Electro-acupuncture stimulation scheme personalized generation method fusing myoelectricity and brain function network state
By integrating electromyography and brain functional network states into an electroacupuncture stimulation protocol, and utilizing the brain-muscle synergistic response index and reinforcement learning algorithm, personalized parameter optimization for electroacupuncture treatment was achieved. This solved the problem of inaccurate target selection in traditional electroacupuncture treatment, and improved the safety and intelligence of the treatment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-03-17
- Publication Date
- 2026-04-14
AI Technical Summary
Existing electroacupuncture treatment methods lack objective assessment and precise target selection based on brain functional activity characteristics, and are insufficient in multimodal neural signal fusion and adaptive optimization of stimulation parameters, making it difficult to accurately predict the optimal electroacupuncture stimulation target for an individual and quantify the therapeutic effect.
By fusing electromyography (EMG) and brain functional network states, a synchronous electroacupuncture-EEG-EMG dataset was constructed. The brain-muscle synergistic response index (BMEI) was calculated, and stimulation parameters were adjusted in real time using reinforcement learning algorithms to optimize the electroacupuncture treatment strategy.
It achieves adaptive control of the electroacupuncture stimulation process, improves individual response accuracy, significantly enhances the safety and stability of treatment, and ensures the effectiveness and intelligence of electroacupuncture treatment in epilepsy and neurorehabilitation scenarios.
Smart Images
Figure CN121846531A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of neuromodulation and electroacupuncture treatment for epilepsy, specifically a method for generating personalized electroacupuncture stimulation protocols that integrates electromyography and brain functional network states. Background Technology
[0002] Epilepsy is a common chronic neurological disorder characterized by recurrent seizures caused by abnormal neuronal discharges in the brain. Long-term seizures not only lead to impairments in cognitive function, motor coordination, and emotional regulation, but may also trigger neural network remodeling and abnormal coupling of brain functional areas. While traditional drug therapy can control seizure frequency to some extent, it suffers from problems such as drug tolerance, significant individual variability, and substantial side effects. In recent years, electroacupuncture, as a non-pharmacological intervention, has been extensively studied in neuromodulation and brain disease rehabilitation, showing potential efficacy, particularly in the adjunctive treatment of epilepsy.
[0003] However, current electroacupuncture treatments largely rely on experience to set stimulation points and parameters, lacking objective assessment and precise target selection based on brain functional activity characteristics. Clinically used EEG analysis methods primarily focus on cortical electrical activity, neglecting the dynamic response characteristics of deep structures such as the amygdala and anterior cingulate cortex, which are related to emotion and seizure regulation. The amygdala plays a crucial role in the generation and propagation of epileptic seizures, and its neural response patterns directly influence the modulatory effects of electroacupuncture stimulation. Therefore, traditional methods relying solely on surface potential signals or anatomical localization are insufficient for accurately predicting the optimal electroacupuncture stimulation target and quantifying therapeutic efficacy for an individual.
[0004] Furthermore, existing technologies still have significant shortcomings in areas such as multimodal neural signal fusion, adaptive optimization of stimulation parameters, and constraints on treatment safety. Most studies lack the synergistic utilization of multi-source data such as EEG, EMG, functional magnetic resonance imaging (fMRI), and positron emission tomography (PET), and also lack real-time evaluation mechanisms for neural network response characteristics during electroacupuncture stimulation. Summary of the Invention
[0005] To address the shortcomings of existing technologies, this invention provides a method for generating personalized electroacupuncture stimulation programs that integrates electromyography and brain functional network states, thereby solving the problems mentioned in the background art.
[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for generating personalized electroacupuncture stimulation programs that integrates electromyography and brain functional network states, comprising the following steps: Step 1: Obtaining data via a high-density EEG electrode array under electroacupuncture stimulation conditions. Wave power spectral density and Wave power spectral density, EMG sensors deployed in the target muscle group to extract the root mean square value of electromyographic discharge. Brain metabolic and blood flow signal data were acquired using functional magnetic resonance imaging (fMRI); stimulation intensity, waveform type, and stimulation frequency parameters of electroacupuncture were recorded; and a synchronous acquisition dataset of electroacupuncture-EEG-EMG was constructed. Step 2: Using data from a synchronized electroacupuncture-EEG-EMG dataset, obtain the correlation coefficient between EEG power and EMG amplitude, the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex, and the rate of change in electrical power in the target brain region. Calculate the brain-muscle synergy response index (BMEI) and compare it with the brain-muscle synergy response threshold (Bth) to determine whether the brain-muscle pathway synergy is satisfactory. If satisfactory, the electroacupuncture stimulation is considered effective; if unsatisfactory, the stimulation parameters are automatically adjusted to optimize the response. Step 3: Under effective electroacupuncture stimulation, perform time-frequency analysis and waveform recognition on the EEG power time series to extract epileptiform discharge characteristics of spikes, slow waves, and sharp-slow wave complexes and locate characteristic segments; calculate the epileptiform discharge power change rate GLB and compare it with the epileptiform discharge threshold Gth to determine whether the current state is a normal synchronous discharge state. If abnormal, discharge suppression is achieved by adjusting the stimulation intensity, waveform type, and frequency; after dynamic correction and verification, stable and abnormal stimulation parameters are included in the safe stimulation parameter set. Step 4: Based on the effective state of electroacupuncture stimulation and combined with the set of safe stimulation parameters, construct the brain-muscle network state space, with the brain electrical power change rate, the root mean square value of electromyography discharge, and the epileptiform discharge power change rate as the state vector; use the brain-muscle synergistic response index (BMEI) as the reward and punishment function to establish an initial reinforcement learning model; conduct offline training, dynamically calculate the reward and punishment signals based on the BMEI change trend and optimize the policy network to obtain the first version of model M0; Step 5: Based on the first version of model M0, introduce an online learning module to continuously update the model parameters and form the second version of model M1; output the optimal stimulation intensity, waveform type and frequency strategy to generate the individual optimal electroacupuncture stimulation parameter set Popt as the long-term treatment parameter configuration.
