Hand rehabilitation training system supporting finger joint angle collection and finger pulp force feedback

Abstract

The invention belongs to the technical field of hand rehabilitation training, and particularly relates to a hand rehabilitation training system supporting finger joint angle acquisition and finger pulp force feedback, which comprises a hardware module, and the hardware module comprises a wearable acquisition device, an interactive display device and a data processing terminal. The data acquisition module synchronously acquires and preprocesses electromyographic signals, finger joint angle data and finger pulp force feedback data, the data acquisition module synchronously acquires and preprocesses the electromyographic signals, the finger joint angle data and the finger pulp force feedback data, and the data acquisition module synchronously acquires and preprocesses the electromyographic signals, the finger joint angle data and the finger pulp force feedback data, so that guarantee is provided for subsequent training and calibration; then the multi-modal closed-loop calibration module constructs a three-closed-loop model by means of a Kalman filtering algorithm, the acquisition error is corrected in real time, the data reliability is greatly improved, a customized scheme is provided for different hand trauma and rehabilitation stages through a trauma exclusive force feedback model library, and the problem of insufficient training or excessive injury caused by a general model is avoided.

Description

Technical Field

[0001] This invention belongs to the field of hand rehabilitation training technology, and in particular relates to a hand rehabilitation training system that supports the acquisition of finger joint angles and feedback of fingertip force. Background Technology

[0002] Current hand rehabilitation training systems suffer from numerous technical shortcomings, failing to meet the demands for precise and personalized rehabilitation. Firstly, data acquisition is often limited to a single dimension, focusing primarily on finger joint angles or fingertip force feedback, lacking simultaneous acquisition and fusion processing of electromyographic signals, joint angles, and fingertip force feedback. Furthermore, single-sensor systems are susceptible to interference, resulting in significant data errors and failing to provide reliable data for rehabilitation assessment. Secondly, training programs lack specificity; generic models do not differentiate between hand injury types (such as metacarpal fractures and tendon injuries) and rehabilitation stages, easily leading to insufficient training intensity or excessive force causing secondary injuries. Thirdly, training methods are often monotonous, primarily involving repetitive mechanical movements, resulting in low patient compliance and a lack of real-time feedback and effective safety warning mechanisms, making it difficult to promptly avoid risks associated with improper movements. Therefore, we propose a hand rehabilitation training system that supports both finger joint angle acquisition and fingertip force feedback. Summary of the Invention

[0003] The purpose of this invention is to address the aforementioned technical problems by providing a hand rehabilitation training system that supports finger joint angle acquisition and fingertip force feedback.

[0004] In view of this, the present invention provides a hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback, comprising a hardware module including a wearable acquisition device, an interactive display device, and a data processing terminal; a data acquisition module that simultaneously acquires electromyographic signals, finger joint angle data, and fingertip force feedback data and performs preprocessing; a multimodal closed-loop calibration module that constructs an electromyographic-angle-force feedback three-loop calibration model based on the Kalman filter algorithm to correct acquisition errors in real time; a trauma-specific force feedback model library module that pre-stores at least four types of hand trauma-specific force feedback curve models, and the models are associated with the trauma type and rehabilitation stage; a training execution module that constructs a virtual training scene based on Unity3D to achieve real-time data feedback and security warnings; and a data interaction and storage module that implements encrypted data transmission, hierarchical storage, and access control.

[0005] Furthermore, the wearable acquisition device includes an electromyography sensor that fits the muscles related to finger movement, an infrared + IMU fusion acquisition module deployed at each finger joint, and pressure sensors distributed on the fingertips of the thumb, index finger, and middle finger; the finger joint angle acquisition range of the infrared + IMU fusion acquisition module is 0-180°.

[0006] Furthermore, the three-loop calibration model of the multimodal closed-loop calibration module includes an electromyography-angle calibration closed loop, which corrects finger joint angle data with reference to electromyography characteristic parameters; an angle-force feedback calibration closed loop, which corrects the force feedback output value based on the physical correlation between finger joint angle and fingertip force feedback; and a force feedback-electromyography calibration closed loop, which verifies the validity of electromyography signals with fingertip force feedback data.

