Peripheral nerve stimulation system based on muscle fatigue prediction and compensation

Abstract

The invention discloses a peripheral nerve stimulation system based on muscle fatigue prediction and compensation, which is used for acquiring electromyographic signals and predicting muscle fatigue states during nerve stimulation, and timely warning and adjusting stimulation to ensure the stability of muscle strength, and comprises a stimulation output module used for stimulating target nerves to activate target muscles according to stimulation parameters; the acquisition and analysis module is used for synchronously acquiring electromyographic signals of target muscles, preprocessing the electromyographic signals and extracting electromyographic signal features; the fatigue prediction module is used for predicting the fatigue state of the muscle according to the electromyographic signal characteristics; the compensation control module is used for adjusting stimulation parameters according to the fatigue state of the muscle and sending the stimulation parameters to the stimulation output module to ensure that the muscle strength is stable in the stimulation process; and the user interaction module is used for performing stimulation control and setting in real time and displaying the stimulation effect and the muscle fatigue state. Muscle fatigue and strength attenuation in the peripheral nerve stimulation process can be dynamically predicted and compensated in time, and the safety, accuracy and reliability of nerve stimulation are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of nerve stimulation, and in particular relates to a peripheral nerve stimulation system based on muscle fatigue prediction and compensation. Background Art

[0002] Peripheral nerve stimulation is a technology widely used in the fields of medicine, rehabilitation and sports medicine. It mainly induces muscle contraction and restores or enhances muscle function by electrically stimulating peripheral nerves.

[0003] For example, Chinese patent document with publication number CN11586953A discloses an implanted electrode and a peripheral nerve stimulation system thereof; Chinese patent document with publication number CN114302758A discloses a method and system for electrical nerve stimulation.

[0004] However, long-term electrical stimulation often leads to muscle fatigue and unstable force output. Existing electrical stimulation systems usually lack the ability to timely judge the state of muscle fatigue and dynamically adjust and optimize stimulation parameters according to the muscle state to adapt to muscle fatigue and force attenuation. It is difficult to achieve continuous, accurate and stable control during the stimulation process, which limits its effectiveness and adaptability in clinical applications and daily use. Summary of the invention

[0005] The present invention provides a peripheral nerve stimulation system based on muscle fatigue prediction and compensation. By predicting and compensating for muscle fatigue and strength attenuation that may occur during long-term stimulation, it ensures continuous output of stable muscle strength and improves the safety, accuracy and reliability of neural electrical stimulation in aspects such as motion control and rehabilitation therapy.

[0006] A peripheral nerve stimulation system based on muscle fatigue prediction and compensation collects electromyographic signals and predicts muscle fatigue status when stimulating nerves, and timely warns and adjusts stimulation to ensure stable muscle strength, including: A stimulation output module, used for stimulating target nerves and activating target muscles according to stimulation parameters; The acquisition and analysis module is used to synchronously acquire the electromyographic signals of the target muscles, perform preprocessing and extract the electromyographic signal features; The fatigue prediction module inputs the electromyographic signal characteristics and stimulation parameters of the past and current moments into the fatigue prediction model to obtain the predicted value of the electromyographic signal characteristics at the future moment; and then inputs the predicted value into the muscle fatigue regression model to obtain the fatigue state of the target muscle; The compensation control module inputs the muscle fatigue state into the strength model, calculates the optimal stimulation parameters using the model predictive control algorithm, and adjusts the nerve stimulation parameters to the optimal stimulation parameters, thereby compensating for the strength loss caused by muscle fatigue and keeping the muscle strength stable during the stimulation process; The user interaction module is used for real-time stimulation control and setting, and displays the stimulation effect and muscle fatigue status.

[0007] Preferably, the target nerve is a peripheral nerve related to movement.

[0008] Optionally, the electromyographic signal of the target muscle is a surface electromyographic signal or an intramuscular electromyographic signal.

[0009] In the acquisition and analysis module, preprocessing is performed and the characteristics of the electromyographic signal are extracted, specifically: The electromyographic signal is preprocessed by removing baseline drift, removing artifacts, filtering, rectifying and signal normalization. After preprocessing, the time domain characteristics and frequency domain characteristics of the electromyographic signal are extracted.

[0010] In the fatigue prediction module, the fatigue prediction model uses a multivariate time series model, takes past and current electromyographic signal characteristics and stimulation parameters as input, and uses temporal dependencies to predict and output future electromyographic signal characteristics.

[0011] Furthermore, an online learning method is used to continuously update and optimize the fatigue prediction model. At each time step, new EMG signal features are input into the fatigue prediction model, and the fatigue prediction model iteratively optimizes the model according to the error between the latest true value and the predicted value.