[0007] Preferably, step one includes: S11. Under electroacupuncture stimulation, a high-density EEG electrode array is deployed on the scalp surface covering key functional areas to collect real-time cortical activity signals during the patient's rehabilitation training tasks; the collected EEG signals are then decomposed into frequency bands to obtain... Wave power spectral density and Wave power spectral density; S12. Deploy surface electromyography (EMG) sensors on the surface of the relevant muscle groups to collect muscle discharge signals in real time, and obtain the root mean square value of EMG discharge through feature extraction. ; S13. Acquire neural activity information of the amygdala and anterior cingulate cortex using functional magnetic resonance imaging (fMRI) equipment, including brain region metabolic and blood flow signal data. S14. During the electroacupuncture stimulation process, the electroacupuncture control module records stimulation parameters in real time, including stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim; through a synchronous time stamping mechanism, it is time-aligned with high-density EEG signals and electromyography (EMG) signals to construct an electroacupuncture-EEG-EMG synchronous acquisition dataset.
[0008] Preferably, step two includes: S21. Extract and centralize the synchronously acquired data from electroacupuncture, EEG, and EMG. Wave power spectral density and Wave power spectral density and root mean square value of electromyography discharge The correlation coefficient between EEG power and EMG amplitude was obtained using a cortical-muscle signal co-analysis method. ; S22. Brain region metabolic and blood flow signal data acquired simultaneously by fMRI and PET were used to obtain the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex using dynamic causal modeling (DCM) and correlation modeling methods. ; S23, based on Wave power spectral density and The power spectral density was determined by using time-series resampling and differential analysis to acquire synchronized corrected EEG power time-series data. Then, a power spectral tracking algorithm was used to calculate the power changes at different time points within the electroacupuncture stimulation cycle, obtaining the rate of change of electrical power in the target brain region. .
[0009] Preferably, step two further includes: S24. The correlation coefficient between the obtained EEG power and EMG amplitude. Changes in functional connectivity between the amygdala and the anterior cingulate cortex and the rate of change of electrical power in the target brain region After dimensionless normalization, the brain muscle synergistic response index (BMEI) was calculated and obtained. S25. By setting a pre-defined brain-muscle synergy response threshold Bth, and comparing and analyzing the brain-muscle synergy response index BMEI with the brain-muscle synergy response threshold Bth, the first assessment results are obtained, including: When the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is qualified, and the current electroacupuncture stimulation is determined to be effective. Continuous monitoring is required. When the brain-muscle synergy response index BMEI is less than the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is not up to standard, and there is a risk of delayed nerve conduction, blocked information transmission, or weakened stimulus response. This triggers the first warning instruction and generates the first strategy: automatically execute the parameter adjustment instruction to reduce the stimulus intensity and adjust the stimulus frequency, optimize and recalculate until the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth.
[0010] Preferably, step three includes: S31. Based on the time series data of EEG power under the effective state of electroacupuncture stimulation, time-frequency analysis and waveform pattern recognition technology are used to extract features from the synchronously corrected EEG signals, obtain typical waveform feature parameters of epileptiform discharges such as spikes, slow waves and spike-slow wave complexes, and construct an epileptiform discharge feature set. S32. Based on the epileptiform discharge feature set and combined with synchronously acquired high-density EEG signal data, the feature mapping and power spectrum analysis methods are used to locate and bandpass filter the epileptiform discharge feature segments; the power spectral density Pspike(t) of the epileptiform discharge segment in the EEG signal at time t is obtained by power integration calculation.
[0011] Preferably, step three further includes: S33. Power spectral density of epileptiform discharge segments in the EEG signal at time t. After dimensionless processing, the sliding time window difference algorithm is used to dynamically calculate the power spectral density change in the continuous time series and obtain the epileptiform discharge power change rate GLB. S34. By setting a preset epileptic discharge threshold Gth, and comparing the epileptiform discharge power change rate GLB with the epileptic discharge threshold Gth, the second evaluation results are obtained, including: When the rate of change of epileptiform discharge power GLB ≤ epileptiform discharge threshold Gth, the current state is determined to be normal synchronous discharge, the current electroacupuncture stimulation mode is maintained, and continuous monitoring is performed. When the rate of change of epileptiform discharge power (GLB) exceeds the epileptiform discharge threshold (Gth), the current state is determined to be an abnormal synchronous discharge state, with a risk of excessive discharge or synchronous enhancement. This triggers a second warning instruction and generates a second strategy: adjusting the stimulation intensity, waveform type, and stimulation frequency. The stimulation intensity is adjusted linearly based on the coordinated trend of the rate of change of epileptiform discharge power and the root mean square value of electromyographic discharge. The waveform type is adaptively matched based on the distribution changes of the power spectrum structure characteristics. The stimulation frequency is based on the rate of change of functional connectivity between the amygdala and the anterior cingulate cortex. Dynamic adjustment is performed; after parameter correction is completed, the power change rate is re-detected until the epileptiform discharge power change rate GLB ≤ epileptiform discharge threshold Gth. S35. After the safety constraint strategy is stably executed and no abnormal discharge is detected, the stimulation parameters are used as a safe and effective configuration to establish a set of safe stimulation parameters.
[0012] Preferably, step four includes: S41. Based on the effective state of electroacupuncture stimulation determined in step two, and combined with the safe stimulation parameter set stabilized in step three, construct the brain-muscle network state space; and set the rate of change of electrical power in the target brain region. Root mean square value of electromyography (EMG) discharge Using the epileptiform discharge power change rate GLB as the state vector st, a brain-muscle network state set is established. S42. Based on the brain-muscle synergistic response index (BMEI) as the reward and punishment function, an individualized stimulus optimization model is constructed using a reinforcement learning framework. Through a joint algorithm of deep Q-learning and policy gradient, a stimulus parameter action set at is established, including stimulus intensity Istim, waveform type Fstim, and stimulus frequency fstim. The state-action mapping space is constructed and the model parameters are initialized using the brain-muscle network state set as input, forming the initial reinforcement learning model Minit.