[0007] Furthermore, the trauma-specific force feedback curve model includes four types of models: metacarpal fracture, tendon injury, nerve injury, and hand burn. Each model includes fingertip force feedback threshold and resistance change rate parameters. Among them, the tendon injury model uses a step-like triggering resistance change, and the nerve injury model uses a gradual pressure triggering.

[0008] Furthermore, the virtual training scenarios of the training execution module include a virtual kitchen and an office, and the scenarios include daily action tasks such as grasping, pinching, and rotating. The safety warning is implemented in three ways: force feedback glove vibration, interactive display device interface pop-up window, and voice prompt, and different warning types correspond to different vibration modes.

[0009] Furthermore, the data acquisition module uses a dual-sensor fusion scheme of infrared positioning and inertial measurement unit to acquire finger joint angles. Each finger's metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint are independently equipped with acquisition units to synchronously acquire bending or extension angles, angular velocities, and angular acceleration data of each joint.

[0010] Furthermore, the fingertip force feedback acquisition of the data acquisition module is realized through distributed pressure sensors. The sensors are attached to the fingertip contact areas of the thumb, index finger, and middle finger, and can capture the pressure peak, pressure change rate, and pressure distribution characteristics in real time during the grasping process. Based on the acquired force feedback data, the training execution module dynamically outputs the corresponding resistance through the micro-drive unit of the force feedback glove.

[0011] Furthermore, the finger joint angle acquisition data and fingertip force feedback data form a linkage control: when the finger joint angle reaches the preset threshold of the external injury-specific force feedback curve model, the fingertip force feedback automatically triggers the corresponding resistance level.

[0012] The beneficial effects of this invention are:

[0013] This system simultaneously acquires and preprocesses three types of data—electromyography, joint angle, and fingertip force feedback—through a data acquisition module, ensuring data quality for subsequent training and calibration. A multimodal closed-loop calibration module utilizes a Kalman filter algorithm to construct a three-loop model, correcting acquisition errors in real time and significantly improving data reliability. A trauma-specific force feedback model library provides customized solutions for different hand injuries and rehabilitation stages, avoiding undertraining or over-injury issues caused by generic models. A virtual training scenario built using Unity3D enhances training engagement and immersion, improving patient compliance. Simultaneously, the data interaction and storage module implements encrypted transmission, hierarchical storage, and access control, protecting patient privacy while facilitating hierarchical access and management by medical staff. Applicable to hospitals, rehabilitation centers, and home rehabilitation settings, this system significantly improves the scientific rigor, relevance, and safety of hand rehabilitation training. Attached Figure Description

[0014] Figure 1 This is a flowchart of a hand rehabilitation training system that supports finger joint angle acquisition and fingertip force feedback, as proposed in this invention. Detailed Implementation

[0015] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0016] In the description of this application, it should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. For ease of description, the dimensions of the various parts shown in the drawings are not drawn to actual scale. Techniques, methods, and devices known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and devices should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters in the following drawings denote similar items; therefore, once an item is defined in one drawing, it need not be further discussed in subsequent drawings.

[0017] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and are not limited in number; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0018] It should be noted that in the description of this application, the directional terms such as "front, back, up, down, left, right", "horizontal, vertical, horizontal" and "top, bottom" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this application and simplifying the description. Unless otherwise stated, these directional terms do not indicate or imply that the device or element referred to must have a specific orientation or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the scope of protection of this application. The directional terms "inner" and "outer" refer to the inner and outer contours relative to the outline of each component itself.

[0019] It should be noted that, in this application, 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 a process, method, article, or apparatus. Without further limitations, 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. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0020] Reference Figure 1A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback includes a hardware module comprising a wearable acquisition device, an interactive display device, and a data processing terminal; a data acquisition module that simultaneously acquires electromyographic signals, finger joint angle data, and fingertip force feedback data and performs preprocessing; a multimodal closed-loop calibration module that constructs an electromyographic-angle-force feedback three-loop calibration model based on the Kalman filter algorithm to correct acquisition errors in real time; a trauma-specific force feedback model library module that pre-stores at least four types of hand trauma-specific force feedback curve models, and the models are associated with the trauma type and rehabilitation stage; a training execution module that constructs a virtual training scene based on Unity3D to achieve real-time data feedback and security warnings; and a data interaction and storage module that implements encrypted data transmission, hierarchical storage, and access control.

