map, which represents the long-term linkage memory and stimulation effect coupling between channels in the target electrical stimulation channel set.

[0128] In this embodiment, the set of generated electrical stimulation parameters includes:

[0129] Based on the connection weights of electrical stimulation channel pairs in the electrical stimulation pathway memory map and their changing trends over multiple consecutive stimulation cycles, the pathway activation intensity of the electrical stimulation channel pairs is calculated. This pathway activation intensity and the intensity level of the obtained qi are used as input variables and substituted into the electrical stimulation parameter adjustment function. This function outputs the optimal combination of current intensity, stimulation frequency, waveform type, and pulse width for each target electrical stimulation channel. Based on the number of pathways and the sum of connection weights each target electrical stimulation channel participates in in the electrical stimulation pathway memory map, the optimal output combination is amplitude-weighted to form an electrical stimulation parameter set. The parameters in this electrical stimulation parameter set are then bound to the corresponding electrical stimulation channels in the target electrical stimulation channel set to constitute a complete channel-level electrical stimulation control output command.

[0130] In this embodiment, the strategy for generating electrical stimulation output includes:

[0131] Extract the numerical change sequence of the connection weight of each electrical stimulation channel pair in the electrical stimulation pathway memory map over multiple consecutive stimulation cycles, and calculate the mean difference, range amplitude, and direction of variation of the numerical change sequence to form a feature vector of the connection weight change trend of the electrical stimulation channel pair.

[0132] Based on the weighted superposition of the characteristic vectors of the changing trends of the connection weights of all electrical stimulation channels involved in the electrical stimulation pathway memory map for each target electrical stimulation channel, the connection stability index and response modulation tendency index of the target electrical stimulation channel are obtained.

[0133] The connection stability index and response adjustment tendency index are combined with the set of electrical stimulation parameters and the intensity level of the target electrical stimulation channel, and input to the electrical stimulation output strategy generation function. The electrical stimulation output strategy generation function dynamically generates the output activation status and activation sequence of different target electrical stimulation channels based on the channel stability, the intensity of the qi level and the gradient of parameter changes.

[0134] Based on the generated output activation status and activation sequence, an electrical stimulation output strategy is generated. The electrical stimulation output strategy includes the start and stop timing of each channel in the target electrical stimulation channel set during the stimulation cycle, the electrical stimulation parameter switching logic, and the regulation cycle rhythm.

[0135] Example 1:

[0136] To verify the feasibility of this invention in practice, it was applied to the daily treatment process of the neurorehabilitation outpatient department of a hospital. This department mainly treats patients with central nervous system diseases such as hemiplegia, facial paralysis, and chronic neuropathic pain after stroke. The degree of neurological function impairment varies significantly among patients. Traditional electroacupuncture treatment relies entirely on the physician's experience in terms of stimulation intensity, frequency, and location, making it impossible to achieve precise individual stimulation control. This results in problems such as low treatment efficiency, strong stimulation discomfort, and low Qi attainment rate.

[0137] In this application scenario, the hospital selected 80 outpatients over a certain period as clinical trial subjects, with an average age of 56.7 years and a gender ratio of 44 males and 36 females. They were randomly divided into an experimental group (40 patients) and a control group (40 patients). The experimental group received brain-clearing and blood-activating acupuncture treatment using the control method described in this invention in conjunction with a multi-channel electrical stimulation therapy device, while the control group received conventional treatment using a traditional manually adjustable electrical stimulation device. The treatment plan was set at 30 minutes per session, 3 times per week, for 4 weeks. All treatments were performed by experienced senior rehabilitation therapists to ensure consistency in operation.