[0013] Preferably, step four further includes: S43. Based on the initial reinforcement learning model Minit, high-density EEG signals, electromyography (EMG) signals, and synchronously recorded stimulation parameter action data from multiple subjects under different electroacupuncture stimulation conditions were selected as training samples. The model underwent multiple rounds of offline training. During the training process, the improvement magnitude of the response under different stimulation actions was calculated based on the temporal change trend of the Brain-Muscle Coordination Response Index (BMEI). When BMEI increased after electroacupuncture stimulation, a positive reward / penalty value was assigned; when BMEI decreased, a negative reward / penalty value was assigned, prompting the model to correct the stimulation strategy. The algorithm dynamically updated the weights of the policy network and the value network accordingly, and adaptively optimized the action selection probability. After multiple rounds of offline training, the model parameters converged, forming the first version of the reinforcement learning model M0.
[0014] Preferably, step five includes: S51. Based on the first version of model M0 formed by offline training, an online learning module is introduced during real-time electroacupuncture stimulation; continuous collection of EEG and EMG signals from the subjects is performed, and the real-time BMEI value is calculated as the state input to dynamically update the model parameters; when the real-time feedback deviates from the predicted reward and punishment trend, the algorithm automatically adjusts the policy gradient direction and corrects the stimulation action output, so that the model gradually adapts to individual differences in neural response; through multiple rounds of online iterative training, a second version of reinforcement learning model M1 is formed. S52. Based on the output of model M1, an individualized stimulation strategy is generated in real time before each electroacupuncture stimulation. The strategy includes the optimal stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim, and is adaptively corrected according to the current brain-muscle network state. When brain functional connectivity is enhanced and electromyographic synchronization is improved, the parameters are maintained. When a decrease in BMEI is detected, the stimulation intensity and waveform type are automatically adjusted, and parameter combinations that can stabilize neural activity and improve synergy are selected first.
[0015] Preferably, step five further includes: S53. During the implementation of individualized stimulus strategies, continuous monitoring is required. Wave power spectral density, Wave power spectral density, root mean square value of electromyography discharge And the BMEI change trend; based on real-time feedback data, the reinforcement learning model is updated continuously, so that it can automatically accumulate experience and correct the strategy weights in subsequent training; after multiple rounds of closed-loop adaptive adjustment, an individual optimal electroacupuncture stimulation parameter set Popt is formed as a long-term treatment parameter configuration, so that the stimulation plan operates in a safe, effective and optimal dynamic balance.
[0016] This invention provides a method for personalized generation of electroacupuncture stimulation protocols that integrates electromyography and brain functional network states. It has the following beneficial effects: (1) This method for personalized generation of electroacupuncture stimulation protocols by integrating electromyography (EMG) and brain functional network status proposes the Brain-Muscle Synergistic Response Index (BMEI) by fMRI functional connectivity data, thereby achieving a quantitative assessment of the synergistic effect of the brain-muscle pathway under electroacupuncture stimulation. Compared with traditional methods that rely on subjective observation or single neural signals, this invention can determine the effectiveness of stimulation in real time and automatically adjust parameters, enabling the electroacupuncture stimulation process to have adaptive control capabilities and higher individual response accuracy.
[0017] (2) This method for personalized generation of electroacupuncture stimulation protocols that integrates electromyography and brain functional network states, through time-frequency analysis of EEG power and calculation of epileptiform discharge power change rate (GLB), enables the identification and suppression of abnormal neural discharges during electroacupuncture stimulation, and establishes a safe stimulation parameter set. This mechanism effectively avoids potential risks such as overstimulation and synchronous enhancement, and significantly improves the safety and stability of electroacupuncture treatment in epilepsy and neurorehabilitation scenarios.
[0018] (3) This method for personalized generation of electroacupuncture stimulation schemes that integrates electromyography (EMG) and brain functional network states constructs a reinforcement learning model based on the brain-muscle network state space. It uses the rate of change of EEG power, the root mean square value of EMG discharge, and GLB as state vectors, and BMEI as the reward / penalty function to achieve adaptive optimization of electroacupuncture parameters. Through a combination of offline training and online learning, the model can gradually learn and output the individual's optimal stimulation parameter set, Popt, thereby achieving truly personalized electroacupuncture treatment strategy generation.
[0019] (4) This personalized electroacupuncture stimulation protocol generation method, which integrates electromyography (EMG) and brain functional network status, establishes a closed-loop system covering the entire process from signal acquisition, effectiveness assessment, safety screening to strategy optimization by integrating multi-source data from EEG, EMG, and fMRI. This closed-loop structure can continuously monitor changes in brain functional connectivity and EMG synergy during electroacupuncture stimulation, dynamically correct model parameters, and maintain a safe, effective, and optimal balance in long-term operation, significantly improving the intelligence and self-evolutionary ability of electroacupuncture therapy. Attached Figure Description
[0020] Figure 1 This is a schematic diagram illustrating the steps of a personalized electroacupuncture stimulation scheme generation method that integrates electromyography and brain functional network states according to the present invention. Detailed Implementation
[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0022] Example 1: Please refer to Figure 1 This invention provides a method for generating personalized electroacupuncture stimulation protocols that integrates electromyography and brain functional network states, comprising the following steps: Step 1: Obtaining data via a high-density EEG electrode array under electroacupuncture stimulation conditions. Wave power spectral density and Wave power spectral density, EMG sensors deployed in the target muscle group to extract the root mean square value of electromyographic discharge. Brain metabolic and blood flow signal data were acquired using functional magnetic resonance imaging (fMRI); stimulation intensity, waveform type, and stimulation frequency parameters of electroacupuncture were recorded; and a synchronous acquisition dataset of electroacupuncture-EEG-EMG was constructed. Step 2: Using data from a synchronized electroacupuncture-EEG-EMG dataset, obtain the correlation coefficient between EEG power and EMG amplitude, the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex, and the rate of change in electrical power in the target brain region. Calculate the brain-muscle synergy response index (BMEI) and compare it with the brain-muscle synergy response threshold (Bth) to determine whether the brain-muscle pathway synergy is satisfactory. If satisfactory, the electroacupuncture stimulation is considered effective; if unsatisfactory, the stimulation parameters are automatically adjusted to optimize the response. Step 3: Under effective electroacupuncture stimulation, perform time-frequency analysis and waveform recognition on the EEG power time series to extract epileptiform discharge characteristics of spikes, slow waves, and sharp-slow wave complexes and locate characteristic segments; calculate the epileptiform discharge power change rate GLB and compare it with the epileptiform discharge threshold Gth to determine whether the current state is a normal synchronous discharge state. If abnormal, discharge suppression is achieved by adjusting the stimulation intensity, waveform type, and frequency; after dynamic correction and verification, stable and abnormal stimulation parameters are included in the safe stimulation parameter set. Step 4: Based on the effective state of electroacupuncture stimulation and combined with the set of safe stimulation parameters, construct the brain-muscle network state space, with the brain electrical power change rate, the root mean square value of electromyography discharge, and the epileptiform discharge power change rate as the state vector; use the brain-muscle synergistic response index (BMEI) as the reward and punishment function to establish an initial reinforcement learning model; conduct offline training, dynamically calculate the reward and punishment signals based on the BMEI change trend and optimize the policy network to obtain the first version of model M0; Step 5: Based on the first version of model M0, introduce an online learning module to continuously update the model parameters and form the second version of model M1; output the optimal stimulation intensity, waveform type and frequency strategy to generate the individual optimal electroacupuncture stimulation parameter set Popt as the long-term treatment parameter configuration.
