Brain-computer interface-based limb function closed-loop collaborative rehabilitation system and application

Abstract

The application provides a limb function closed-loop cooperative rehabilitation system based on a brain-computer interface and an application, and belongs to the fields of rehabilitation medical devices and brain-computer interface technologies. The system comprises: a non-invasive brain-computer interface intention recognition module that collects motor cortex electroencephalogram signals of a user, recognizes a movement intention, and outputs a corresponding control instruction; a hand rehabilitation execution module that drives a paretic hand to perform finger flexion and extension rehabilitation training after receiving the control instruction; an upper limb rehabilitation training module that provides a patient with multi-plane strength and coordination rehabilitation training of the upper limbs; and an AI rehabilitation data integration and cloud management platform that records full rehabilitation cycle data of the user and generates a quantitative rehabilitation scheme. The application realizes double-mode cross verification by combining BCI to recognize a movement intention and an electromyographic sensor, changes passive traction into active rehabilitation, integrates a soft glove and a VR scene to break through the scene limitation of high-end equipment, realizes closed-loop management through an AI cloud platform, achieves full-cycle personalized rehabilitation, activates neural remodeling, and improves the life quality of the patient.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation medical devices and brain-computer interface technology, specifically to a closed-loop collaborative rehabilitation system for limb function based on brain-computer interface and its application. Background Technology

[0002] Stroke (commonly known as "apoplexy") is characterized by high incidence, high disability rate, high mortality rate, and high recurrence rate. The total number of stroke patients in my country has exceeded 28 million, with upper limb dysfunction affecting 50% to 80% of acute-phase patients, nearly half of whom suffer from chronic permanent disability. The core task of post-stroke rehabilitation is to achieve the remodeling of the functional areas of the central nervous system, which requires the patient's active participation in the rehabilitation process, using the brain to send motor intentions to drive limb training.

[0003] Existing limb rehabilitation systems suffer from the following shortcomings: First, there is a contradiction between active participation and passive execution. Traditional exoskeleton rehabilitation equipment mostly adopts a passive traction mode, requiring patients to exert force during training. While suitable for physical training of the musculoskeletal system, it cannot address the core rehabilitation issues caused by central nervous system damage. Second, advanced rehabilitation technologies are difficult to democratize. While some innovative products have made breakthroughs using brain-computer interfaces or electromyographic interfaces, these are primarily centralized devices used in high-end hospital rehabilitation departments. These devices are expensive and complex to operate, making them difficult to implement in homes and communities. Third, fragmented rehabilitation methods lead to discontinuous effectiveness. Hand grasping training, upper limb strength training, balance function training, and daily movement reproduction are handled by multiple independent rehabilitation devices and processes, lacking a unified, systematic management platform, resulting in a lack of management after discharge. Therefore, there is an urgent need for a closed-loop collaborative rehabilitation system and application for limb function that enables active patient participation, multi-scenario coverage, and full-cycle management. Summary of the Invention

[0004] The purpose of this invention is to provide a closed-loop collaborative rehabilitation system and application for limb function that integrates brain-computer interface intention recognition, multimodal rehabilitation module and Internet of Things cloud platform, in order to solve the technical problems of low patient participation, high barriers to high-end rehabilitation technology and fragmented rehabilitation process in existing limb function rehabilitation technology.