[0021] This application first initiates and establishes a communication connection between the wearable acquisition device, interactive display device, and data processing terminal in the hardware module, completing the device linkage initialization. Subsequently, the data acquisition module synchronously starts the acquisition of electromyography, finger joint angle, and fingertip force feedback data. The raw data undergoes noise reduction, filtering, and normalization preprocessing to remove invalid interference data and generate standardized data for subsequent modules. Next, the multimodal closed-loop calibration module drives the three-loop calibration model through a Kalman filter algorithm, performing real-time cross-validation and error correction on the three types of preprocessed data to output accurate data. Simultaneously, the trauma-specific force feedback model library module, based on the patient's entered trauma type and current... In the pre-rehabilitation phase, the system automatically matches a corresponding dedicated force feedback curve model, determines core parameters such as force feedback threshold and resistance change rate, and transmits them to the training execution module. The training execution module, based on a virtual training scene built with Unity3D, integrates the calibrated data with the matched model parameters, drives the action interaction in the virtual scene in real time, and synchronously monitors data anomalies during the training process, triggering a security warning mechanism. Finally, the data interaction and storage module encrypts and transmits all training data, stores it hierarchically according to patient privacy level and training cycle, and enables differentiated data access for medical staff and patients through an access control mechanism. At the same time, the data is retained for rehabilitation effect tracking and program optimization.

[0022] In the example of this application, the wearable acquisition device includes an electromyography sensor that fits the muscles related to finger movement, an infrared + IMU fusion acquisition module deployed at each finger joint, and pressure sensors distributed on the fingertips of the thumb, index finger, and middle finger; the finger joint angle acquisition range of the infrared + IMU fusion acquisition module is 0-180°.

[0023] As a preferred example of the present invention, the wearable data acquisition device achieves comprehensive capture of core hand rehabilitation data through a multi-sensor collaborative layout. Its core function lies in the targeted deployment and collaborative acquisition of each sensor. Electromyography (EMG) sensors are closely fitted to the muscles related to finger movement, capturing EMG signals generated during muscle contraction and relaxation in real time. These signals are then converted into electrical signals and transmitted to the data processing terminal, providing raw data for action intention recognition and rehabilitation effect evaluation. An infrared + IMU fusion acquisition module is deployed at each finger joint. The infrared sensor is responsible for static joint angle positioning, while the IMU inertial measurement unit captures dynamic joint motion parameters. The data from both are fused... The combined and complementary technologies effectively compensate for the measurement errors of single sensing methods. Furthermore, based on a 0-180° acquisition range, it fully covers the entire range of motion of finger joints, from flexion to extension, adapting to the range of motion requirements at different rehabilitation stages. Distributed pressure sensors precisely conform to the contact areas of the thumb, index, and middle fingers—key execution points for core hand movements such as grasping and pinching. The sensors can capture pressure signals in real time during movement, converting them into quantifiable data to provide precise input for fingertip force feedback control. The overall device features a lightweight design that conforms to the hand's contours without restricting movement, ensuring accurate data acquisition during natural patient training.

[0024] It is worth noting that the wearable collection device is made of a composite material of medical-grade silicone and lightweight carbon fiber, and is designed to fit the contours of the hand. The Velcro adjustable straps can be used to adapt to different hand shapes, and the device does not affect the natural flexion and extension of the fingers when worn.

[0025] The electromyography (EMG) sensor uses a dry electrode design, eliminating the need for conductive gel. It features three acquisition points that fit snugly against the muscles involved in finger movement, capturing real-time changes in electrical signals during muscle contraction and relaxation.