[0023] In this embodiment, by fusing EEG, EMG, and brain functional network signals, a brain-muscle synergistic response index (BMEI) is constructed and combined with a reinforcement learning algorithm to achieve intelligent, adaptive, and individualized optimization of electroacupuncture stimulation parameters. This method can assess the synergy and safety of the brain-muscle pathway in real time during electroacupuncture stimulation, automatically identify the effective state of stimulation, and dynamically correct stimulation intensity, waveform, and frequency parameters. With the linkage of offline and online learning mechanisms, the system can continuously optimize and generate an individual optimal electroacupuncture stimulation parameter set (Popt), enabling the electroacupuncture treatment process to operate in a dynamic balance of safety, precision, and efficiency, significantly improving the level of intelligence and individualized treatment effects in clinical applications.
[0024] Example 2: This example is an explanation of Example 1. Please refer to the example provided. Figure 1 Specifically, step one includes: S11. Under electroacupuncture stimulation, a high-density EEG electrode array is deployed on the scalp surface covering key functional areas to collect real-time cortical activity signals during the patient's rehabilitation training tasks; the collected EEG signals are then decomposed into frequency bands to obtain... Wave power spectral density and Wave power spectral density; S12. Deploy surface electromyography (EMG) sensors on the surface of the relevant muscle groups to collect muscle discharge signals in real time, and obtain the root mean square value of EMG discharge through feature extraction. ; S13. Acquire neural activity information of the amygdala and anterior cingulate cortex using functional magnetic resonance imaging (fMRI) equipment, including brain region metabolic and blood flow signal data. S14. During the electroacupuncture stimulation process, the electroacupuncture control module records stimulation parameters in real time, including stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim; through a synchronous time stamping mechanism, it is time-aligned with high-density EEG signals and electromyography (EMG) signals to construct an electroacupuncture-EEG-EMG synchronous acquisition dataset.
[0025] In this embodiment, high-density EEG, EMG, and fMRI brain region metabolic and blood flow signals were simultaneously acquired under electroacupuncture stimulation conditions. Stimulation parameters were recorded in real time and time-aligned, constructing an electroacupuncture-EEG-EMG synchronous acquisition dataset. This technology achieves high-precision synchronous acquisition and integration of multimodal neural and muscle signals, providing a reliable data foundation for subsequent brain-muscle synergistic analysis and optimization of individualized electroacupuncture stimulation protocols, thus improving the scientific rigor and precision of treatment plan design.
[0026] Example 3: This example is an explanation of Example 1. Please refer to the provided text. Figure 1 Specifically, step two includes: S21. Extract and centralize the synchronously acquired data from electroacupuncture, EEG, and EMG. Wave power spectral density and Wave power spectral density and root mean square value of electromyography discharge The correlation coefficient between EEG power and EMG amplitude was obtained using a cortical-muscle signal co-analysis method. ; S22. Brain region metabolic and blood flow signal data acquired simultaneously by fMRI and PET were used to obtain the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex using dynamic causal modeling (DCM) and correlation modeling methods. ; S23, based on Wave power spectral density and The power spectral density was determined by using time-series resampling and differential analysis to acquire synchronized corrected EEG power time-series data. Then, a power spectral tracking algorithm was used to calculate the power changes at different time points within the electroacupuncture stimulation cycle, obtaining the rate of change of electrical power in the target brain region. .
[0027] In this embodiment, by performing brain-muscle signal synergistic analysis on synchronously acquired data from electroacupuncture, electroencephalography (EEG), and electromyography (EMG), and combining the calculation of functional connectivity change rate and brain region electrical power change rate, a quantitative assessment of brain-muscle pathway synergy is achieved. This technology can accurately determine the effectiveness of electroacupuncture stimulation and provide a scientific basis for automatically optimizing stimulation parameters, thereby improving the accuracy and controllability of individualized treatment plans.
[0028] Example 4: This example is an explanation of Example 1. Please refer to the provided text. Figure 1 Specifically, step two also includes: S24. The correlation coefficient between the obtained EEG power and EMG amplitude. Changes in functional connectivity between the amygdala and the anterior cingulate cortex and the rate of change of electrical power in the target brain region After dimensionless normalization, the Brain-Muscle Synergistic Response Index (BMEI) is calculated using the following formula: ; In the formula, w1, w2, and w3 represent weighting coefficients; The correlation coefficient between EEG power and EMG amplitude has a high weighting on the brain-muscle synergistic response index and is a key indicator that directly reflects the synchronicity and synergy of the brain-muscle pathway. The second highest weighting represents the influence of the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex on the brain-muscle synergistic response index, reflecting the driving role of brain region functional network regulation on muscle response. : Characterizes the influence of the rate of change of electrical power in the target brain region on the brain muscle synergistic response index, with a moderate weight, reflecting the contribution of local brain region electrical activity to overall synergy; By constructing the Brain-Muscle Synergistic Response Index (BMEI), which is a weighted fusion of EEG-EMG correlation, changes in brain region functional connectivity, and changes in brain region electrical power, the level of synergy of the brain-muscle pathway under electroacupuncture stimulation can be quantified, providing a scientific basis for determining the effectiveness of electroacupuncture stimulation, optimizing individualized strategies, and setting safety constraints.
