An active real-time closed-loop electrical stimulation method and system for Parkinson's disease
By identifying and utilizing the motor imaginative EEG signals of Parkinson's patients and adjusting the parameters of brain implanted electrical stimulation system in real time, the problems of individualized differences and many side reactions in the existing technology are solved, and more efficient and safe treatment is achieved Effect.
Patent Information
- Application Number
- CN202210047709.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-01-17
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-01-17
AI Technical Summary
The existing brain implantation electrical stimulation therapy has problems such as large individualization differences, unstable efficacy and many side effects in the treatment of Parkinson's disease, and lacks effective biomarkers and dynamic management methods.
By collecting EEG signals in patients' motion imagination and actual movement, identifying instructions in EEG signals in motor imagination, real-time adjustment of DBS parameters is achieved, and spatial domain features are extracted using the co-spatial pattern algorithm, generating biomarkers and storing them to healthy organisms The label library performs closed-loop regulation based on the electroencephalogram signal and motion imagination signal under abnormal states, and adjusts the electrical stimulation parameters in real time.
It improves the efficacy of electrical stimulation, reduces side reactions, realizes dynamic management centered on patient needs, and enhances the individualization and safety of treatment.
Smart Images

Figure CN114259651B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to biomedical engineering technology, in particular to an active real-time closed-loop electrical stimulation system for Parkinson's disease. Background Art
[0002] Parkinson's disease (PD) is a common degenerative neurological disease. The global prevalence of people over 60 years old is about 1.7%. It is predicted that by 2030, the number of PD patients worldwide will reach 15 million, and nearly half of them will be in my country. Neurodegenerative diseases have a long course and are difficult to cure. The medical burden is heavy, which seriously affects the healthy development of the country's economy and society. Since the beginning of the disease, the symptoms gradually worsen, causing a sharp decline in the patient's quality of life and placing a heavy burden on the patient and his family.
[0003] Brain implant electrical stimulation therapy (DBS) is the mainstream treatment for drug-refractory Parkinson's disease, and more than 200,000 Parkinson's disease patients worldwide receive this therapy. However, clinical DBS technology faces many challenges. First, the individual differences in patient efficacy are large. Taking motor symptoms as an example, the short-term improvement rate after surgery ranges from 20% to 80%. Clinical, imaging, and electrophysiological data characteristics show certain predictive performance, but the prediction accuracy is insufficient in individuals, and there is a lack of effective biomarkers; second, the patient's postoperative management burden is heavy, and patients and caregivers need to go back and forth to the hospital for programming many times to achieve the best treatment method, and the stimulation parameters cannot be automatically adjusted according to the patient's condition. In addition, clinical DBS is an open-loop DBS, that is, the selection of DBS stimulation parameters such as intensity, frequency, pulse shape, and phase is based only on previous empirical evidence and remains unchanged during programming. Long-term inappropriate stimulation may induce new adverse reactions, such as gait disorders and articulation problems, which bring significant troubles to patients' lives.
[0004] Some limitations can be addressed by applying closed-loop neural stimulation. In closed-loop neural stimulation, the electrophysiological or biological correlation between pathological biomarkers and the neurological disease to be treated is continuously monitored, and DBS stimulation parameters are adjusted in real time according to their fluctuations to reduce the negative effects of constant stimulation. However, in the current research on closed-loop stimulation, a single and effective biomarker with a pathological biomarker expected value has not yet been found. The key to improving the efficacy of closed-loop stimulation and minimizing the degree of side effects lies in the search for predictive markers, the understanding of the working mechanism of DBS, and dynamic management centered on patient needs.