[0138] In the experimental group's treatment process, the signal encoding module was first used to collect skin electrical signals, electromyographic feedback, and neural delay signals in real time for each electrical stimulation channel, constructing a standardized temporal response matrix for each channel. Subsequently, the K-WTA mechanism and sliding window analysis method were applied to perform sparse filtering, dynamically identifying the set of channels with high activation stability in the current cycle. Individual physiological feedback information was collected to construct a nonlinear response model, generating individual response maps, enabling spatial thermodynamic analysis and visualization of the patient's response to each electrical stimulation channel.

[0139] The "Deqi" (arrival of Qi) recognition module automatically identifies the "Deqi" state and outputs a quantitative classification based on channel scoring synchronization and feedback enhancement trends. If "Deqi" is determined, the system activates a memory map update function based on an improved Hebbian mechanism, using feedback modulation factors to enhance the synaptic connection weights between truly effective channels. Simultaneously, it extracts multi-cycle connection weight change trends to form an electrical stimulation pathway memory map, and dynamically adjusts parameters such as current intensity and pulse width to generate new stimulation strategies. Finally, the system combines channel stability indicators, "Deqi" levels, and output strategy functions to construct a closed-loop control of the electrical stimulation output rhythm, forming an individualized and evolvable multi-channel electrical stimulation scheme.

[0140] To quantitatively evaluate the clinical efficacy and application value of the system, the treatment effects, stimulation efficiency, parameter adjustment efficiency, and subjective stimulation comfort were compared between the two groups of patients. The statistical indicators are shown in the table below:

[0141] Table 1. Evaluation of the Clinical Application Effect of the Control Method of the Brain-Activating Electrical Stimulation Therapy Instrument

[0142]

[0143]

[0144] The data analysis results in Table 1 show that the control method of the brain-activating acupuncture electrostimulation therapy device described in this invention is significantly superior to traditional treatment methods in several key indicators. Firstly, in terms of neurological function improvement, the experimental group showed an average increase of 13.2 points in Fugl-Meyer scores, far exceeding the 6.8 points in the control group, indicating that this invention has a stronger neuromodulation effect in restoring patients' motor function. Secondly, the incidence of "deqi" (a sensation of sensation) was as high as 85.0% in the experimental group, compared to 47.5% in the control group, demonstrating that the "deqi" recognition mechanism based on multi-source feedback recognition has higher recognition sensitivity and stability in clinical practice. Regarding parameter adjustment efficiency, the experimental group required an average of only 1.2 adjustments per treatment, compared to 3.1 adjustments in the control group, significantly reducing the frequency of manual intervention and demonstrating the efficiency and intelligence of the closed-loop control algorithm. Patient complaints of subjective discomfort also decreased significantly; only 2 patients in the experimental group complained of discomfort, compared to 9 in the control group, indicating a clear advantage in stimulation comfort. Furthermore, the experimental group activated an average of 7.4 electrostimulation channels per treatment, nearly twice that of the control group, proving that the system has higher stimulation coverage and more comprehensively activates neural response areas. Finally, regarding system stability, the experimental group's control system operated without faults, while the control group's failure rate reached 5%, further highlighting the superior performance of this invention in terms of equipment control, stability, and treatment safety. In summary, this invention achieves breakthrough improvements in individualization, precision, efficiency, and user experience.

[0145] During testing, the system demonstrated extremely high stability, with no strategy failures or output logic errors caused by electrode detachment. The overall treatment process exhibited strong adaptability, and the individual response profile could update the output logic in real time according to changes in the patient's neural sensitivity, ensuring the dynamic adaptability of the stimulation strategy.

[0146] In summary, the control method of the electrostimulation therapy device based on Tongnao Huoluo acupuncture proposed in this invention has demonstrated good individual adaptability, intelligent stimulation regulation capability and therapeutic effectiveness in clinical applications, providing a feasible path and engineering foundation for the subsequent development of Tongnao Huoluo therapy towards precision treatment and intelligent control.