[0029] S25. By setting a pre-defined brain-muscle synergy response threshold Bth, and comparing and analyzing the brain-muscle synergy response index BMEI with the brain-muscle synergy response threshold Bth, the first assessment results are obtained, including: When the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is qualified, and the current electroacupuncture stimulation is determined to be effective. Continuous monitoring is required. When the brain-muscle synergy response index BMEI is less than the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is not up to standard, and there is a risk of delayed nerve conduction, blocked information transmission, or weakened stimulus response. This triggers the first warning instruction and generates the first strategy: automatically execute the parameter adjustment instruction to reduce the stimulus intensity and adjust the stimulus frequency, optimize and recalculate until the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth.
[0030] The method for obtaining the brain-muscle synergistic response threshold Bth: Through experimental acquisition and statistical analysis of EEG and EMG signals from a large number of subjects under different electroacupuncture stimulation conditions, the synergistic variation ranges of EEG power, EMG discharge amplitude, and functional connectivity change rate were extracted. Combined with the dynamic response characteristics of the brain-muscle network and individual differences, a reasonable critical value for brain-muscle synergistic response was determined. Referring to brain-muscle function assessment standards in the field of neurorehabilitation, clinical EEG-EMG synchronicity indicators, and expert experience, this threshold was established to accurately reflect the synergistic level of the brain-muscle pathway, promptly identify risks of neural conduction delay, weakened pathway response, or abnormal synchronization, thereby ensuring the effectiveness and safety of the electroacupuncture stimulation protocol.
[0031] In this embodiment, the brain-muscle synergy response index (BMEI) is calculated and compared with a preset threshold (Bth) to achieve real-time assessment of the brain-muscle pathway synergy. When the response is insufficient, the system can automatically adjust the electroacupuncture stimulation parameters and perform iterative optimization to ensure that the brain-muscle pathway remains in an effective state, thereby improving the safety and individualized adaptability of electroacupuncture treatment.
[0032] Example 5: This example is an explanation of Example 1. Please refer to the example provided. Figure 1 Specifically, step three includes: S31. Based on the time series data of EEG power under the effective state of electroacupuncture stimulation, time-frequency analysis and waveform pattern recognition technology are used to extract features from the synchronously corrected EEG signals, obtain typical waveform feature parameters of epileptiform discharges such as spikes, slow waves and spike-slow wave complexes, and construct an epileptiform discharge feature set. S32. Based on the epileptiform discharge feature set and combined with synchronously acquired high-density EEG signal data, the feature mapping and power spectrum analysis methods are used to locate and bandpass filter the epileptiform discharge feature segments; the power spectral density Pspike(t) of the epileptiform discharge segment in the EEG signal at time t is obtained by power integration calculation.
[0033] In this embodiment, by performing time-frequency analysis and waveform recognition on the electroencephalogram (EEG) signals under effective electroacupuncture stimulation, the characteristics of epileptiform discharges are accurately extracted and their power spectral density is calculated, thereby achieving real-time localization and quantification of abnormal discharges. This provides a scientific basis for subsequent personalized stimulation regulation, thereby improving the accuracy and safety of treatment.
[0034] Example 6: This example is an explanation of Example 1. Please refer to the provided text. Figure 1 Specifically, step three also includes: S33. Power spectral density of epileptiform discharge segments in the EEG signal at time t. After dimensionless processing, a sliding time window difference algorithm is used to dynamically calculate the power spectral density change within a continuous time series, obtaining the epileptiform discharge power change rate GLB, as shown in the following formula: ; In the formula, Indicates the sampling time interval; S34. By setting a preset epileptic discharge threshold Gth, and comparing the epileptiform discharge power change rate GLB with the epileptic discharge threshold Gth, the second evaluation results are obtained, including: When the rate of change of epileptiform discharge power GLB ≤ epileptiform discharge threshold Gth, the current state is determined to be normal synchronous discharge, the current electroacupuncture stimulation mode is maintained, and continuous monitoring is performed. When the rate of change of epileptiform discharge power (GLB) exceeds the epileptiform discharge threshold (Gth), the current state is determined to be an abnormal synchronous discharge state, with a risk of excessive discharge or synchronous enhancement. This triggers a second warning instruction and generates a second strategy: adjusting the stimulation intensity, waveform type, and stimulation frequency. The stimulation intensity is adjusted linearly based on the coordinated trend of the rate of change of epileptiform discharge power and the root mean square value of electromyographic discharge. The waveform type is adaptively matched based on the distribution changes of the power spectrum structure characteristics. The stimulation frequency is based on the rate of change of functional connectivity between the amygdala and the anterior cingulate cortex. Dynamic adjustment is performed; after parameter correction is completed, the power change rate is re-detected until the epileptiform discharge power change rate GLB ≤ epileptiform discharge threshold Gth. S35. After the safety constraint strategy is stably executed and no abnormal discharge is detected, the stimulation parameters are used as a safe and effective configuration to establish a set of safe stimulation parameters.
[0035] The method for obtaining the epileptic discharge threshold Gth: Epileptiform discharge characteristic analysis and power spectrum statistics were performed on high-density EEG signals from a large number of subjects under electroacupuncture stimulation and non-stimulation conditions. The power variation amplitude distribution of spikes, slow waves, and sharp-slow-wave complexes was extracted. Combined with the dynamic regulation characteristics of brain region electrical activity and individual differences, a reasonable critical value for epileptic discharge power variation was determined. Referring to epileptic seizure risk assessment criteria, EEG power spectrum abnormality judgment indicators, and the experience of neurological experts, this threshold was formulated to accurately reflect the risk of abnormal synchronous discharge, promptly identify the possibility of excessive discharge or synchronous enhancement, and ensure the stability and safety of neural activity during electroacupuncture stimulation.