[0005] To achieve the above objectives, this invention provides a closed-loop collaborative rehabilitation system for limb function based on a brain-computer interface. The system includes: a non-invasive brain-computer interface intention recognition module, comprising an EEG signal acquisition array, a signal preprocessing unit, and a motor intention classifier, used to acquire the user's motor cortex EEG signals, identify at least one motor intention among five-finger grasping, thumb-to-finger, and release / relaxation, and output corresponding control commands; and a hand rehabilitation execution module, comprising a flexible fiber-reinforced pneumatic actuator, a multi-finger independent pneumatic drive unit, and an EEG signal cross-validation and tactile feedback unit, which, together with the non-invasive... The brain-computer interface intent recognition module is wirelessly connected and, upon receiving control commands, drives the affected hand to perform finger flexion and extension rehabilitation training. The upper limb rehabilitation training module includes a three-dimensional force-sensing robotic arm, a mirror therapy VR display unit, and daily life movement simulation scenarios, providing patients with multi-plane strength and coordination rehabilitation training for the upper limbs. The AI ​​rehabilitation data integration and cloud management platform includes a patient's personal health record layer, an AI personalized engine with automatic rehabilitation prescription adaptation capabilities, a remote physician access platform, and a patient's family monitoring interface. It records user data throughout the entire rehabilitation cycle and generates quantitative rehabilitation plans, achieving closed-loop management of the entire path from perception to execution to evaluation to feedback.

[0006] Optionally, in the non-invasive brain-computer interface intent recognition module, EEG signal preprocessing includes bandpass filtering (0.5~40Hz), power frequency notch filtering, and artifact removal.

[0007] Optionally, the motion intent classifier uses a convolutional neural network model to map EEG signal features to motion intent categories, and the classifier is trained using clinical EEG data.

[0008] Optionally, the hand rehabilitation execution module integrates an electromyography (EMG) sensor for collecting surface EMG signals of the wrist flexors and extensor muscles on the affected side. The EMG signals are cross-validated with the EEG intentions recognized by the non-invasive brain-computer interface intention recognition module to reduce the false positive rate.

[0009] Optionally, the hand rehabilitation execution module provides three adaptive training modes: passive mode, assisted mode, and resistance mode, which can be automatically switched according to the user's Brunnstrom stage or specified by a rehabilitation therapist through remote prescription.

[0010] Optionally, the upper limb rehabilitation training module further includes a functional electrical stimulation unit, which outputs biphasic constant current pulses to the target muscle group within a predetermined time delay after the BCI detects the movement intention and within a preset time period to enhance neuromuscular activation.

[0011] Optionally, the simulated daily life actions include a library of six categories of daily actions: drinking water from a cup, turning on a switch or knob, picking up chopsticks to grab objects, carrying a bag and walking, buttoning clothes, and picking up coins. The training is guided by virtual reality or augmented reality.

[0012] Optionally, the AI ​​personalization engine employs a prognostic analysis algorithm based on a federated learning architecture to dynamically predict the optimal training path based on the patient's historical rehabilitation data.

[0013] Optionally, the limb function closed-loop collaborative rehabilitation system also includes a lower limb gait coordination training module. This module detects EMG signals of the flexor muscle groups and gait cycle posture parameters, and links them with the motor cortex excitability signals detected by the brain-computer interface to synchronously provide gait assistance torque.

[0014] On the other hand, the present invention also provides an application of a brain-computer interface-based closed-loop collaborative rehabilitation system for limb function in the preparation of rehabilitation products for the entire life cycle and multiple scenarios of upper limb daily living activities after cerebral infarction.

[0015] Through the above technical solutions, this invention accurately identifies fine motor intentions such as "five-finger grasping" using a non-invasive BCI module, and performs EEG-EMG dual-modal cross-validation using electromyography sensors, achieving a fundamental shift from passive traction to intention-driven processes. This effectively addresses the clinical pain point of traditional rehabilitation equipment, which struggles to activate neural remodeling due to its passive training mode. By constructing a modular execution terminal integrating a soft robotic glove, a three-dimensional force-sensing robotic arm, and VR daily scene simulation, and in conjunction with the federated learning prognostic algorithm of an AI cloud platform, rehabilitation training has expanded from single movements to cover all daily functions such as holding a cup to drink water and buttoning clothes. This breaks down the technical barriers that limit high-end rehabilitation equipment to hospital departments and meets the application needs of home and community scenarios. Through AI rehabilitation data integration and cloud management platform for dynamic tracking of patient data throughout the entire life cycle, and the linkage design of the lower limb gait coordination module, a closed-loop management of the entire path—perception, execution, assessment, and feedback—is achieved. This fills the gap of fragmented existing rehabilitation processes and achieves the goal of providing personalized and continuous rehabilitation prescriptions for patients at different stages, significantly improving the overall rehabilitation efficiency and quality of life of stroke patients.