[0012] In the fatigue prediction module, the muscle fatigue regression model establishes a mapping between EMG signal features and fatigue states by learning the relationship between EMG signal features in historical data and calibrated fatigue states; after training, when new EMG signal features are input, the muscle fatigue regression model gives the muscle fatigue state based on the signal features.

[0013] In the compensation control module, the strength model includes the relationship between muscle strength output, electrical stimulation parameters, and muscle fatigue state, and the muscle strength output is obtained by inputting electrical stimulation parameters and muscle fatigue state.

[0014] Furthermore, after the strength model is established, the model predictive control algorithm defines the optimization goal as minimizing the error between the muscle output force and the target force, and predicts future muscle strength changes based on the muscle fatigue state and stimulation parameters; in each control cycle, based on the error between the target force and the actual force, the electrical stimulation parameters are optimized under constraints and the electrical stimulation output is adjusted to make the muscle output force close to the target force; and by recalculating the optimization problem after a control cycle, the muscle force prediction and stimulation parameter adjustment are rolled over to adapt to the muscle strength drop caused by different muscle fatigue states, thereby ensuring the stability and accuracy of muscle force output.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention can simultaneously collect and analyze electromyographic signals during the stimulation process, monitor the fatigue state of the muscles in real time and predict the fatigue state of the muscles in the future, provide timely warning of muscle fatigue, avoid muscle damage and large deviation in force output, and ensure the safety of electrical stimulation and motion control.

[0016] 2. Compared with traditional methods, the present invention not only solves the problem of being unable to timely detect the state of muscle fatigue during electrical stimulation, but also can adjust the stimulation output in time according to the muscle fatigue state based on the real-time feedback of electrophysiological signals to compensate for the strength attenuation caused by muscle fatigue, thereby achieving stable muscle strength output that meets the target and ensuring the accuracy and reliability of electrical stimulation movement control.

[0017] 3. The muscle fatigue prediction and compensation mechanism of the present invention can dynamically adjust the stimulation parameters according to different muscle fatigue states, and continuously optimize and update the prediction and control algorithms during the stimulation process, which greatly improves the adaptability and reliability of the neural electrical stimulation system, and can provide users with more accurate, stable and safer electrical stimulation effects during motion control and rehabilitation. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a flow chart of a peripheral nerve stimulation method based on muscle fatigue prediction and compensation according to an embodiment of the present invention.

[0019] Figure 2 Schematic diagram of an algorithm for predicting the characteristics of electromyographic signals at future moments using a fatigue prediction model in an embodiment of the present invention.

[0020] Figure 3 Schematic diagram of an algorithm for obtaining muscle fatigue status using a muscle fatigue regression model in an embodiment of the present invention.

[0021] Figure 4 This is a schematic diagram of using a model predictive control algorithm to perform fatigue compensation optimization control in an embodiment of the present invention.

[0022] Figure 5 This is a flowchart of a program for predicting and compensating muscle fatigue based on electromyographic signals in an embodiment of the present invention.

[0023] Figure 6 This is a structural diagram of a peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be pointed out that the embodiments described below are intended to facilitate the understanding of the present invention and do not have any limiting effect on the present invention.

[0025] First, the principle of the change of electromyographic signal when the muscle is fatigued and the adjustment of stimulation to compensate for the muscle fatigue and maintain the strength stability involved in the embodiment of the present invention is briefly introduced.

[0026] The changes in electromyographic signals during muscle fatigue are mainly reflected in: 1. Time domain: When muscles become fatigued, the amplitude of the EMG signal will show an overall downward trend.

[0027] 2. Spectrum: When muscles are fatigued, the EMG signal usually shows a decrease in high-frequency components and an increase in low-frequency components.

[0028] When muscles are fatigued during electrical stimulation, the stimulation parameters are adjusted to compensate for the force attenuation caused by muscle fatigue: 1. Adjust the stimulation frequency: By increasing the stimulation frequency, the muscle contraction speed is enhanced, thereby compensating for the decrease in muscle strength caused by muscle fatigue.

[0029] 2. Adjust the stimulation intensity: By increasing the stimulation intensity, more nerve fibers and corresponding muscles are activated, thereby providing additional force output when the muscles are fatigued and compensating for the force attenuation caused by muscle fatigue.

[0030] 3. Adjust the stimulation pulse width: Increase the amount of charge injected into the nerves so that more nerve fibers are activated, generating greater muscle force, thereby maintaining force output when the muscles are fatigued.