[0036] In this embodiment, by dynamically calculating the epileptiform discharge power change rate GLB and comparing it with the threshold Gth, real-time early warning and adaptive control of abnormal synchronous discharge can be achieved. The intensity, waveform and frequency of electroacupuncture stimulation can be automatically adjusted to ensure that the stimulation parameters are stably executed within a safe range, thereby significantly improving the safety and reliability of treatment.
[0037] Example 7: This example is an explanation of Example 1. Please refer to the example provided. Figure 1 Specifically, step four includes: S41. Based on the effective state of electroacupuncture stimulation determined in step two, and combined with the safe stimulation parameter set stabilized in step three, construct the brain-muscle network state space; and set the rate of change of electrical power in the target brain region. Root mean square value of electromyography (EMG) discharge Using the epileptiform discharge power change rate GLB as the state vector st, a brain-muscle network state set is established. S42. Based on the brain-muscle synergistic response index (BMEI) as the reward and punishment function, an individualized stimulus optimization model is constructed using a reinforcement learning framework. Through a joint algorithm of deep Q-learning and policy gradient, a stimulus parameter action set at is established, including stimulus intensity Istim, waveform type Fstim, and stimulus frequency fstim. The state-action mapping space is constructed and the model parameters are initialized using the brain-muscle network state set as input, forming the initial reinforcement learning model Minit.
[0038] In this embodiment, by constructing a brain-muscle network state space and building an initial reinforcement learning model based on the brain-muscle synergistic response index (BMEI), the state-action mapping and individualized optimization of electroacupuncture stimulation parameters are realized, providing an intelligent foundation for subsequent precise and dynamic adjustment and improving the pertinence and effectiveness of the treatment strategy. Example
[0039] This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 Specifically, step four also includes: S43. Based on the initial reinforcement learning model Minit, high-density EEG signals, electromyography (EMG) signals, and synchronously recorded stimulation parameter action data from multiple subjects under different electroacupuncture stimulation conditions were selected as training samples. The model underwent multiple rounds of offline training. During the training process, the improvement magnitude of the response under different stimulation actions was calculated based on the temporal change trend of the Brain-Muscle Coordination Response Index (BMEI). When BMEI increased after electroacupuncture stimulation, a positive reward / penalty value was assigned; when BMEI decreased, a negative reward / penalty value was assigned, prompting the model to correct the stimulation strategy. The algorithm dynamically updated the weights of the policy network and the value network accordingly, and adaptively optimized the action selection probability. After multiple rounds of offline training, the model parameters converged, forming the first version of the reinforcement learning model M0.
[0040] In this embodiment, reinforcement learning offline training is performed using data from multiple subjects. The brain-muscle synergistic response index (BMEI) is used to dynamically adjust the reward and punishment mechanism, thereby achieving adaptive optimization of the electroacupuncture stimulation strategy. This enables the first version of the reinforcement learning model M0 to effectively generalize among different individuals, improving the accuracy and reliability of individualized treatment plans.
[0041] Example 9: This example is an explanation of Example 1. Please refer to the provided text. Figure 1 Specifically, step five includes: S51. Based on the first version of model M0 formed by offline training, an online learning module is introduced during real-time electroacupuncture stimulation; continuous collection of EEG and EMG signals from the subjects is performed, and the real-time BMEI value is calculated as the state input to dynamically update the model parameters; when the real-time feedback deviates from the predicted reward and punishment trend, the algorithm automatically adjusts the policy gradient direction and corrects the stimulation action output, so that the model gradually adapts to individual differences in neural response; through multiple rounds of online iterative training, a second version of reinforcement learning model M1 is formed. S52. Based on the output of model M1, an individualized stimulation strategy is generated in real time before each electroacupuncture stimulation. The strategy includes the optimal stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim, and is adaptively corrected according to the current brain-muscle network state. When brain functional connectivity is enhanced and electromyographic synchronization is improved, the parameters are maintained. When a decrease in BMEI is detected, the stimulation intensity and waveform type are automatically adjusted, and parameter combinations that can stabilize neural activity and improve synergy are selected first.
[0042] In this embodiment, the online learning module collects EEG and EMG feedback in real time and dynamically updates the reinforcement learning model M1, enabling the individualized electroacupuncture stimulation strategy to adaptively adjust according to the brain-muscle network state, ensuring that the stimulation protocol maintains optimal synergy and safety at different times and among different subjects.
[0043] Example 10: This example is an explanation of Example 1. Please refer to the provided text. Figure 1 Specifically, step five also includes: S53. During the implementation of individualized stimulus strategies, continuous monitoring is required. Wave power spectral density, Wave power spectral density, root mean square value of electromyography discharge And the BMEI change trend; based on real-time feedback data, the reinforcement learning model is updated continuously, so that it can automatically accumulate experience and correct the strategy weights in subsequent training; after multiple rounds of closed-loop adaptive adjustment, an individual optimal electroacupuncture stimulation parameter set Popt is formed as a long-term treatment parameter configuration, so that the stimulation plan operates in a safe, effective and optimal dynamic balance.
[0044] In this embodiment, by continuously monitoring EEG, EMG, and the brain-muscle synergistic response index (BMEI) and continuously updating the reinforcement learning model, closed-loop adaptive adjustment is achieved, thereby forming an individual optimal electroacupuncture stimulation parameter set (Popt), which maintains a dynamic balance between safety, effectiveness, and optimality in the stimulation protocol during long-term treatment.
[0045] The threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by those skilled in the art for each set of sample data; as long as it does not affect the ratio between the parameter and the quantized value, it is acceptable.