[0016] Other features and advantages of the present invention will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate embodiments of the present invention and form part of the specification. They are used together with the following detailed description to explain the embodiments of the present invention, but do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the overall structure of the brain-computer interface-based closed-loop collaborative rehabilitation system for limb function of the present invention. Figure 2 This is a structural layout diagram of the upper limb rehabilitation training module in this invention; Figure 3This is an exploded structural diagram of the soft robot rehabilitation glove structure in this invention; Figure 4 This is a closed-loop flowchart of brain-computer interface intent recognition and rehabilitation execution in this invention; Figure 5 This is a schematic diagram of the architecture of the AI ​​rehabilitation data integration and cloud management platform in this invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0019] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with the relevant provisions of national laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0020] Please refer to Figure 1 This invention provides a brain-computer interface-based closed-loop collaborative rehabilitation system for limb function. The system includes a non-invasive brain-computer interface intention recognition module, a hand rehabilitation execution module, an upper limb rehabilitation training module, a lower limb gait coordination training module, and an AI rehabilitation data integration and cloud management platform. The non-invasive brain-computer interface intention recognition module may include an EEG signal acquisition array, a signal preprocessing unit, and a motor intention classifier. It is used to acquire the user's motor cortex EEG signals, identify at least one motor intention among five-finger grasping, thumb-to-finger, and release / relaxation, and output corresponding control commands.

[0021] Specifically, the non-invasive brain-computer interface intention recognition module (BCI) adopts a wearable headband structure, worn on the forehead region and corresponding positions of the parietal motor cortex of the patient's head. It includes an EEG signal acquisition array, a signal conditioning unit, a wireless communication unit, and a power supply unit. The EEG signal acquisition array can adopt a dry electrode structure, which is made of conductive polymer composite material or metal-plated flexible electrodes to reduce the dependence on conductive gel in the use of traditional wet electrodes and improve the convenience of long-term wear in home settings.

[0022] In a preferred embodiment of the present invention, the EEG signal acquisition array includes 8 to 16 dry electrode channels, each covering the motor cortex region, with a focus on acquiring motor-related EEG signals corresponding to the C3 and C4 regions. The sampling frequency can be set to 250Hz to 500Hz to obtain EEG change information generated during motor imagery.

[0023] In this embodiment of the invention, the signal preprocessing unit may include an instrument amplification module, a digital filtering module, a power frequency interference suppression module, and a motion artifact removal module. The digital filtering module performs bandpass filtering on the acquired EEG signals, for example, using a frequency range of 0.5Hz to 40Hz for effective signal extraction, achieving a signal-to-noise ratio ≥80dB. The power frequency interference suppression module reduces periodic noise caused by the external electromagnetic environment. The motion artifact removal module eliminates artifact interference caused by slight head movements, muscle activity, and other factors.

[0024] Please refer to Figure 4 In this embodiment of the invention, the motion intention classifier employs a deep learning-based motion intention recognition model, such as a one-dimensional convolutional neural network model (1D-CNN, which classifies EEG signals into five-finger grasping intention, thumb-to-finger intention, and release / relaxation intention; the Motocimagery-BC system can decode three discrete grasping strength levels of 20%, 50%, and 80% from 64-channel EEG signals; and a Bluetooth LE5.3 wireless transmission module transmits the recognized intention signal to the rehabilitation execution module in real time). The motion intention classifier extracts features from the preprocessed EEG signal, fusing and analyzing the temporal and frequency domain variation features, as well as the spatial correlation features between different EEG channels, and maps the EEG signal to the corresponding motion intention category. The motion intention classifier can utilize pre-collected EEG training data for model training, continuously optimizing network parameters to improve the adaptability of motion intention recognition among different patients. The recognition results output by the motion intention classifier are transmitted to the hand rehabilitation execution module via a wireless communication unit. The wireless communication unit can use low-power Bluetooth communication to achieve real-time data interaction between the brain-computer interface intention recognition module and the rehabilitation execution module.