[0005] Motor imagery therapy activates a specific area of a certain activity in the brain based on motor memory, thereby achieving the purpose of improving motor function. For Parkinson's disease, the brain is the root cause of the disease, and motor dysfunction is the result and manifestation of the disease. Motor imagery therapy is a bridge that can connect the patient's central nervous system and peripheral movement, providing a new control method for patients with movement disorders. Through motor imagery, the selection of expected signals in the closed-loop control of different motor dysfunctions (such as tremor, muscle rigidity, pathological gait, etc.) is realized. Compared with traditional closed-loop control, motor imagery is more humanistic and can flexibly switch electrical stimulation control schemes to achieve patient-centered, guided selection and adjustment of closed-loop expected signals, thereby maximizing the improvement of electrical stimulation efficacy and reducing the degree of side effects. Summary of the invention
[0006] In view of the above problems, the purpose of the present invention is to provide an active real-time closed-loop electrical stimulation system for Parkinson's disease. By collecting the EEG signals of the patient during motor imagery and actual movement, identifying the instructions in the motor imagery EEG signals, achieving an overall understanding of the patient's own health and disease state and resolving the biomarkers and pathological characteristics of specific actions, and then obtaining the difference between the biomarkers in the pathological characteristics and the healthy state, and achieving real-time adjustment of DBS parameters, it provides an active digital real-time closed-loop electrical stimulation platform with humanistic care for the study of Parkinson's disease.
[0007] To achieve the above object, the technical solution adopted by the present invention is:
[0008] An active digital real-time closed-loop electrical stimulation system for Parkinson's disease, characterized in that the working process of the system is:
[0009] 1) Collecting EEG signals of patients in different normal motion states, then preprocessing the collected EEG signals of patients in normal motion to remove interference and highlight EEG features, and then decoding the EEG features of the preprocessed EEG signals, that is, using the common space pattern algorithm to extract spatial domain features, and performing class activation mapping visualization processing on the feature contribution, to obtain biomarkers of specific actions with high spatial and temporal resolution and store them in a healthy biomarker library; the healthy biomarker library is all the corresponding biomarkers captured in the same motion state within a period of time during the current normal motion state of the patient, that is, different motion states and corresponding time-related biomarkers are matched together, and the healthy biomarker library stores biomarkers of different motion states;
[0010] 2) Detect abnormal states during the patient's movement. Abnormal states are accompanied by motor imagery. In abnormal states, motor imagery EEG signals and abnormal EEG signals during abnormal movement are collected in sequence; monitor whether the current motor imagery process is effective through heart rate monitoring. If motor imagery is effective, perform EEG preprocessing and EEG feature decoding on the collected motor imagery EEG signals to obtain corresponding action features, generate corresponding state instructions, and select corresponding health biomarkers from the health biomarker library as the expected features of the patient's abnormal state according to the state instructions; select different biomarkers for closed-loop control according to different state instructions;
[0011] 3) When motor imagery is effective, abnormal EEG signals during the patient's abnormal movement are collected at the same time. The abnormal EEG signals are preprocessed and decoded to obtain pathological features. The extracted pathological features and the expected features corresponding to the state instructions determined by motor imagery are sent to the controller together. The controller outputs the actual control signal to make the pathological features and the expected features continuously tend to be consistent, realize the real-time adjustment of the electrical stimulation parameters, and complete the closed-loop active regulation.
[0012] On the other hand, the present invention also provides an active digital real-time closed-loop electrical stimulation system for Parkinson's disease, the system comprising an EEG acquisition unit, a brain-computer interface unit and a closed-loop control unit;
[0013] The EEG acquisition unit includes a heart rate monitoring module, a motor imagery and actual movement EEG signal acquisition module, and the actual movement includes normal movement and abnormal movement;
[0014] The heart rate detection module is used to monitor whether the increase in the patient's heart rate during motor imagery is consistent with that during actual exercise execution, and to determine whether motor imagery is actually performed;
[0015] The EEG acquisition unit is composed of an amplifier (Symtop, Beijing, China), an Agcl powder electrode cap, 19 electrode sheets, and 4 additional signal channels; the electrode sheets are placed at the corresponding positions of the subject's scalp according to the 10-20 international standard electrode system; the 4 additional signal channels are used to remove eye artifacts, record horizontal and vertical electrooculograms, and monitor the subject's eye movements and blinking during the acquisition process;
[0016] The brain-computer interface unit includes an EEG preprocessing module and a feature decoding module. The EEG preprocessing module is used to remove interference from EEG signals to highlight EEG features, and the feature decoding module is used to decode EEG features to obtain EEG features with high spatial and temporal resolution.