[0147] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A control system for an electrical stimulation therapy device used in acupuncture for promoting blood circulation and relieving brain congestion, characterized in that, include: The signal encoding module is used to acquire raw electrophysiological signals from multiple electrical stimulation channels, perform preprocessing, and generate a temporal response matrix for the electrical stimulation channels. The target channel screening module is used to apply the K-WTA mechanism to perform multi-cycle sparse screening on the electrical stimulation channel temporal response matrix to dynamically select a set of target electrical stimulation channels with continuous activation characteristics. The sensory feedback modeling module is used to combine the skin conductance changes and electromyographic response delays corresponding to each electrical stimulation channel in the target electrical stimulation channel set to construct a nonlinear response model between the electrical stimulation channel and the individual neural response. The Qi-obtaining identification module is used to analyze the temporal characteristics of multi-channel synchronous activation in the nonlinear response model, and to determine whether the current state of Qi-obtaining is entered based on preset criteria, and output the Qi-obtaining classification result. The memory update module is used to update the channel connection weights based on the linkage frequency of channel pairs in the target electrical stimulation channel set after detecting the Qi state, and generate a stimulation pathway memory map by using the Hebbian learning rules. The output control module is used to generate a set of stimulation parameters for the next stimulation cycle based on the stimulation pathway memory map and the results of the Qi acquisition grading. The closed-loop scheduling module is used to generate a closed-loop electrical stimulation output strategy by combining the current set of stimulation parameters, the set of target electrical stimulation channels, and the evolution trend of the stimulation pathway memory map.

2. A control method for an electrical stimulation therapy device used in acupuncture for promoting blood circulation and relieving brain congestion, characterized in that, The modules are connected in the following way: Raw electrophysiological signals from multiple electrical stimulation channels were acquired and preprocessed to generate a temporal response matrix for each electrical stimulation channel. Based on the K-WTA mechanism, a multi-cycle sparse screening process is performed on the temporal response matrix of electrical stimulation channels to select a set of target electrical stimulation channels that remain continuously activated in multiple stimulation cycles. The changes in skin electrical signals, electromyographic response amplitude, and stimulation response delay corresponding to each electrical stimulation channel in the target electrical stimulation channel set were collected, and a nonlinear response model between the electrical stimulation channel and the individual neural response was constructed to form an individual response atlas. Based on the activation synchronicity and feedback change trend between electrical stimulation channels in the individual response spectrum, it is determined whether the current state of obtaining qi has been reached, and the intensity level of obtaining qi is generated. When the state of obtaining qi is determined, the linkage frequency of electrical stimulation channel pairs in the target electrical stimulation channel set is statistically analyzed, and the channel connection weights are updated based on the Hebbian rule to construct an electrical stimulation pathway memory map. Based on the electrical stimulation pathway memory map and the intensity level of the obtained qi, a set of electrical stimulation parameters consisting of current intensity, stimulation frequency, waveform type and pulse width is generated; An electrical stimulation output strategy is generated based on the set of electrical stimulation parameters, the set of target electrical stimulation channels, and the weight change trend in the electrical stimulation pathway memory map.

3. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation, as described in claim 2, is characterized in that... The preprocessing includes bandpass filtering, Hilbert transform envelope extraction, amplitude normalization, median filtering for noise reduction, and time series alignment.

4. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation according to claim 2, characterized in that, The selection of the target electrical stimulation channel set includes: Within each stimulation cycle, the response amplitudes of all electrical stimulation channels are collected, and the response vectors of the electrical stimulation channels are constructed. The response vectors of the electrical stimulation channels in multiple consecutive cycles are arranged in chronological order to form the temporal response matrix of the electrical stimulation channels. The electrical stimulation channel response vector corresponding to each period in the electrical stimulation channel time-series response matrix is ​​input into the K-WTA mechanism, and the response amplitude is sorted in descending order. The top K electrical stimulation channels with the highest response amplitude are selected, and the remaining electrical stimulation channels are set to zero, resulting in a sparse electrical stimulation channel time-series response matrix. With a sliding time window period of T, the temporal response matrix of the sparse electrical stimulation channels is cumulatively statistically analyzed, and the number of times each electrical stimulation channel is selected by the K-WTA mechanism in T consecutive periods is recorded, denoted as n. i , where i is the electrical stimulation channel number; Based on statistical results, the activation ratio S of each electrical stimulation channel within the sliding time window is calculated. i =n i / T, set the activation ratio threshold θ∈[0,1], when the activation ratio S i When ≥θ, the corresponding electrical stimulation channel is marked as a stable activation channel; All electrical stimulation channels that meet the activation ratio threshold are grouped into a target electrical stimulation channel set, and the target electrical stimulation channel set and the sparsed electrical stimulation channel temporal response matrix are output.

5. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation according to claim 2, characterized in that, The formation of the individual response profile includes: The changes in skin electrical signal amplitude, electromyographic response amplitude, and neural response delay of each electrical stimulation channel in the target electrical stimulation channel set during the electrical stimulation cycle are collected. These three factors together constitute the individual neural response of that electrical stimulation channel. The changes in skin electrical signal amplitude, electromyographic response amplitude, and neural response delay are organized according to the electrical stimulation channel number to construct an individual neural response vector at the electrical stimulation channel level. The current intensity, stimulation frequency, and pulse width applied to each electrical stimulation channel in the current cycle are collected to form a corresponding electrical stimulation parameter vector, which is then combined with the individual neural response vector at the electrical stimulation channel level to generate a parameter comparison structure between the channel-level stimulation input and the individual neural response. Based on preset weighting coefficients, the three indicators in the individual neural response are standardized, and the scores of the indicators are calculated separately. The scores are then superimposed to generate a nonlinear response score for each electrical stimulation channel. The numbers of all electrical stimulation channels in the target electrical stimulation channel set are paired with nonlinear response scores to form a nonlinear response matrix; Based on the nonlinear response matrix, the relative magnitude, spatial distribution relationship and physiological location mapping between channels of the nonlinear response score values ​​are set. The score matrix is ​​transformed into a spatial response heatmap, and a three-dimensional channel mapping structure including the response amplitude layer, the neural delay layer and the physiological mapping layer is constructed. The three-dimensional channel mapping structure is defined as an individual response map.

6. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation according to claim 2, characterized in that, The generated gas strength level includes: Extract the nonlinear response score values ​​corresponding to the target electrical stimulation channel set in the individual response map to form the nonlinear response score vector of the target electrical stimulation channel set; Based on the nonlinear response score vector of the target electrical stimulation channel set, pairwise score correlation calculation is performed between all electrical stimulation channels to obtain the electrical stimulation channel score correlation coefficient matrix, and the mean of the electrical stimulation channel score correlation coefficient matrix is ​​calculated. The mean is defined as the activation synchronicity index. The temporal changes of individual neural response vectors at the electrical stimulation channel level are extracted from individual response maps, and a feedback change trend vector is constructed for each electrical stimulation channel. The feedback change trend vector includes the change trend of skin electrodermal signal amplitude, the change trend of electromyographic response amplitude, and the change trend of neural response delay. The feedback trend vectors of all electrical stimulation channels are statistically aggregated. If there are multiple electrical stimulation channels that simultaneously satisfy the trends of increased skin electromyography signal amplitude, enhanced electromyography response, and shortened neural response delay in multiple consecutive stimulation cycles, it is determined that there is an individual state of enhanced neural response feedback. When the activation synchronicity index is higher than the activation synchronicity threshold and there is an enhanced state of individual neural response feedback, the target electrical stimulation channel set is determined to have reached the state of obtaining qi. The activation synchronicity index and the duration of the enhanced state of individual neural response feedback are input into a preset Qi-obtaining level evaluation function to generate a Qi-obtaining intensity level.

7. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation according to claim 2, characterized in that, Constructing a memory map of electrical stimulation pathways includes: The input neuron activation signal generated by each electrical stimulation channel in the target electrical stimulation channel set during the current stimulation cycle is defined as the electrical stimulation input encoded signal x. i , where i is the electrical stimulation channel number; Receive the output neuron activation signal y of each electrical stimulation channel in the target electrical stimulation channel set during the current cycle. j This constitutes the output response signal of the electrical stimulation channel, where j is the electrical stimulation channel number; Collect the multi-source physiological feedback signal vector f = {f1, f2, ..., f} corresponding to each electrical stimulation channel. k The physiological feedback signal includes the amplitude of the skin electrodermal signal, the amplitude of the electromyographic response, and the subjective qi-delivery score, and the feedback modulation factor α(f)∈[0,1] is calculated; The connection weights of the channel pairs are updated using the Hebbian learning rule, and the update formula is as follows: Δw ij =η·α(f)·x i ·y j ; Where η∈(0,1) is the preset learning rate, x i y j These are the current cycle activation values ​​of the input and output neurons, respectively. The linkage frequency of all channel pairs in the target electrical stimulation channel set within M consecutive stimulation cycles is statistically analyzed. The linkage frequency refers to the sum of the cycle counts in which the channel pairs simultaneously satisfy the K-WTA activation and gas-generating state conditions within the same cycle. Whenever a pair of channels is selected as an active channel by the K-WTA mechanism in the same cycle, and both are above the current cycle's gas intensity level threshold, it is considered a linkage event. The cumulative connection weights, linkage frequencies, and spatial topological relationships of each channel pair are integrated to form a directed graph structure. Nodes represent electrical stimulation channels, edges represent memory connections, edge weights are connection weights, and edge frequencies are linkage frequencies. The above directed graph structure is defined as an electrical stimulation pathway memory map, which represents the long-term linkage memory and stimulation effect coupling between channels in the target electrical stimulation channel set.

8. The control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation according to claim 2, characterized in that, The set of generated electrical stimulation parameters includes: Based on the connection weights of electrical stimulation channel pairs in the electrical stimulation pathway memory map and their changing trends over multiple consecutive stimulation cycles, the pathway activation intensity of the electrical stimulation channel pairs is calculated. The pathway activation intensity and the intensity level of the obtained qi are substituted into the electrical stimulation parameter adjustment function as input variables. The electrical stimulation parameter adjustment function is used to output the optimal combination of current intensity, stimulation frequency, waveform type and pulse width for each target electrical stimulation channel. Based on the number of pathways and the sum of connection weights in which each target electrical stimulation channel participates in the electrical stimulation pathway memory map, the output optimal combination is amplitude-weighted to form an electrical stimulation parameter set.

9. A control method for an electrical stimulation therapy device for acupuncture to promote blood circulation and relieve brain stagnation, as described in claim 2, characterized in that, The strategy for generating electrical stimulation output includes: Extract the numerical change sequence of the connection weight of each electrical stimulation channel pair in the electrical stimulation pathway memory map over multiple consecutive stimulation cycles, calculate the mean difference, range amplitude, and direction of variation of the numerical change sequence, and construct the feature vector of the trend of connection weight change of electrical stimulation channel pair. Based on the weighted superposition of the characteristic vectors of the changing trends of the connection weights of all electrical stimulation channels involved in the electrical stimulation pathway memory map for each target electrical stimulation channel, the connection stability index and response modulation tendency index of the target electrical stimulation channel are obtained. The connection stability index and response adjustment tendency index are combined with the set of electrical stimulation parameters and the intensity level of the target electrical stimulation channel, and input to the electrical stimulation output strategy generation function. The electrical stimulation output strategy generation function dynamically generates the output activation status and activation sequence of different target electrical stimulation channels based on the channel stability, the intensity of the qi level and the gradient of parameter changes. Based on the generated output activation status and activation sequence, an electrical stimulation output strategy is generated.

Images