[0031] like Figure 1 As shown, a peripheral nerve stimulation method based on muscle fatigue prediction and compensation comprises the following steps: Step 101, stimulating the target nerve according to the initial stimulation parameters to generate movement of the target muscle.

[0032] According to treatment needs or motion control goals, users can pre-set motion control goals, configure initial stimulation parameters such as stimulation frequency, pulse amplitude and pulse width according to the preset goals, apply stimulation pulses to the target area through stimulation electrodes implanted in the target nerves, activate the corresponding nerve area and induce contraction of the target muscles to perform the predetermined motion tasks.

[0033] Step 102, collect the electromyographic signal of the target muscle, perform preprocessing and extract signal features.

[0034] The muscle electromyographic signals are collected in real time through the acquisition electrodes in contact with the target muscles, and preliminary data processing is performed, including removing baseline drift, filtering, rectification and signal normalization, to improve signal quality and reduce external interference. The preprocessed electromyographic signals are subjected to time domain and frequency domain feature extraction. The time domain features include the amplitude, power and root mean square value (RMS) of the electromyographic signals, while the frequency domain features focus on the spectrum characteristics of each frequency band, including high frequency components (HF), low frequency components (LF) and mean power frequency (MPF).

[0035] Step 103: predicting the fatigue state of the target muscle according to the extracted electromyographic signal features.

[0036] The current EMG signal features are combined with the current stimulation parameters and input into the fatigue prediction model to obtain the predicted value of the EMG signal features at the next moment. The predicted value is then input into the muscle fatigue regression model to obtain the muscle fatigue state and promptly remind the user. In this process, the prediction model is optimized through online iteration based on the comparison between the actual value of the collected signal and the predicted value calculated by the model, making the prediction result more accurate and reliable.

[0037] Step 104, adjusting stimulation parameters according to the muscle fatigue state to ensure stable muscle strength during the stimulation process.

[0038] The muscle fatigue state is input into the strength model, and the model predictive control algorithm is used to calculate and update the optimal stimulation plan, adjust the neural stimulation parameters, and compensate for the power loss caused by muscle fatigue. The compensation control algorithm will calculate the amount of stimulation parameter adjustment required to compensate for muscle fatigue, and combine the parameters such as electrical stimulation frequency, amplitude and cycle and the stimulation mode to generate the optimal adjustment plan, and adjust the stimulation output to help the fatigued muscles maintain a stable power output. During the entire stimulation process, the system continuously generates compensation plans dynamically based on the electromyographic signals in real time, and adjusts the stimulation output so that the movement control goals can always be effectively achieved during the entire stimulation process, ensuring the persistence and stability of muscle strength. At the same time, the prediction results and compensation effects are fed back to the system to further optimize the prediction accuracy of the model. In addition, the model is regularly retrained with new data to adapt to individual differences and long-term changes.

[0039] like Figure 2As shown in the figure, the fatigue prediction model uses the time series prediction method to extract the electromyographic signal features at the current moment. At the same time, the electrical stimulation parameters at the current moment are also important input features. The electrical stimulation features need to be used as the input of the time series model together with the electromyographic signal features. The model design uses the long short-term memory network (LSTM) to capture the long-term dependencies in the time series data. The data set is pre-divided into training set, validation set and test set: the training set is used for model training, the validation set is used for hyperparameter tuning, and the test set is used for model evaluation. The trained model combines the electromyographic signal features and electrical stimulation parameters at the past and current moments to predict the future electromyographic signals in real time. Specifically, the model will input the signal features and stimulation parameters such as stimulation frequency, amplitude and pulse width at the past and current moments into the network for processing through a series of time windows, and generate prediction results. In order to continuously improve the prediction accuracy, the system also uses online learning methods to continuously update and optimize the model. At each time step, new signal features will be input into the model, and the model will iteratively optimize the model based on the error between the latest true value and the predicted value.

[0040] like Figure 3 As shown in the figure, the muscle fatigue regression model first establishes a mapping between the EMG signal features and the fatigue state by learning the relationship between the EMG signal and the calibrated fatigue state in the historical data. After training, when a new EMG signal is input, the regression model can give the fatigue state of the muscle according to the signal features. The EMG signal features collected in advance or the predicted EMG signal features can be used through the regression model to obtain the current fatigue state of the muscle or the predicted value of the future fatigue state, which provides support for further warning and compensation of muscle fatigue.