[0026] The infrared + IMU fusion acquisition module is deployed independently for each finger's metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint, totaling 15 groups. The IMU inertial measurement unit includes an accelerometer and a gyroscope to capture joint dynamic angular velocity and angular acceleration data; the two are integrated through a weighted fusion algorithm, with an acquisition range of 0-180°, covering the entire range of finger flexion and extension.

[0027] The pressure sensor uses a thin-film pressure sensor that fits the contact area of ​​the thumb, index finger, and middle fingertips. It can capture pressure peaks, pressure change rates, and pressure distribution characteristics, avoiding lag in force feedback control due to delays.

[0028] The interactive display device uses a 10.1-inch touchscreen to display virtual training scenarios, real-time data curves, rehabilitation progress and early warning information. It supports multi-touch operation, making it convenient for patients to adjust training parameters independently.

[0029] The data processing terminal uses an embedded processor and integrates a Bluetooth and Wi-Fi dual-mode communication module. It is responsible for receiving data collected by wearable devices, running calibration algorithms, and driving virtual scenes.

[0030] The synchronous acquisition mechanism adopts a hardware-triggered synchronization method. The data processing terminal sends a synchronous clock signal to drive the electromyography sensor, infrared + IMU fusion module, and pressure sensor to start acquisition simultaneously, ensuring that the timestamps of the three types of data are aligned and avoiding calibration deviations caused by asynchronous data.

[0031] Electromyographic signals are first de-denoised using a wavelet denoising algorithm to remove baseline drift and electromagnetic interference, then filtered by a Butterworth low-pass filter to extract the effective signal, and finally the min-max normalization method is used to map the signal to the [0,1] interval to eliminate individual differences in muscle characteristics.

[0032] The joint angle data is used to detect outliers in the fused infrared and IMU data, removing abnormal data caused by equipment displacement or sudden jitter. The data is then smoothed by moving average filtering to output a continuous and stable sequence of joint angles, angular velocities, and angular accelerations.

[0033] The fingertip force feedback data is filtered by median to remove instantaneous pulse interference. The peak pressure, steady-state pressure and rate of change are calculated and normalized to maintain dimensional consistency with the angle data.

[0034] After preprocessing, invalid data is removed by threshold judgment, a standardized dataset is generated, and stored in a temporary cache for use by the calibration module.

[0035] In the example of this application, the three-loop calibration model of the multimodal closed-loop calibration module includes an electromyography-angle calibration closed loop, which corrects finger joint angle data with reference to electromyography characteristic parameters; an angle-force feedback calibration closed loop, which corrects the force feedback output value based on the physical correlation between finger joint angle and fingertip force feedback; and a force feedback-electromyography calibration closed loop, which verifies the validity of electromyography signals with fingertip force feedback data.

[0036] As a preferred example of the present invention, the multimodal closed-loop calibration module is based on a three-loop calibration model constructed using the Kalman filter algorithm. It achieves dynamic calibration of three types of data through mutual verification and cyclic correction logic. The core lies in forming a closed-loop feedback adjustment mechanism. During the electromyography-angle calibration closed-loop operation, the electromyography characteristic parameters preprocessed by the data acquisition module are used as a reference. Combined with the physiological correlation between muscle activity and joint movement, the synchronously acquired finger joint angle data is corrected to eliminate angle measurement errors caused by factors such as device displacement and motion interference, ensuring that the angle data is consistent with the actual muscle activity state. The angle-force feedback calibration closed-loop relies on… The physical relationship between finger joint angle and fingertip force feedback is established. Based on the calibrated joint angle data, the output value of fingertip force feedback is adjusted to avoid the mismatch between force feedback and the actual joint movement state, thus improving the rationality of force feedback control. The force feedback-EMG calibration closed loop uses the accurately collected fingertip force feedback data as the verification standard to determine whether the synchronously acquired EMG signal is a valid signal, eliminating invalid EMG data generated by electromagnetic interference and involuntary muscle contraction, ensuring that subsequent training programs are based on reliable data. The three closed loops operate synchronously and interact with each other to form a three-dimensional calibration system, realizing real-time correction and dynamic optimization of data errors.