[0017] The closed-loop control unit includes a controller and an electrical stimulation unit, which analyzes the pathological characteristics of the patient's abnormal movement process EEG, and selects the expected characteristics under the corresponding normal state by analyzing the instructions obtained from the motor imagery EEG. The two are used as controller inputs and output actual control signals to adjust the DBS stimulation parameters in real time.
[0018] Furthermore, the EEG preprocessing module includes three parts: zero phase shift filtering, independent component analysis and manual data screening. The zero phase shift filtering module is used to control the frequency range of the EEG signal band; the independent component analysis module is used to remove the noise generated by the interference of electrooculography during the EEG acquisition process; the manual data screening is used to remove the EEG noise generated by the subject's unconscious head movement or body movement during the acquisition process, and the EEG noise amplitude of the manual data screening is greater than 80μV and less than -80μV;
[0019] The feature decoding module includes two parts: a co-spatial model and a class activation map. The co-spatial model extracts spatial domain features from the processed EEG, and the class activation map module visualizes the contribution of the extracted features to find the features with the greatest contribution to health and disease states as the corresponding action EEG features.
[0020] Compared with the prior art, the present invention has the following beneficial effects:
[0021] The system of the present invention provides a dynamic closed-loop management centered on patient needs and a suitable biomarker adjustment method for real-time closed-loop control of the nervous system. Motor imagery therapy is used to help patients actively identify symptoms of the disease, select suitable biomarkers as the basis for DBS regulation, identify the expected characteristics that change with the movement state, expand the scope of care, and help patients maintain treatment without increasing the burden.
[0022] The present invention uses motor imagery in the treatment of Parkinson's disease, and uses motor imagery signals as command signals of the control strategy, participates in the control process, and is used for the selection of desired biomarkers. Parkinson's disease can cause motor dysfunction (such as tremor, muscle rigidity, pathological gait, etc.). Motor imagery can serve as a bridge between the patient's central nervous system and peripheral movement, and helps to achieve active closed-loop management centered on patient needs, minimize the degree of side effects, and meet the individualized difference adjustment of patient efficacy. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] Figure 1 Overall block diagram of the active real-time closed-loop electrical stimulation system of the present invention.
[0024] Figure 2 EEG acquisition electrode distribution map.
[0025] Figure 3 Circuit schematic diagram of the heart rate monitoring module of the present invention.
[0026] Figure 4 Flowchart of the EEG preprocessing module in the present invention.
[0027] Figure 5 Flowchart of the EEG feature decoding module in the present invention. DETAILED DESCRIPTION
[0028] The present invention is further described in detail below with reference to the accompanying drawings.