[0041] like Figure 4As shown in the figure, the process of using the model predictive control algorithm to optimize motion control by stimulation compensation is demonstrated. First, a force model describing muscle response is established. This model contains the relationship between muscle force output, electrical stimulation parameters and muscle fatigue, and the muscle force output is obtained by inputting electrical stimulation parameters and muscle fatigue state. On this basis, the MPC algorithm seeks the optimal electrical stimulation parameters by setting an optimization problem. Its goal is to minimize the error between the muscle output force and the set target force, while satisfying the physical constraints of electrical stimulation (the output range of the stimulator) and the physiological limitations of the muscle (the safe stimulation range). Each optimization process is based on the prediction results at the current moment, and the control strategy is adjusted through real-time feedback to achieve stable force output. In the implementation phase, after the MPC controller calculates the optimal electrical stimulation parameters, it adjusts the electrical stimulation device for stimulation. This process is constantly adjusted in real time. The system calculates the error based on the comparison between the actual muscle force output and the target force, and updates the prediction model. At this time, the MPC dynamically adjusts the control parameters based on the error feedback, continuously optimizes the prediction model and control strategy, ensures that the muscle output force is consistent with the target force, and keeps the motion control stable.

[0042] like Figure 5 As shown in the figure, a process of predicting and compensating muscle fatigue based on electromyographic signals is shown. First, after configuring the initial stimulation parameters and outputting the stimulation, the electromyographic signals are collected and the fatigue state is detected and judged, and the user is reminded whether the muscle is fatigued. If the user chooses to stop the stimulation, the stimulation output is turned off and the stimulation process ends. If the user does not choose to stop the stimulation, the next step is to select whether to adjust the stimulation. If the stimulation is not adjusted, the stimulation is output according to the original stimulation parameters until the stimulation stops. If the stimulation is adjusted, the stimulation scheme is optimized according to the muscle fatigue state, and the stimulation parameters are reconfigured to output the stimulation for fatigue compensation, and then the above procedure is continued.

[0043] like Figure 6 As shown, a peripheral nerve stimulation system based on muscle fatigue prediction and compensation is mainly composed of five parts: a stimulation output module 1, an acquisition and analysis module 2, a fatigue prediction module 3, a compensation control module 4 and a user interaction module 5.

[0044] The stimulation output module 1 comprises a parameter configuration unit 11, which is responsible for generating stimulation parameters; and a current output unit 12, which is responsible for generating stimulation current and outputting it to the target area.

[0045] The acquisition and analysis module 2 includes an acquisition and amplification unit 21, which is responsible for the acquisition and amplification of the electromyographic signal of the target muscle; and a processing and analysis unit 22, which is responsible for the preprocessing and feature extraction of the electromyographic signal.

[0046] The fatigue prediction module 3 comprises an electromyography prediction unit 31, which is responsible for predicting the electromyography signal characteristics at a future moment; and a fatigue calculation unit 32, which is responsible for calculating the muscle fatigue state according to the electromyography signal characteristics.

[0047] The compensation control module 4 includes a strength model unit 41, which is responsible for constructing the response relationship between the muscle and the stimulation parameters; and an optimization control unit 42, which is responsible for generating the optimal stimulation parameters according to the latest state.

[0048] The user interaction module 5 includes a user interface 51, which is responsible for real-time display of signals, stimulation parameters, fatigue status, etc., and for stimulation settings; a data interface 52, which is responsible for data transmission and reception between the user end and the device end.

[0049] In one embodiment, the acquisition and amplification unit 21 in the acquisition and analysis module 2 adopts the RHD2132 acquisition chip, which is responsible for acquiring and amplifying electrophysiological signals. The processing and analysis unit 22 mainly includes FPGA, which is responsible for signal processing and data analysis; the current output unit 12 is mainly composed of a high-precision time phase control circuit and a stimulation interface, which is responsible for outputting the specified stimulation waveform and can support 16 channels of synchronous and asynchronous stimulation. The above two parts can communicate through optical coupling, which can greatly reduce the impact of the stimulation output module 1 on the acquisition and analysis module 2. The system is divided into offline mode and online mode. The offline mode includes the acquisition and analysis module 2 and the stimulation output module 1. At this time, the FPGA collects data and stores it in the SD card. Different functions are realized according to different algorithms written. The offline mode can support at least 24 hours of continuous recording and stimulation. Compared with the offline mode, the online mode adds a user interaction module 5. The electrophysiological signal transmits the data to the user interface 51 through the data interface 52 (Bluetooth or Ethernet). The electrophysiological signal can be observed online in real time. At the same time, the stimulation parameters of the stimulation system can be configured online, and the electrical stimulation can be manually triggered according to the needs.