[0025] Please refer to Figure 3 In this embodiment of the invention, the hand rehabilitation execution module may include a flexible fiber reinforced pneumatic actuator, a multi-finger independent pneumatic drive unit, and an EEG signal cross-validation and tactile feedback unit, which is wirelessly connected to a non-invasive brain-computer interface intention recognition module. After receiving control commands, it drives the affected hand to perform finger flexion and extension rehabilitation training.

[0026] Specifically, the hand rehabilitation execution module adopts a soft robotic rehabilitation glove structure, including a flexible fiber-reinforced pneumatic actuator set on the back of the fingers and a pneumatic drive component for controlling air pressure changes. The flexible fiber-reinforced pneumatic actuator includes a flexible silicone body, a fiber-reinforced constraint layer and multiple series air chamber structures. There are 23 series air chambers on the back of the fingers. When inflated, it drives finger flexion, and when deflated, it drives finger extension. Each finger can be controlled independently.

[0027] In a preferred embodiment of the invention, each finger is provided with an independent pneumatic actuation channel. The series-connected air chambers generate a bending driving force when inflated, enabling the finger to flex. When deflated, the finger extends through elastic recovery. The flexible silicone body can be configured with different hardness ranges to meet the comfort needs of different patients. For example, a flexible material with a Shore hardness range of 20A to 40A can be used, allowing the actuator to balance driving force output capability and wearing comfort. The pneumatic drive assembly includes a miniature air pump, an electromagnetic control valve, and a pressure sensor. Different levels of assistance force are controlled by adjusting the air pump output pressure. For example, in assisted training mode, different levels of assistance force are output according to the patient's exercise capacity; in resistance training mode, the degree of active muscle involvement is increased by increasing the resistance output.

[0028] In this embodiment of the invention, the hand rehabilitation execution module integrates an electromyography (EMG) signal-assisted closed-loop tactile feedback unit. The EMG signal-assisted closed-loop tactile feedback unit includes a surface EMG sensor located in the patient's forearm region. This sensor collects EMG signals generated by target muscle groups such as the wrist flexors and extensor digitorum muscles on the affected side. The EMG signals are cross-validated with the brainwave motor intention output by the non-invasive brain-computer interface intention recognition module. When the brainwave recognition result matches the EMG activity state, the corresponding action control weight is increased; when there is a deviation, the motor intention recognition result is corrected to reduce the probability of erroneous actions. The EMG acquisition frequency can be set according to the sensor performance, for example, using a range of 500Hz to 2000Hz for EMG signal acquisition.

[0029] In this embodiment of the invention, the hand rehabilitation execution module provides three adaptive training modes: passive mode, assisted mode, and resistance mode. These modes can be automatically switched according to the user's Brunnstrom stage or specified by a rehabilitation therapist via remote prescription.

[0030] Specifically, in passive training mode, for patients with weak motor skills, the system controls a pneumatic actuator to complete a full grasping and releasing motion according to a preset movement trajectory, providing joint mobility training and visual-motor feedback stimulation. In assisted training mode, the system generates auxiliary control parameters based on the motor intention recognized by the brain-computer interface. After the patient actively generates a motor intention, the pneumatic actuator provides auxiliary driving force to complete the target action. For example, the system dynamically adjusts the assistance ratio based on the patient's current muscle strength, gradually reducing the assistance force and guiding the patient to increase their active motor participation. In active resistance mode, the system outputs impedance control parameters based on the patient's motor completion, providing adjustable resistance to train the patient's active muscle strength and exercise endurance.

[0031] Please refer to Figure 2In this embodiment of the invention, the upper limb rehabilitation training module may include a three-dimensional force-sensing robotic arm, a mirror therapy VR display unit, and a daily life movement simulation scenario, providing patients with multi-plane strength and coordination rehabilitation training for the upper limbs.