[0029] The present invention is directed to an active real-time closed-loop electrical stimulation system for Parkinson's disease, as shown in the attached Figure 1 The system includes three parts: an EEG acquisition unit, a brain-computer interface unit, and a closed-loop control unit. The EEG acquisition unit consists of a heart rate monitoring module, a motor imagery and human movement EEG signal acquisition module. The heart rate detection module consists of a MAX30102 chip and its peripheral circuits; the motor imagery and human movement EEG acquisition modules share a set of equipment, which consists of an amplifier (Symtop, Beijing, China), an Agcl powder electrode cap, 19 electrode sheets, and 4 additional signal channels;
[0030] The brain-computer interface unit includes an EEG preprocessing module and an EEG feature decoding module, and is connected to the EEG acquisition unit; the EEG preprocessing module performs preprocessing such as filtering, independent component analysis, and manual screening on the acquired signals; the EEG feature decoding module uses a common spatial paternal (CSP) algorithm to extract features from the preprocessed EEG signals, and uses a class activation mapping method to visualize the contribution of the extracted features, and uses the feature with the greatest contribution to health and disease states as the EEG feature of the action, and the EEG feature includes both the following biomarkers, action features, and pathological features;
[0031] The closed-loop control unit includes a controller and an electrical stimulation unit, which analyzes the pathological characteristics of the patient's abnormal movement process EEG, and selects the expected characteristics under the corresponding normal state by analyzing the instructions obtained from the motor imagery EEG. The two are used as controller inputs and output actual control signals to adjust the DBS stimulation parameters in real time.
[0032] The specific workflow of the system is:
[0033] 1) First, the EEG signals of the patient's different normal movement states are collected, and then the EEG signals collected during the normal movement of the patient are subjected to EEG preprocessing such as zero phase shift filtering, independent component analysis, and artificial digital filtering (artificial screening) to remove interference and highlight the EEG features, and then the EEG features of the preprocessed EEG signals are decoded, that is, the spatial domain features are extracted using the common space pattern algorithm, and the feature contribution is visualized by class activation mapping to obtain biomarkers of specific actions with high spatial and temporal resolution and store them in a healthy biomarker library. The healthy biomarker library is all the corresponding biomarkers captured by the same movement state within a period of time during the normal movement state of the current patient, that is, different movement states and corresponding and time-related biomarkers are matched together, and the healthy biomarker library stores biomarkers of different movement states, and the biomarkers of each movement state may be different, determined according to individual conditions.
[0034] 2) Then, the abnormal state of the patient's movement process is detected. The abnormal state will be accompanied by motor imagination. In the abnormal state, the motor imagination EEG signals and the abnormal EEG signals during the abnormal movement are collected in sequence; the heart rate monitoring module is used to monitor whether the current motor imagination process is effective. If the motor imagination is effective, the collected motor imagination EEG signals are subjected to EEG preprocessing and EEG feature decoding through the brain-computer interface unit to obtain the corresponding action features, and generate corresponding state instructions (the state instructions represent the patient's abnormal behavior, where instruction A corresponds to abnormal gait and slow movement, instruction B represents tremor, and instruction C represents stiffness, etc.), and the corresponding health biomarkers are selected from the health biomarker library according to the state instructions as the expected features of the patient's abnormal state (the expected features represent the patient's true intention of motor imagination is A walking, B grasping, and C standing); different biomarkers are selected according to different state instructions for closed-loop regulation.
[0035] 3) At the same time, abnormal EEG signals during the patient's abnormal movement are collected. After EEG preprocessing and EEG feature decoding, pathological features (the cause of the disease) are obtained. The extracted pathological features and the expected features corresponding to the state instructions determined by motor imagination are sent to the controller together. The controller outputs the actual control signal to make the pathological features and the expected features continue to converge, realize real-time adjustment of electrical stimulation parameters, and complete closed-loop control.
[0036] Reference Figure 2The EEG acquisition unit includes 19 electrodes, which are placed at the corresponding positions of the subject's scalp according to the 10-20 international standard electrode system, namely Fp1, Fp2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T7, T8, P7, P8, Fz, Cz, Pz. The sampling frequency of the amplifier is 500Hz. In addition, 4 additional signal channels are collected to record horizontal and vertical electrooculograms, which are used to monitor the subject's eye movements and blinks during the acquisition process and remove eye artifacts. The subject is in a quiet room. The staff puts on the electrode cap for the subject and collects the EEG signal through the amplifier. The entire acquisition process is monitored by the staff, and the normal movement process is in a quiet room to ensure that the subject collects EEG in a clear and safe state. Abnormal movement process: in this process, after the patient feels the onset of the disease, he should first perform motor imagery and imagine normal movement movements. The motor imagery EEG signals pass through the brain-computer interface unit to obtain state instructions, and the state instructions guide the selection of expected features. After that, the actual movement begins (for example, if the hand starts to tremble and cannot grasp steadily, imagine the corresponding movement to grasp steadily, and then perform the grasping movement). The abnormal state EEG signals pass through the brain-computer interface unit to obtain pathological characteristics. The expected characteristics and pathological characteristics are used as the two inputs of the controller, and the output stimulation intensity adjusts the patient's DBS parameters until the pathological characteristics are consistent with the expected characteristics.