[0050] In one embodiment, MCU (STM32RCT6) is used as the main control of the parameter configuration unit 11, operating at a main frequency of 72MHz. The MCU communicates with the external DAC through a high-speed SPI interface under DMA control, and the communication rate can reach 36MHz. The model of the external DAC is AD5674, which has 16 independent DAC units with 12bits accuracy. The MCU controls the current output of the constant current circuit by comparing the voltage output by the DAC with the reference voltage Vref in the constant current circuit. The stimulation output module 1 also has a safety design, which is mainly manifested in two points: 1. ADG1412 is used to achieve grounding in the non-stimulation phase to prevent leakage current from affecting animal tissue and signal quality. ADG1412 is a normally closed switch. Under normal circumstances, the output and GND are short-circuited, and the on-resistance is only 1.5Ω. The electrode impedance of microwire electrodes and flexible electrodes commonly used for implanting animal tissues is between 10KΩ-100KΩ, and there is only 0.001%-0.01% leakage current, which can be basically ignored, to protect animal tissue and ensure signal quality.

[0051] 2. Through high-speed optocoupler (ISO7341) communication with FPGA, it can support 25Mbps communication rate. While meeting the communication rate requirements, it can achieve complete electrical isolation between the stimulation part and the data processing part (FPGA), further ensuring the signal quality.

[0052] The embodiments described above provide a detailed description of the technical solutions and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not intended to limit the present invention. Any modifications, supplements and equivalent substitutions made within the scope of the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A peripheral nerve stimulation system based on muscle fatigue prediction and compensation, characterized in that: When stimulating nerves, it collects electromyographic signals and predicts muscle fatigue, and promptly warns and adjusts stimulation to ensure stable muscle strength, including: A stimulation output module, used for stimulating target nerves and activating target muscles according to stimulation parameters; The acquisition and analysis module is used to synchronously acquire the electromyographic signals of the target muscles, perform preprocessing and extract the electromyographic signal features; The fatigue prediction module inputs the electromyographic signal characteristics and stimulation parameters of the past and current moments into the fatigue prediction model to obtain the predicted value of the electromyographic signal characteristics at the future moment; and then inputs the predicted value into the muscle fatigue regression model to obtain the fatigue state of the target muscle; The compensation control module inputs the muscle fatigue state into the strength model, calculates the optimal stimulation parameters using the model predictive control algorithm, and adjusts the nerve stimulation parameters to the optimal stimulation parameters, thereby compensating for the strength loss caused by muscle fatigue and keeping the muscle strength stable during the stimulation process; The user interaction module is used for real-time stimulation control and setting, and displays the stimulation effect and muscle fatigue status.

2. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: The target nerves are peripheral nerves related to movement.

3. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: The EMG signal of the target muscle is a surface EMG signal or an intramuscular EMG signal.

4. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: In the acquisition and analysis module, preprocessing is performed and the characteristics of the electromyographic signal are extracted, specifically: The electromyographic signal is preprocessed by removing baseline drift, removing artifacts, filtering, rectifying and signal normalization. After preprocessing, the time domain characteristics and frequency domain characteristics of the electromyographic signal are extracted.

5. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: In the fatigue prediction module, the fatigue prediction model uses a multivariate time series model, takes past and current electromyographic signal characteristics and stimulation parameters as input, and uses temporal dependencies to predict and output future electromyographic signal characteristics.

6. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 5, characterized in that: An online learning method is used to continuously update and optimize the fatigue prediction model. At each time step, new electromyographic signal features are input into the fatigue prediction model, and the fatigue prediction model iteratively optimizes the model based on the error between the latest true value and the predicted value.

7. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: In the fatigue prediction module, the muscle fatigue regression model establishes a mapping between the EMG signal features and the fatigue state by learning the relationship between the EMG signal features and the calibrated fatigue state in the historical data; After training, when new EMG signal features are input, the muscle fatigue regression model gives the muscle fatigue state according to the signal features.

8. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: In the compensation control module, the strength model includes the relationship between muscle strength output, electrical stimulation parameters, and muscle fatigue state, and the muscle strength output is obtained by inputting electrical stimulation parameters and muscle fatigue state.

9. The peripheral nerve stimulation system based on muscle fatigue prediction and compensation according to claim 1, characterized in that: In the compensation control module, after the strength model is established, the model predictive control algorithm defines the optimization goal as minimizing the error between the muscle output force and the target force, and predicts the future muscle force changes according to the muscle fatigue state and stimulation parameters; in each control cycle, according to the error between the target force and the actual force, the electrical stimulation parameters are optimized under the constraints and the electrical stimulation output is adjusted to make the muscle output force close to the target force; By recalculating the optimization problem after a control cycle, the muscle strength prediction and stimulation parameter adjustments are updated in a rolling manner to adapt to the muscle strength decline caused by different muscle fatigue states, thereby ensuring the stability and accuracy of muscle strength output.

Images