[0032] Specifically, the three-dimensional force-sensing robotic arm includes a multi-degree-of-freedom motion mechanism and force-sensing components to detect the force, direction of movement, and spatial position changes during the patient's upper limb movement. It can provide six-dimensional tensile / resistance / support forces from 0 to 50 N, and monitor the patient's three-plane mobility in real time. The robotic arm can provide assistive, supportive, or resistive forces in different directions and adjust output parameters according to the patient's training status. The mirror therapy VR display unit uses virtual reality or augmented reality devices to display images of the healthy limb's movements, inducing motor imagery of the affected side and synchronously driving the robotic arm / glove. Robot-assisted mirror therapy (RMT) has been proven to be highly safe and has high compliance, and is particularly beneficial for patients with severe movement disorders.

[0033] In this embodiment of the invention, the upper limb rehabilitation training module further includes a functional electrical stimulation (FES) unit. This unit uses surface electrostimulation to stimulate key muscle groups such as the extensor digitorum, lateral wrist flexor, and biceps brachii. The FES unit synchronously triggers stimulation of the target muscle groups based on the movement intention signal output from the brain-computer interface (BCI). The FES unit includes stimulation electrodes, a constant current pulse output module, and a stimulation control module. When the BCI detects a movement intention, it outputs biphasic constant current pulses to the target muscle groups within a preset response time. For example, it discharges synchronously with the BCI decision, emitting 0-60mA biphasic constant current pulses within 200 milliseconds after EEG recognition of "thought grasping," thereby enhancing muscle arousal and neuromuscular junction signal transmission. The FES stimulation parameters are dynamically adjusted by the closed-loop control module based on electromyographic feedback and training effects to enhance neuromuscular activation.

[0034] In a preferred embodiment of the present invention, the daily living action simulation scenario includes multiple daily living action training scenarios, and guides the patient to complete the target action training through a virtual reality display device or an augmented reality interactive device.

[0035] Specifically, the daily living skills training scenarios include a library of actions such as drinking water from a cup, turning on and off switches, picking up chopsticks, carrying a bag and walking, buttoning clothes, and picking up coins. These scenarios are then simulated using virtual reality guidance to reinforce independent living skills. Virtual reality composite task training has been proven to significantly improve the dexterity of the affected hand, with therapeutic effects lasting for more than 6 months. Patients complete action tasks in a virtual environment using VR or AR interactive devices. The system simultaneously collects the completion time, trajectory, force, and accuracy of the actions, and uploads the relevant data to an AI rehabilitation data integration and cloud management platform.

[0036] In this embodiment of the invention, the lower limb gait coordination training module is an optional configuration module. This module detects the EMG signals of the flexor muscle group and the gait cycle posture parameters, and links them with the motor cortex excitability signals detected by the brain-computer interface to synchronously provide gait assistance torque.

[0037] Specifically, a lightweight exoskeleton device worn on the flexor muscles and ankle joint for real-time electromyography (EMG) and posture detection can be linked with the aforementioned brain-computer interface motor cortex intensity signals (synchronously activating lower limb muscle groups to improve the symmetry of the affected leg swing and forward thrust during walking). The lower limb gait coordination training module includes an EMG detection device, a posture detection device, and an auxiliary drive mechanism. The EMG detection device is used to collect the activity status of target lower limb muscle groups, and the posture detection device is used to acquire information on joint angles, gait cycles, and changes in body posture. Based on the motor intention detected by the brain-computer interface and the lower limb movement status, the system controls the auxiliary drive mechanism to output corresponding auxiliary torque, achieving coordinated training of upper and lower limb motor functions.

[0038] Please refer to Figure 5 In this embodiment of the invention, the AI ​​rehabilitation data integration and cloud management platform may include a patient personal health record layer, an AI personalization engine with automatic rehabilitation prescription adaptation capability, a physician remote access platform, and a patient family monitoring interface, record user data throughout the entire rehabilitation cycle and generate quantitative rehabilitation plans, and realize closed-loop management of the entire path of "perception-execution-evaluation-feedback".