[0037] It is worth noting that the Kalman filter parameters are set as follows: the state equation is constructed based on the muscle-joint dynamics model, which describes the physiological relationship between electromyography signal, joint angle, and finger force; the process noise matrix Q is set as a diagonal matrix; and the observation noise matrix R is set according to the sensor noise characteristics. The parameters are dynamically optimized through an adaptive adjustment mechanism to improve calibration robustness.

[0038] The electromyography-angle calibration closed loop uses preprocessed electromyography characteristic parameters as a reference benchmark and combines the physiological correspondence between muscle activity and joint movement to correct joint angle data, eliminating angle deviations caused by slight device displacement and poor skin contact, and ensuring that the angle data is consistent with the actual muscle activity state.

[0039] The angle-force feedback calibration closed loop relies on the physical relationship between finger joint angle and fingertip force. Based on the calibrated joint angle data, it adjusts the fingertip force feedback output value to avoid mismatch between force feedback and joint movement state due to sensor sensitivity differences, thereby improving the rationality of force feedback control.

[0040] The force feedback-EMG calibration closed loop uses the collected fingertip force feedback data as the verification standard to determine whether the synchronously acquired EMG signals are valid movement signals, and eliminates invalid data generated by electromagnetic interference and involuntary muscle contractions, ensuring that subsequent training programs are based on reliable data.

[0041] After three closed-loop calibrations, the output data is synchronously transmitted to the training execution module and the data storage module.

[0042] In the examples of this application, the trauma-specific force feedback curve model includes four types of models: metacarpal fracture, tendon injury, nerve injury, and hand burn. Each model includes fingertip force feedback threshold and resistance change rate parameters. Among them, the tendon injury model uses a step-like triggering resistance change, the nerve injury model uses a gradual pressure triggering, the threshold of the metacarpal fracture model increases linearly, and the threshold of the hand burn model decreases by 30% overall.

[0043] As a preferred example of the present invention, the core working principle of the trauma-specific force feedback model library module is to achieve personalized adaptation of rehabilitation training. The module pre-stores dedicated force feedback curve models for four types of hand injuries: metacarpal fractures, tendon injuries, nerve injuries, and hand burns. Each model is associated with the type of injury, the rehabilitation stage, and the corresponding force feedback threshold and resistance change rate parameters, forming a standardized model database. Once the patient activates the system, they input their injury type and current rehabilitation stage. The module uses a keyword matching algorithm to retrieve the corresponding customized model from the database. Based on the differentiated design of different injury models, the system outputs parameters as needed: the tendon injury model outputs stepwise resistance change parameters, gradually increasing the resistance trigger threshold according to the rehabilitation progress to avoid secondary injury caused by sudden force; the nerve injury model outputs gradually changing pressure trigger parameters, slowly adjusting the pressure feedback intensity to adapt to the decreased sensitivity of nerve perception; the metacarpal fracture model outputs linearly increasing threshold parameters, conforming to the recovery rhythm of bone load-bearing capacity during fracture healing; and the hand burn model automatically reduces the standard threshold by 30% and outputs it, protecting the burn wound by weakening the force feedback intensity. Finally, the matched model parameters are transmitted to the training execution module, providing a basis for customized training.

[0044] It is worth noting that the metacarpal fracture model is suitable for patients with fractures of the first to fifth segments of the metacarpal bone. The force feedback threshold is set to increase linearly according to the rehabilitation stage. The threshold in the acute phase is 30%-50% of that in the recovery phase, and the resistance change rate is 0.5 N / s to avoid excessive force on the fracture site. In the sequelae phase, the resistance change rate is increased to 1.0 N / s to gradually strengthen muscle strength.

[0045] Tendon injury model: Suitable for patients with flexor and extensor tendon injuries, it adopts a step-by-step trigger resistance change, increasing the resistance by 1 level for each week of rehabilitation training. Only 1-2 levels of resistance are activated in the acute phase, 3-4 levels are activated in the recovery phase, and 5 levels are activated in the sequelae phase. At the same time, a tendon stretching protection threshold is set to avoid excessive activity that could lead to re-injury of the tendon.