[0046] The above formulas are all derived from software simulation using a large amount of data and are selected to be close to the actual values. The coefficients in the formulas are set by those skilled in the art according to the actual situation. The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any equivalent substitutions or changes made by those skilled in the art within the technical scope disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the protection scope of the present invention.
Claims
1. A method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states, characterized in that, Includes the following steps: Step 1: Obtaining data via a high-density EEG electrode array under electroacupuncture stimulation conditions. Wave power spectral density and Wave power spectral density, EMG sensors deployed in the target muscle group to extract the root mean square value of electromyographic discharge. Brain metabolic and blood flow signal data were acquired using functional magnetic resonance imaging (fMRI); stimulation intensity, waveform type, and stimulation frequency parameters of electroacupuncture were recorded; and a synchronous acquisition dataset of electroacupuncture-EEG-EMG was constructed. Step 2: Using data from a synchronized electroacupuncture-EEG-EMG dataset, obtain the correlation coefficient between EEG power and EMG amplitude, the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex, and the rate of change in electrical power in the target brain region. Calculate the brain-muscle synergy response index (BMEI) and compare it with the brain-muscle synergy response threshold (Bth) to determine whether the brain-muscle pathway synergy is satisfactory. If satisfactory, the electroacupuncture stimulation is considered effective; if unsatisfactory, the stimulation parameters are automatically adjusted to optimize the response. Step 3: Under effective electroacupuncture stimulation, perform time-frequency analysis and waveform recognition on the EEG power time series to extract epileptiform discharge characteristics of spikes, slow waves, and sharp-slow wave complexes and locate characteristic segments; calculate the epileptiform discharge power change rate GLB and compare it with the epileptiform discharge threshold Gth to determine whether the current state is a normal synchronous discharge state. If abnormal, discharge suppression is achieved by adjusting the stimulation intensity, waveform type, and frequency; after dynamic correction and verification, stable and abnormal stimulation parameters are included in the safe stimulation parameter set. Step 4: Based on the effective state of electroacupuncture stimulation and combined with the set of safe stimulation parameters, construct the brain-muscle network state space, with the brain electrical power change rate, the root mean square value of electromyographic discharge, and the epileptiform discharge power change rate as the state vector. The Brain-Muscle Co-responsive Index (BMEI) was used as the reward and punishment function to establish an initial reinforcement learning model. Offline training was performed, and reward and punishment signals were dynamically calculated and the policy network was optimized based on the BMEI change trend to obtain the first version of model M0. Step 5: Based on the first version of model M0, introduce an online learning module to continuously update the model parameters and form the second version of model M1; output the optimal stimulation intensity, waveform type and frequency strategy to generate the individual optimal electroacupuncture stimulation parameter set Popt as the long-term treatment parameter configuration.
2. The method for personalized generation of electroacupuncture stimulation programs integrating electromyography and brain functional network states according to claim 1, characterized in that, Step one includes: S11. Under electroacupuncture stimulation, a high-density EEG electrode array is deployed on the scalp surface covering key functional areas to collect real-time cortical activity signals during the patient's rehabilitation training tasks; the collected EEG signals are then decomposed into frequency bands to obtain... Wave power spectral density and Wave power spectral density; S12. Deploy surface electromyography (EMG) sensors on the surface of the relevant muscle groups to collect muscle discharge signals in real time, and obtain the root mean square value of EMG discharge through feature extraction. ; S13. Acquire neural activity information of the amygdala and anterior cingulate cortex using functional magnetic resonance imaging (fMRI) equipment, including brain region metabolic and blood flow signal data. S14. During the electroacupuncture stimulation process, the electroacupuncture control module records stimulation parameters in real time, including stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim; through a synchronous time stamping mechanism, it is time-aligned with high-density EEG signals and electromyography (EMG) signals to construct an electroacupuncture-EEG-EMG synchronous acquisition dataset.
3. The method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states according to claim 1, characterized in that, Step two includes: S21. Extract and centralize the synchronously acquired data from electroacupuncture, EEG, and EMG. Wave power spectral density and Wave power spectral density and root mean square value of electromyography discharge The correlation coefficient between EEG power and EMG amplitude was obtained using a cortical-muscle signal co-analysis method. ; S22. Brain region metabolic and blood flow signal data acquired simultaneously by fMRI and PET were used to obtain the rate of change in functional connectivity between the amygdala and the anterior cingulate cortex using dynamic causal modeling (DCM) and correlation modeling methods. ; S23, based on Wave power spectral density and The power spectral density was determined by using time-series resampling and differential analysis to acquire synchronized corrected EEG power time-series data. Then, a power spectral tracking algorithm was used to calculate the power changes at different time points within the electroacupuncture stimulation cycle, obtaining the rate of change of electrical power in the target brain region. .
4. The method for personalized generation of electroacupuncture stimulation programs integrating electromyography and brain functional network states according to claim 1, characterized in that, Step two also includes: S24. The correlation coefficient between the obtained EEG power and EMG amplitude. Changes in functional connectivity between the amygdala and the anterior cingulate cortex and the rate of change of electrical power in the target brain region After dimensionless normalization, the brain muscle synergistic response index (BMEI) was calculated and obtained. S25. By setting a pre-defined brain-muscle synergy response threshold Bth, and comparing and analyzing the brain-muscle synergy response index BMEI with the brain-muscle synergy response threshold Bth, the first assessment results are obtained, including: When the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is qualified, and the current electroacupuncture stimulation is determined to be effective. Continuous monitoring is required. When the brain-muscle synergy response index BMEI is less than the brain-muscle synergy response threshold Bth, it indicates that the brain-muscle pathway synergy is not up to standard, and there is a risk of delayed nerve conduction, blocked information transmission, or weakened stimulus response. This triggers the first warning instruction and generates the first strategy: automatically execute the parameter adjustment instruction to reduce the stimulus intensity and adjust the stimulus frequency, optimize and recalculate until the brain-muscle synergy response index BMEI is greater than or equal to the brain-muscle synergy response threshold Bth.
