Multifunctional finger rehabilitation training device

Abstract

The present application relates to the technical field of medical devices, and discloses a multifunctional finger rehabilitation training device, which comprises a frame formed by an anti-skid base and a stand column, a hole plate training module installed on the frame, a surface of the hole plate training module being provided with through holes and being internally provided with a pressure sensor array, a grip strength training module, a rotating training module, an electric control module, and a micro-control unit, an analog-digital conversion interface, a storage unit and a communication interface of the electric control module being electrically connected with the pressure sensor array, the displacement sensor and the photoelectric encoder, and the electric control module being used for collecting, calibrating and analyzing training data and storing or transmitting the training data.

Description

Technical Field

[0001] This invention relates to the field of medical device technology, and more specifically, to a multifunctional finger rehabilitation trainer. Background Technology

[0002] Currently, finger rehabilitation training devices are mainly mechanical structures, commonly including fixed perforated plates, simple hand grippers, or rotating discs. Their training effects rely on patients' self-practice, lacking precise quantitative rehabilitation assessment methods. While some existing technologies incorporate single sensors to monitor specific training indicators—for example, pressure sensors to detect finger pressure and displacement sensors to record the opening and closing strokes of hand grippers—these devices often focus only on localized data and fail to form a holistic assessment framework across training modules.

[0003] On the other hand, traditional rehabilitation training devices generally only perform threshold judgments or trend curve analysis at the data processing level, failing to comprehensively analyze the multi-dimensional characteristics of the training process. For example, they struggle to identify compensatory sliding movements of the fingers during perforated plate training, quantify the cumulative fatigue effect during grip strength training, and distinguish between stable rhythms and sudden tremors during rotational training. Due to the lack of joint assessment of key dimensions such as movement quality, fatigue risk, and coordination, existing technologies are significantly insufficient in terms of the accuracy and personalization of rehabilitation training.

[0004] Therefore, there is an urgent need for a finger rehabilitation trainer that can simultaneously collect multi-dimensional data in multiple training modules and perform joint modeling and analysis of different features through an electronic control system to output structured rehabilitation label results, in order to solve the problems of isolated data, one-sided analysis and lack of personalized assessment in existing devices. Summary of the Invention

[0005] In view of this, the present invention proposes a multifunctional finger rehabilitation trainer to solve the above problems.

[0006] The present invention proposes a multifunctional finger rehabilitation training device, comprising:

[0007] The frame includes an anti-slip base and uprights fixed to the upper surface of the anti-slip base;

[0008] An orifice plate training module is mounted on the frame. The surface of the orifice plate training module is provided with through holes, and a pressure sensor array is installed inside the orifice plate training module.

[0009] A grip strength training module is installed on the frame. The grip strength training module includes an upper pressure plate, a lower pressure plate, and a double spring structure. The bottom surface of the lower pressure plate is connected to the anti-slip base, and the top surface is connected to the double spring structure. The bottom surface of the upper pressure plate is connected to both the double spring structure and the lower pressure plate. The double spring structure includes a force-bearing plate, guide rails, and springs. The force-bearing plate and the upper pressure plate are connected by two parallel springs, and the force-bearing plate and the lower pressure plate are connected by two guide rails. A displacement sensor is fixed at the connection between the force-bearing plate and the guide rails.

[0010] A rotation training module is mounted on the frame. The rotation training module includes a textured rotating disk and a rotating shaft, and a photoelectric encoder is provided on the rotating shaft.

[0011] The electronic control module is located inside the anti-slip base and is electrically connected to the pressure sensor array, displacement sensor and photoelectric encoder respectively. The electronic control module includes a microcontroller unit, an analog-to-digital conversion interface, a storage unit and a communication interface. It is used to perform analog-to-digital conversion and calibration processing on the collected analog signals, generate a dataset for analysis, and store the analysis results or transmit them to an external terminal through the communication interface.

[0012] Furthermore, the electronic control module is configured as follows:

[0013] The spatial distribution model of the finger pressure signal output by the pressure sensor array of the orifice plate training module is performed, and the center point position of the finger pressure distribution, the dispersion of the finger pressure distribution, and the dynamic offset trajectory of the finger pressure distribution are calculated. The dynamic offset trajectory is used to identify irregular sliding or compensation actions of the finger in the orifice plate training.

[0014] The finger opening and closing stroke signal output by the displacement sensor of the grip strength training module is coupled with the force response of the grip strength training module to generate a grip strength-displacement response matrix. The grip strength elastic recovery coefficient, grip strength fatigue accumulation index, and grip strength coordination index are extracted from the grip strength-displacement response matrix. The grip strength fatigue accumulation index is obtained by the energy decay ratio of the multi-cycle response matrix and is used to evaluate the rate of endurance decline. The force response is calculated based on the displacement signal and spring parameters.

[0015] The rotation angle signal output by the photoelectric encoder of the rotation training module is periodically segmented to establish a set of rhythmic segments of rotational motion. In each set of rhythmic segments, the rotation angle consistency coefficient, the rotation rhythm deviation degree, and the rotation sudden abnormality factor are calculated. The rotation sudden abnormality factor is obtained by detecting the number of short-time angular velocity jumps within the segment and is used to identify abnormal tremors or involuntary shaking in rotational motion.

[0016] The dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the coordination index of grip force, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation are combined to form a cross-module comprehensive training feature vector. The comprehensive training feature vector is matched with the corresponding numerical range conditions or feature combinations in the preset rehabilitation label library, and the action quality level label, fatigue risk label, and coordination level label are output according to the matching results.