[0046] Nerve injury model: Suitable for patients with median and ulnar nerve injuries. Due to the decreased nerve perception sensitivity of patients, a gradual pressure trigger is used, with the resistance slowly increasing from 0N to the target threshold at an increase rate of 0.3N / s. At the same time, the fault tolerance range of the force feedback threshold is expanded to avoid movement deviation due to perception delay. During the training process, changes in electromyographic signals are monitored simultaneously to help judge the nerve recovery status.

[0047] Hand burn model: Suitable for patients with superficial second-degree to deep second-degree burns. To protect the burn wound, the standard force feedback threshold is reduced by 30% overall, the resistance change rate is controlled within 0.2 N / s, and the maximum force feedback value is limited to ≤20 N to avoid wound rupture under pressure. The model is synchronously linked to the wound healing status and the threshold parameters are dynamically adjusted.

[0048] After the patient starts the system, they enter the type of injury, the location of the injury, and the time of the operation. The system automatically determines the current rehabilitation stage, retrieves the corresponding exclusive model from the model library through a keyword matching algorithm, and outputs core parameters such as force feedback threshold, resistance change rate, and safety tolerance range, which are then transmitted to the training execution module.

[0049] In the example of this application, the virtual training scenarios of the training execution module include a virtual kitchen and an office. The scenarios include daily action tasks such as grasping, pinching, and rotating. Safety warnings are implemented in three ways: force feedback glove vibration, interactive display device interface pop-up, and voice prompts. Different warning types correspond to different vibration modes.

[0050] As a preferred example of the present invention, the training execution module constructs virtual kitchens, offices, and other scenes close to daily life based on the Unity3D engine. It presets daily action tasks such as grasping, pinching, and rotating. The patient wears a wearable data acquisition device and performs the corresponding actions in the real space. The joint angle and finger force feedback data collected by the device are calibrated and transmitted to the module in real time, driving the character in the virtual scene to complete synchronous actions, forming an interactive closed loop of real actions combined with virtual feedback. This helps the patient adapt to daily action needs during rehabilitation training. At the same time, the module monitors the matching degree between the training data and the parameters of the trauma-specific model in real time. When risk situations such as excessive force feedback or abnormal joint angle occur, a triple safety warning mechanism is triggered: the vibration unit built into the force feedback glove outputs the corresponding vibration mode according to the warning type; the interactive display device interface pops up a warning pop-up indicating the cause of the abnormality; and the voice module plays prompt information synchronously, reminding the patient to adjust the range and intensity of the movements from multiple dimensions to avoid secondary injury and ensure training safety.

[0051] It is worth noting that two core scenarios, a virtual kitchen and an office, are set up. Each scenario contains 10+ interactive tasks that are close to daily life and are divided into basic, intermediate, and advanced levels according to difficulty. Patients can choose according to their recovery stage or the system can automatically recommend tasks based on model parameters.

[0052] Virtual-real interaction mechanism: Calibrated joint angles and fingertip force data are transmitted to the engine in real time, driving the virtual hand model to complete synchronized movements. At the same time, the force feedback glove, through its built-in micro electromagnetic drive unit, dynamically outputs corresponding resistance based on the model parameters, helping patients build muscle memory. For example, when grasping a virtual water cup, as the finger bending angle increases, the force feedback resistance gradually increases, and the virtual water cup synchronously displays the force state, enhancing the immersive training experience.

[0053] The system monitors the matching degree between training data and model preset parameters in real time. When the following abnormal situations occur, a triple warning linkage is triggered:

[0054] Force feedback exceeding the preset threshold: When the force feedback glove emits a short vibration when it exceeds the preset threshold by 10%, a red pop-up window appears on the interactive screen indicating "Force exceeds the limit, it is recommended to reduce the force", and a voice prompt is broadcast simultaneously; when it exceeds the limit by 20%, the resistance is automatically reduced to a safe range, and the training is paused until the patient adjusts the movement.