5. The method for personalized generation of electroacupuncture stimulation programs integrating electromyography and brain functional network states according to claim 1, characterized in that, Step three includes: S31. Based on the time series data of EEG power under the effective state of electroacupuncture stimulation, time-frequency analysis and waveform pattern recognition technology are used to extract features from the synchronously corrected EEG signals, obtain typical waveform feature parameters of epileptiform discharges such as spikes, slow waves and spike-slow wave complexes, and construct an epileptiform discharge feature set. S32. Based on the epileptiform discharge feature set and combined with synchronously acquired high-density EEG signal data, the feature mapping and power spectrum analysis methods are used to locate and bandpass filter the epileptiform discharge feature segments; the power spectral density Pspike(t) of the epileptiform discharge segment in the EEG signal at time t is obtained by power integration calculation.
6. The method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states according to claim 1, characterized in that, Step three also includes: S33. Power spectral density of epileptiform discharge segments in the EEG signal at time t. After dimensionless processing, the sliding time window difference algorithm is used to dynamically calculate the power spectral density change in the continuous time series and obtain the epileptiform discharge power change rate GLB. S34. By setting a preset epileptic discharge threshold Gth, and comparing the epileptiform discharge power change rate GLB with the epileptic discharge threshold Gth, the second evaluation results are obtained, including: When the rate of change of epileptiform discharge power GLB ≤ epileptiform discharge threshold Gth, the current state is determined to be normal synchronous discharge, the current electroacupuncture stimulation mode is maintained, and continuous monitoring is performed. When the rate of change of epileptiform discharge power (GLB) exceeds the epileptiform discharge threshold (Gth), the current state is determined to be an abnormal synchronous discharge state, with a risk of excessive discharge or synchronous enhancement. This triggers a second warning instruction and generates a second strategy: adjusting the stimulation intensity, waveform type, and stimulation frequency. The stimulation intensity is adjusted linearly based on the coordinated trend of the rate of change of epileptiform discharge power and the root mean square value of electromyographic discharge. The waveform type is adaptively matched based on the distribution changes of the power spectrum structure characteristics. The stimulation frequency is based on the rate of change of functional connectivity between the amygdala and the anterior cingulate cortex. Dynamic adjustment is performed; after parameter correction is completed, the power change rate is re-detected until the epileptiform discharge power change rate GLB ≤ epileptiform discharge threshold Gth. S35. After the safety constraint strategy is stably executed and no abnormal discharge is detected, the stimulation parameters are used as a safe and effective configuration to establish a set of safe stimulation parameters.
7. The method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states according to claim 1, characterized in that, Step four includes: S41. Based on the effective state of electroacupuncture stimulation determined in step two, and combined with the safe stimulation parameter set stabilized in step three, construct the brain-muscle network state space; and set the rate of change of electrical power in the target brain region. Root mean square value of electromyography (EMG) discharge Using the epileptiform discharge power change rate GLB as the state vector st, a brain-muscle network state set is established. S42. Based on the brain-muscle synergistic response index (BMEI) as the reward and punishment function, an individualized stimulus optimization model is constructed using a reinforcement learning framework. Through a joint algorithm of deep Q-learning and policy gradient, a stimulus parameter action set at is established, including stimulus intensity Istim, waveform type Fstim, and stimulus frequency fstim. The state-action mapping space is constructed and the model parameters are initialized using the brain-muscle network state set as input, forming the initial reinforcement learning model Minit.
8. The method for personalized generation of electroacupuncture stimulation programs integrating electromyography and brain functional network states according to claim 1, characterized in that, Step four also includes: S43. Based on the initial reinforcement learning model Minit, high-density EEG signals, electromyography (EMG) signals, and synchronously recorded stimulation parameter action data from multiple subjects under different electroacupuncture stimulation conditions were selected as training samples. The model underwent multiple rounds of offline training. During the training process, the improvement magnitude of the response under different stimulation actions was calculated based on the temporal change trend of the Brain-Muscle Coordination Response Index (BMEI). When BMEI increased after electroacupuncture stimulation, a positive reward / penalty value was assigned; when BMEI decreased, a negative reward / penalty value was assigned, prompting the model to correct the stimulation strategy. The algorithm dynamically updated the weights of the policy network and the value network accordingly, and adaptively optimized the action selection probability. After multiple rounds of offline training, the model parameters converged, forming the first version of the reinforcement learning model M0.
9. The method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states according to claim 1, characterized in that, Step five includes: S51. Based on the first version of model M0 formed by offline training, an online learning module is introduced during real-time electroacupuncture stimulation; continuous collection of EEG and EMG signals from the subjects is performed, and the real-time BMEI value is calculated as the state input to dynamically update the model parameters; when the real-time feedback deviates from the predicted reward and punishment trend, the algorithm automatically adjusts the policy gradient direction and corrects the stimulation action output, so that the model gradually adapts to individual differences in neural response; through multiple rounds of online iterative training, a second version of reinforcement learning model M1 is formed. S52. Based on the output of model M1, an individualized stimulation strategy is generated in real time before each electroacupuncture stimulation. The strategy includes the optimal stimulation intensity Istim, waveform type Fstim, and stimulation frequency fstim, and is adaptively corrected according to the current brain-muscle network state. When brain functional connectivity is enhanced and electromyographic synchronization is improved, the parameters are maintained. When a decrease in BMEI is detected, the stimulation intensity and waveform type are automatically adjusted, and parameter combinations that can stabilize neural activity and improve synergy are selected first.
10. The method for personalized generation of electroacupuncture stimulation protocols integrating electromyography and brain functional network states according to claim 1, characterized in that, Step five also includes: S53. During the implementation of individualized stimulus strategies, continuous monitoring is required. Wave power spectral density, Wave power spectral density, root mean square value of electromyography discharge And the BMEI change trend; based on real-time feedback data, the reinforcement learning model is updated continuously, so that it can automatically accumulate experience and correct the strategy weights in subsequent training; after multiple rounds of closed-loop adaptive adjustment, an individual optimal electroacupuncture stimulation parameter set Popt is formed as a long-term treatment parameter configuration, so that the stimulation plan operates in a safe, effective and optimal dynamic balance.