[0037] Reference Figure 3 The heart rate monitoring module uses the MAX30102 chip, which is composed of the MAX30102 chip and its peripheral circuit. The SDA, SCL and INT# pins of the MAX30102 chip are connected to the STM32F103 microcontroller, and the SDA and SCL pins are also connected to one end of the resistor R1 and the resistor R2 respectively. The other ends of the resistor R1 and the resistor R2 are connected and then connected to the INT# pin of the MAX30102 chip through the resistor R3; the other end of the resistor R1 is simultaneously connected to the VDD pin and the voltage VCC of the MAX30102 chip, one end of the capacitor C3, and one end of the capacitor C2. The other end of the capacitor C3 and the other end of the capacitor C2 are both grounded; the VLED+ pin of the MAX30102 chip is connected to one end of the capacitor C1 and the voltage VCC, and the other end of the capacitor C1 is grounded. The increase in heart rate during motor imagery and actual exercise is consistent. The degree of increase in the patient's heart rate during motor imagery can be used to determine whether the motor imagery is effective.
[0038] Reference Figure 4The EEG preprocessing module includes three parts: zero phase shift filtering, independent component analysis and manual screening. The zero phase shift filtering is used to control the frequency range of the EEG signal band; the independent component analysis module is used to remove the noise caused by the interference of the electrooculogram during the EEG acquisition process; and the manual screening is used to remove the EEG noise caused by the subject's unconscious head or body movement during the acquisition process. The EEG noise amplitude of the manual data screening is often greater than 80μV and less than -80μV.
[0039] The zero phase shift filter controls the filtering range within 1 to 45 Hz, and the phase shift after filtering will not affect the division of dynamic brain functional networks. X(z) is the original EEG signal, and H(z) is the filtering function. First, forward filtering is performed to obtain X(z)H(z), and then the filtered sequence is reversed (time domain flip operation) to obtain X(1 / z)H(1 / z), and then backward filtering is performed through a unified filter to obtain X(1 / z)H(1 / z)H(z). The sequence obtained by filtering is finally reversed (time domain flip operation), and finally the signal Y(k) processed by zero phase shift filtering is obtained.
[0040] The independent component analysis is used to remove the noise caused by the interference of electrooculogram in the process of EEG acquisition. The natural gradient expansion algorithm is adopted, that is, Y is set as the source signal, A is the mixing matrix, S=AY is the decomposed independent component, and after obtaining each component, the electrooculogram signal and the electrocardiogram signal are distinguished according to the topographic map of each component, and the components belonging to the electrooculogram signal and the electrocardiogram signal are set to zero, and then P=WS, W is the unmixing matrix, and the original signal with the artifacts removed on the scalp is inversely obtained.
[0041] Reference Figure 5 The EEG feature decoding module includes two parts: a co-spatial model and a class activation map. The co-spatial model extracts spatial domain features from the processed EEG, the input signal is a motor imagery EEG signal, and the output is a spatial distribution component of the motor imagery EEG signal.
[0042] The class activation mapping (CAM) module has a convolutional neural network structure, and the feature vector is passed backward layer by layer through its layer structure. The deep features are further extracted and combined from the shallow features. The last convolutional layer has the highest level of semantic and spatial information. The weighted average of the activation feature map and the weights of each feature map is used to obtain its own judgment basis (i.e., which areas in the feature map are used by deep learning to identify the microstate class), explaining why the model predicts its own prediction.