[0039] Specifically, the AI ​​rehabilitation data integration and cloud management platform adopts a multi-layered architecture consisting of home terminals, mobile applications, and cloud servers. The patient's personal health record layer stores the patient's training records, including training sessions, movement completion rate, EEG recognition accuracy, EMG trends, and rehabilitation evaluation indicators. It also locally stores encrypted rehabilitation logs (e.g., a complete historical record of training movement sets, strength values, EEG classification accuracy, completion rate, FMA scores, etc.) for ≥90 days. The AI ​​personalization engine uses a predictive model based on machine learning or federated learning architecture to predict the next stage of training trends based on the patient's historical rehabilitation data. It can generate daily difficulty coefficients based on user performance and automatically amplify training intensity levels (e.g., reducing the intensity when the force is insufficient and increasing it when it is accurate and forceful). The system utilizes federated learning-based prognostic curve analysis to optimize rehabilitation strategies. The federated learning architecture optimizes rehabilitation prediction models through model parameter sharing without directly uploading patients' original private data. The cloud platform employs encrypted data transmission methods, such as SSL / TLS communication protocols and data access permission management mechanisms, to enhance the security of patient rehabilitation data. Rehabilitation physicians or therapists can view patient rehabilitation activity data via mobile devices or the web, issue prescriptions online, and provide remote guidance. It also offers rehabilitation check-in absence reminders, continuous training cycle reminders, and health monitoring items to maintain consistent adherence to the rehabilitation plan.

[0040] In a preferred embodiment of the invention, the AI ​​personalization engine dynamically adjusts training parameters based on the patient's rehabilitation status: when a decrease in the action completion rate is detected, the training difficulty is reduced or the assistance ratio is increased; when an increase in active movement ability is detected, the resistance training ratio is increased; when a decrease in the accuracy of movement intention recognition is detected, the model parameters are re-optimized. The physician remote access platform allows rehabilitation physicians to view the patient's training status and remotely adjust the training prescription; the patient's family monitoring interface provides training reminders, abnormal status warnings, and rehabilitation progress viewing functions.

[0041] In a preferred embodiment, using a soft rehabilitation glove and EEG frontal / lobular dry electrodes, the user, wearing an 8-channel EEG dry electrode headband and soft rehabilitation gloves, completes a 15-second resting baseline measurement (e.g., LDL EEG background wave recording) according to on-screen instructions. The user then performs three rounds of grasping and lifting intention visualization training, 10 repetitions per round, and the system records the user's motor imagery classifier parameters. The daily rehabilitation plan includes: 10 minutes of hand and wrist range of motion training (passive assisted mode); 15 minutes of grasping and holding training (user actively combined with glove-assisted mode); 10 minutes of daily living activity simulation using a virtual reality bridging rehabilitation training platform; and 5 minutes of FES electrical stimulation consolidation training (a brief muscle arousal is initiated after all training is completed). With a usage frequency of at least 5 times per week and an average training duration of at least 45 minutes, after three consecutive weeks of use (ideal estimate), the expected improvement in Ful-Meer upper limb motor function score is ≥1 point (Minimum Clinically Important Difference (MCID) is approximately 68 points). In the active-assisted training mode, a brain-computer interface recognition accuracy of approximately ≥75% is sufficient to drive non-invasive control and active subjective participation. This data point aligns with the 2026 consensus standard: active-participatory rehabilitation robots enhance active training and the core of brain plasticity.

[0042] In a preferred embodiment, the entire process of remote monitoring and prescription issuance is achieved via a mobile phone or a rehabilitation physician in a rehabilitation hospital through a hand glove integrated with a BMI device. After daily training, the muscle activation effect can be consolidated through FES (Follicular Unit Extraction). It is equipped with functions such as automatic scoring of indicators like FMA-UE (Follicular Unit Analysis) and electromyography data reporting sessions. This is particularly suitable for hemiplegic patients with partial function on the affected side but difficulty lifting objects.