[0055] Abnormal joint angle: When the joint angle exceeds the safe range of motion, the glove will emit a long vibration, a pop-up window will display the current angle and the safe range, and a voice prompt will say "Abnormal joint angle, please pay attention to the amplitude".

[0056] Equipment malfunction: When the sensor signal is interrupted or communication is abnormal, the glove will vibrate continuously, a pop-up window will indicate the type of malfunction, and the voice will guide the patient to check the equipment. At the same time, the collected data will be saved to avoid data loss.

[0057] In the example of this application, the data acquisition module for acquiring finger joint angles adopts a dual-sensor fusion scheme of infrared positioning + inertial measurement unit. The metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint of each finger are independently equipped with acquisition units, and the bending or extension angle, angular velocity and angular acceleration data of each joint are acquired synchronously.

[0058] As a preferred example of this invention, the finger joint angle acquisition adopts an infrared positioning + IMU fusion acquisition scheme. The infrared sensor, by emitting and receiving infrared signals, locates the static position of the joint and calculates the initial angle. The IMU inertial measurement unit simultaneously captures angular velocity and angular acceleration data during joint movement. A data fusion algorithm integrates the data from both types of sensors, overcoming the shortcomings of large dynamic measurement errors from a single infrared sensor and insufficient static positioning accuracy from a single IMU. This achieves accurate angle acquisition in both dynamic and static scenarios. Each finger's metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint have independently deployed acquisition units. Each unit synchronously starts acquisition work, capturing the flexion or extension angle, angular velocity, and angular acceleration data of the corresponding joint, avoiding the limitations of traditional overall acquisition methods that cannot distinguish the state of individual joints. The acquisition units independently transmit the joint data to the data processing terminal, where preprocessing forms a motion state dataset for each joint, facilitating real-time monitoring of the rehabilitation progress of individual joints by medical personnel.

[0059] In the example of this application, the fingertip force feedback acquisition of the data acquisition module is realized through distributed pressure sensors. The sensors are attached to the fingertip contact areas of the thumb, index finger, and middle finger, and can capture the pressure peak, pressure change rate, and pressure distribution characteristics in real time during the grasping process. Based on the acquired force feedback data, the training execution module dynamically outputs the corresponding resistance through the micro-drive unit of the force feedback glove.

[0060] As a preferred example of the present invention, the acquisition of fingertip force feedback data enables precise control of force feedback and enhancement of training effects. Distributed pressure sensors are deployed in close contact with the fingertips of the thumb, index finger, and middle finger to capture pressure signals in real time during the patient's grasping and pinching movements. These signals are converted into three types of quantitative data: pressure peak value, pressure change rate, and pressure distribution characteristics. This comprehensively reflects the patient's force control ability and movement standardization in grasping movements. After preprocessing, the data is transmitted to the training execution module and the multimodal closed-loop calibration module. The training execution module receives the calibrated force feedback data and, combined with the parameters of the trauma-specific model, dynamically outputs corresponding resistance through the micro-drive unit built into the force feedback glove. When the patient's force is insufficient, the drive unit reduces the resistance to encourage the completion of the movement; when the force is close to the threshold, the resistance is appropriately increased to enhance the training effect; when the force exceeds the standard, the resistance is quickly adjusted and an early warning is triggered. This dynamic resistance output mode can flexibly adapt to the training needs of different rehabilitation stages, helping patients gradually improve their force control ability and muscle memory, while avoiding poor training effects or injury risks caused by fixed resistance.

[0061] In the example of this application, the finger joint angle acquisition data and the fingertip force feedback data form a linkage control: when the finger joint angle reaches the threshold preset by the external injury-specific force feedback curve model, the fingertip force feedback automatically triggers the corresponding resistance level.