[0043] The calculation of CAM for class c tags is shown in formula (1).
[0044]
[0045] Among them, f k (x, y) represents the feature map obtained by the kth convolution kernel in the last convolution layer (x, y represent the coordinates of the elements in the feature map). It represents the “contribution” of the k-th feature map to the classification of class c. The calculation formula is shown in formula (2).
[0046]
[0047] where Y c represents the score of class c, and Z represents the size of the feature map. From formula (2), we can see that the “contribution” of the kth feature map is calculated based on the average gradient of each element from class c to the last convolutional layer.
[0048] The closed-loop control unit includes a controller and a stimulator. The controller includes two inputs, one input is the pathological characteristics of the patient's EEG under abnormal conditions, and the other input is the characteristics of the patient's corresponding normal movement state, that is, the expected characteristics. The expected characteristics are obtained by analyzing the normal movement state of the human body and stored in the health biomarker library. By analyzing the EEG signals of motor imagination during the onset of the disease, the call instructions are obtained and the appropriate expected characteristics are selected. The controller outputs a control signal to adjust the electrical stimulation parameters to achieve patient-initiated real-time closed-loop control.
[0049] Any matters not described in the present invention are applicable to the prior art.
Claims
1. An active digital real-time closed-loop electrical stimulation method for Parkinson's disease, characterized in that: The workflow of this method is: 1) Collecting EEG signals of patients in different normal motion states, then preprocessing the collected EEG signals of patients in normal motion to remove interference and highlight EEG features, and then decoding the EEG features of the preprocessed EEG signals, that is, using the common space pattern algorithm to extract spatial domain features, and performing class activation mapping visualization processing on the feature contribution, to obtain biomarkers of specific actions with high spatial and temporal resolution and store them in a healthy biomarker library; the healthy biomarker library is all the corresponding biomarkers captured in the same motion state within a period of time during the current normal motion state of the patient, that is, different motion states and corresponding time-related biomarkers are matched together, and the healthy biomarker library stores biomarkers of different motion states; 2) Detect abnormal states during the patient's movement, which are accompanied by motor imagery. Under abnormal conditions, the EEG signals of motor imagery and abnormal EEG signals during abnormal movement are collected successively; The heart rate monitoring is used to monitor whether the current motor imagination process is effective. If the motor imagination is effective, the collected motor imagination EEG signals are subjected to EEG preprocessing and EEG feature decoding to obtain corresponding action features, and corresponding state instructions are generated. According to the state instructions, corresponding health biomarkers are selected from the health biomarker library as expected features of the patient's abnormal state; different biomarkers are selected according to different state instructions for closed-loop regulation; 3) When motor imagery is effective, abnormal EEG signals during the patient's abnormal movement are collected at the same time. The abnormal EEG signals are preprocessed and decoded to obtain pathological features. The extracted pathological features and the expected features corresponding to the state instructions determined by motor imagery are sent to the controller together. The controller outputs the actual control signal to make the pathological features and the expected features continuously tend to be consistent, realize the real-time adjustment of the electrical stimulation parameters, and complete the closed-loop active regulation.
2. The active digital real-time closed-loop electrical stimulation method for Parkinson's disease according to claim 1, characterized in that: The state instructions represent the patient's behavioral performance when abnormal, where instruction A corresponds to abnormal gait and slow movement, instruction B represents tremor, and instruction C represents stiffness; the expected characteristics represent the patient's actual intention of motor imagination, which is A walking, B grasping, and C standing.