[0043] On the other hand, embodiments of the present invention also provide an application of a brain-computer interface-based closed-loop collaborative rehabilitation system for limb function in the preparation of rehabilitation products for the entire life cycle and multiple scenarios of upper limb daily living activities after cerebral infarction.

[0044] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0046] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0048] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0049] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0050] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0051] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0052] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A closed-loop collaborative rehabilitation system for limb function based on brain-computer interface, characterized in that, The closed-loop collaborative rehabilitation system for limb function includes: The non-invasive brain-computer interface intention recognition module includes an EEG signal acquisition array, a signal preprocessing unit, and a motor intention classifier. It is used to acquire the user's motor cortex EEG signals, identify at least one motor intention among five-finger grasping, thumb-to-finger, and release-relaxation, and output the corresponding control command. The hand rehabilitation execution module includes a flexible fiber-reinforced pneumatic actuator, a multi-finger independent pneumatic drive unit, and an EEG signal cross-validation and tactile feedback unit. It is wirelessly connected to the non-invasive brain-computer interface intention recognition module and drives the affected hand to perform finger flexion and extension rehabilitation training after receiving control commands. The upper limb rehabilitation training module includes a three-dimensional force-sensing robotic arm, a mirror therapy VR display unit, and daily life movement simulation scenarios, providing patients with multi-plane strength and coordination rehabilitation training for the upper limbs. The AI ​​rehabilitation data integration and cloud management platform includes a patient personal health record layer, an AI personalization engine with automatic rehabilitation prescription adaptation capabilities, a physician remote access platform, and a patient family monitoring interface. It records user data throughout the entire rehabilitation cycle and generates quantitative rehabilitation plans, achieving closed-loop management of the entire path from perception to execution to evaluation to feedback.

2. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, In the non-invasive brain-computer interface intent recognition module, EEG signal preprocessing includes bandpass filtering (0.5~40Hz), power frequency notch filtering, and artifact removal.

3. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The motion intent classifier uses a convolutional neural network model to map EEG signal features to motion intent categories, and the classifier is trained using clinical EEG data.

4. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The hand rehabilitation execution module integrates an electromyography (EMG) sensor for collecting EMG signals from the surface of the wrist flexors and extensor muscles on the affected side. These EMG signals are cross-validated with the EEG intentions recognized by the non-invasive brain-computer interface intention recognition module to reduce the false positive rate.

5. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The hand rehabilitation execution module provides three adaptive training modes: passive mode, assisted mode, and resistance mode. These modes can be automatically switched according to the user's Brunnstrom stage or specified by a rehabilitation therapist via remote prescription.

6. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The upper limb rehabilitation training module also includes a functional electrical stimulation unit, which outputs biphasic constant current pulses to the target muscle group within a predetermined time delay after the non-invasive brain-computer interface intention recognition module detects the movement intention and within a preset time period, so as to enhance neuromuscular activation.

7. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The simulated daily life actions include six categories of daily actions: drinking water from a cup, turning on and off a knob, picking up chopsticks, carrying a bag and walking, buttoning clothes, and picking up coins. The training is guided by virtual reality or augmented reality.

8. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The AI ​​personalization engine employs a prognostic analysis algorithm based on a federated learning architecture to dynamically predict the optimal training path based on the patient's historical rehabilitation data.

9. The brain-computer interface-based closed-loop collaborative rehabilitation system for limb function according to claim 1, characterized in that, The closed-loop collaborative rehabilitation system for limb function also includes a lower limb gait coordination training module. This module detects EMG signals of the flexor muscle groups and gait cycle posture parameters, and links them with the excitability signals of the motor cortex detected by the brain-computer interface to synchronously provide gait assistance torque.

10. The application of a brain-computer interface-based closed-loop collaborative rehabilitation system for limb function as described in any one of claims 1-9 in the preparation of rehabilitation products for the entire life cycle and multiple scenarios of upper limb daily living activities after cerebral infarction.

Images