[0062] As a preferred example of this invention, the linkage control of finger joint angle and fingertip force feedback data enables coordinated training that matches movement standardization with force adaptation. During system operation, the data acquisition module synchronously captures finger joint angle data and fingertip force feedback data. After calibration by the multimodal closed-loop calibration module, the data is transmitted in real-time to the linkage control unit. The linkage control unit presets a joint angle threshold corresponding to the injury-specific force feedback model and continuously compares the real-time joint angle with the preset threshold. When the joint angle does not reach the preset threshold, the control unit instructs the force feedback module to maintain a low-resistance state to avoid excessive resistance and joint strain. When the joint angle reaches the preset threshold, it automatically triggers the corresponding level of resistance output, with resistance parameters strictly matching the injury type and rehabilitation stage. When the joint angle exceeds the safe range, it immediately reduces resistance and simultaneously triggers a safety warning. This linkage control mechanism can dynamically adjust the training intensity according to the patient's real-time movement status, ensuring that the training intensity matches the current rehabilitation stage while guiding the patient to standardize movement range and gradually improve joint mobility and force control.

[0063] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

Claims

1. A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback, characterized in that, include: The hardware module includes a wearable data acquisition device, an interactive display device, and a data processing terminal. The data acquisition module synchronously acquires electromyographic signals, finger joint angle data, and fingertip force feedback data and performs preprocessing. A multimodal closed-loop calibration module is provided, which constructs a three-loop calibration model of electromyography-angle-force feedback based on the Kalman filter algorithm to correct acquisition errors in real time. The trauma-specific force feedback model library module pre-stores at least four types of hand trauma-specific force feedback curve models, and the models are associated with the trauma type and rehabilitation stage; The training execution module is based on Unity3D to construct a virtual training scene and realize real-time data feedback and security warning; The data interaction and storage module enables encrypted data transmission, hierarchical storage, and access control.

2. The hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The wearable acquisition device includes: an electromyography sensor that fits the muscles related to finger movement, an infrared + IMU fusion acquisition module deployed at each finger joint, and pressure sensors distributed on the fingertips of the thumb, index finger, and middle finger; the finger joint angle acquisition range of the infrared + IMU fusion acquisition module is 0-180°.

3. The hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The three-loop calibration model of the multimodal closed-loop calibration module includes: Electromyography-angle calibration closed loop, using electromyography characteristic parameters as a reference to correct finger joint angle data; Angle-force feedback calibration closed loop corrects the force feedback output value based on the physical correlation between finger joint angle and fingertip force feedback; Force feedback-EMG calibration closed loop, using fingertip force feedback data to verify the effectiveness of EMG signals.

4. The hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The trauma-specific force feedback curve model includes four types of models: metacarpal fracture, tendon injury, nerve injury, and hand burn. Each model includes fingertip force feedback threshold and resistance change rate parameters. Among them, the tendon injury model uses a step-like triggering resistance change, and the nerve injury model uses a gradual pressure triggering.

5. A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The virtual training scenarios of the training execution module include a virtual kitchen and an office, and the scenarios include daily action tasks such as grasping, pinching, and rotating. The safety warning is implemented in three ways: force feedback glove vibration, interactive display device interface pop-up window, and voice prompt, and different warning types correspond to different vibration modes.

6. A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The data acquisition module uses a dual-sensor fusion scheme of infrared positioning and inertial measurement unit to acquire finger joint angles. Each finger's metacarpophalangeal joint, proximal interphalangeal joint, and distal interphalangeal joint are independently equipped with acquisition units to synchronously acquire bending or extension angles, angular velocities, and angular acceleration data of each joint.

7. A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The data acquisition module acquires fingertip force feedback through distributed pressure sensors. The sensors fit the fingertip contact areas of the thumb, index finger, and middle finger, and can capture pressure peaks, pressure change rates, and pressure distribution characteristics in real time during the grasping process. The training execution module dynamically outputs the corresponding resistance based on the collected force feedback data through the micro-drive unit of the force feedback glove.

8. A hand rehabilitation training system supporting finger joint angle acquisition and fingertip force feedback according to claim 1, characterized in that, The finger joint angle acquisition data and fingertip force feedback data form a linkage control: when the finger joint angle reaches the preset threshold of the external injury-specific force feedback curve model, the fingertip force feedback automatically triggers the corresponding resistance level.

Images