3. An active digital real-time closed-loop electrical stimulation system for Parkinson's disease, the system comprising an EEG acquisition unit, a brain-computer interface unit and a closed-loop control unit; characterized in that: The EEG acquisition unit includes a heart rate monitoring module, a motor imagery and actual movement EEG signal acquisition module, and the actual movement includes normal movement and abnormal movement; The heart rate monitoring module is used to monitor whether the increase in the patient's heart rate during motor imagery is consistent with that during actual exercise execution, and to determine whether motor imagery is actually performed; The EEG acquisition unit is composed of an amplifier and an Agcl powder electrode cap, 19 electrode sheets and 4 additional signal channels; the electrode sheets are placed at the corresponding positions of the subject's scalp according to the 10-20 international standard electrode system; the 4 additional signal channels are used to remove eye artifacts, record horizontal and vertical electrooculograms respectively, and monitor the subject's eye movements and blinking during the acquisition process; The brain-computer interface unit includes an EEG preprocessing module and a feature decoding module. The EEG preprocessing module is used to remove interference from EEG signals to highlight EEG features, and the feature decoding module is used to decode EEG features to obtain EEG features with high spatial and temporal resolution. The closed-loop control unit includes a controller and an electrical stimulation unit, which analyzes the pathological characteristics of the patient's abnormal movement process EEG, and selects the expected characteristics under the corresponding normal state by analyzing the instructions obtained from the motor imagery EEG. The two are used as controller inputs and output actual control signals to adjust the DBS stimulation parameters in real time.
4. The active digital real-time closed-loop electrical stimulation system for Parkinson's disease according to claim 3, characterized in that: The EEG preprocessing module includes three modules: zero phase shift filtering, independent component analysis and manual data screening. The zero phase shift filtering module is used to control the frequency range of the EEG signal band; the independent component analysis module is used to remove the noise caused by the interference of electrooculogram during the EEG acquisition process; the manual data screening module is used to remove the EEG noise caused by the unconscious head movement or body movement of the subject during the acquisition process. The EEG noise amplitude of the manual data screening is greater than 80 and less than -80 ; The feature decoding module includes two parts: a co-spatial model and a class activation map. The co-spatial model extracts spatial domain features from the processed EEG, and the class activation map module visualizes the contribution of the extracted features to find the features with the greatest contribution to health and disease states as the corresponding action EEG features.
5. The active digital real-time closed-loop electrical stimulation system for Parkinson's disease according to claim 3, characterized in that: The sampling frequency of the amplifier is 500 Hz. The subject is in a quiet room. The staff puts an electrode cap on the subject and collects the EEG signal through the amplifier. The whole collection process is monitored by the staff. During the normal movement process, the EEG is collected in a quiet room to ensure that the subject is awake and safe. During the abnormal movement process, the patient should first perform motor imagery after feeling the onset of the disease, imagining normal movement movements. The motor imagery EEG signal passes through the brain-computer interface unit to obtain state instructions, and the state instructions guide the selection of desired features.
6. The active digital real-time closed-loop electrical stimulation system for Parkinson's disease according to claim 3, characterized in that: The heart rate monitoring module adopts the MAX30102 chip, which is composed of the MAX30102 chip and its peripheral circuit. The SDA, SCL and INT# pins of the MAX30102 chip are connected to the STM32F103 single-chip computer, and the SDA and SCL pins are also connected to one end of the resistor R1 and the resistor R2 respectively. The other ends of the resistors R1 and R2 are connected and then connected to the INT# pin of the MAX30102 chip through the resistor R3; the other end of the resistor R1 is simultaneously connected to the VDD pin and the voltage VCC of the MAX30102 chip, one end of the capacitor C3, and one end of the capacitor C2, and the other end of the capacitor C3 and the other end of the capacitor C2 are both grounded; the VLED+ pin of the MAX30102 chip is connected to one end of the capacitor C1 and the voltage VCC, and the other end of the capacitor C1 is grounded; the increase in heart rate during motor imagery and actual exercise is consistent, and whether the motor imagery is effective is judged according to the degree of increase in the patient's heart rate during motor imagery.