[0017] Furthermore, the pressure sensor array is evenly distributed in a ring along the edge of the through hole of the orifice plate training module, with each through hole corresponding to at least three pressure sensors, used to collect contact reaction forces from different directions when a finger is inserted and pressure is applied.

[0018] The output of the pressure sensor array is mapped to form a two-dimensional pressure distribution matrix;

[0019] The electronic control module calculates the position of the center point of the finger pressure distribution based on the two-dimensional pressure distribution matrix using the weighted centroid method, and calculates the dispersion of the finger pressure distribution using the second-order center distance.

[0020] The electronic control module continuously records the changing trajectory of the center point of the finger pressure distribution during the training cycle, and represents the dynamic offset trajectory of the finger pressure distribution with the cumulative offset and the instantaneous offset speed.

[0021] Furthermore, when the electronic control module couples and calculates the finger opening and closing stroke signal output by the displacement sensor of the grip strength training module with the force response of the grip strength training module, the displacement sensor collects the displacement of the upper pressure plate relative to the lower pressure plate in real time.

[0022] The electronic control module calculates the force response of the grip strength training module based on the displacement and the spring parameters of the double spring structure, and maps the displacement and force response into a grip strength-displacement response matrix.

[0023] The stiffness difference between the loading and unloading segments is extracted from the gripping force-displacement response matrix and defined as the gripping force elastic recovery coefficient.

[0024] The ratio sequence of single-cycle energy to initial cycle energy is calculated in the grip force-displacement response matrix over multiple consecutive cycles, and the cumulative grip force fatigue index is obtained by accumulating the ratios.

[0025] In the grip force-displacement response matrix of the same period, the ratio of the actual curve area to the ideal rectangular envelope area is used as the grip force coordination index.

[0026] Furthermore, when calculating the grip fatigue cumulative index, the electronic control module divides the grip force-displacement response matrix of the continuous training cycle into segments for each cycle, calculates the ratio of the load-unload loop energy of each cycle to the load-unload loop energy of the first cycle, obtains the energy decay sequence, and uses the cumulative average value of the energy decay sequence as the grip fatigue cumulative index; when the grip fatigue cumulative index is higher than the first threshold, it is determined to be a high-risk fatigue state; when the grip fatigue cumulative index is between the first threshold and the second threshold, it is determined to be a medium-risk fatigue state; when the grip fatigue cumulative index is lower than the second threshold, it is determined to be a low-risk fatigue state.

[0027] Furthermore, when establishing a set of rhythmic segments for rotational motion, the electronic control module divides the rotation angle signal output by the photoelectric encoder of the rotation training module into multiple equal-length windows according to the time sequence of the training session, and extracts a complete rotation angle cycle as a rhythmic segment within each time window to form a set of rhythmic segments for rotational motion. The electronic control module uses the average angular velocity, standard deviation of angular velocity, and duration of each rhythmic segment as segment parameters for calculating the rotation angle consistency coefficient and rotation rhythm deviation.

[0028] Furthermore, when calculating the rotational sudden anomaly factor, the electronic control module divides each segment in the set of rhythmic segments of the rotational motion into multiple short-time windows. Within each short-time window, it calculates the rate of change of the rotational angular velocity. When the instantaneous rate of change of the rotational angular velocity exceeds the anomaly threshold set based on the mean and standard deviation of the angular velocity of that segment, an anomaly event is recorded. The electronic control module weights and sums the number and amplitude of all anomaly events to obtain the rotational sudden anomaly factor.

[0029] Furthermore, when forming a cross-module comprehensive training feature vector, the electronic control module treats the dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation as independent components to construct an eight-dimensional comprehensive training feature vector. The eight-dimensional comprehensive training feature vector is then normalized to ensure that each component is within a uniform numerical range, thereby guaranteeing the comparability of cross-module features.

[0030] Furthermore, when the electronic control module matches the comprehensive training feature vector with the rehabilitation label library, it compares each component of the comprehensive training feature vector with the numerical range conditions in the rehabilitation label library one by one. During the matching process, a confidence level calculation method is used. When a component of the comprehensive training feature vector is located in the central region of the numerical range conditions, a high confidence level is assigned. When a component is located in the edge region of the numerical range conditions, a low confidence level is assigned. Finally, based on the weighted result of the confidence level of each label, the movement quality level label, fatigue risk label, and coordination level label are output.

[0031] Furthermore, when outputting the label results, the electronic control module arranges the action quality level label, fatigue risk label, and coordination level label in hierarchical order, and adds a confidence level to each label. When the confidence level of at least one label is lower than the lower confidence level limit, a confirmation prompt label is added. When the confidence level of all labels is higher than the upper confidence level limit, a stability label is added. The combination of labels with stability labels is transmitted to the external terminal through the communication interface in the form of rehabilitation label code.

[0032] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0033] The perforated plate training module, grip strength training module, and rotation training module are integrated into a single frame, eliminating the need for patients to switch between different training devices and improving the continuity and convenience of rehabilitation training. A pressure sensor array is arranged in a ring within the perforated plate training module, a displacement sensor is installed at the connection between the force plate and the guide rail in the grip strength training module, and a photoelectric encoder is mounted on the rotation axis of the rotation training module, forming a multi-dimensional sensing network covering pressure, displacement, and rotation angle, enabling comprehensive acquisition of movement characteristics. The electronic control module converts the raw signals from different training modules into quantifiable features, including the dynamic offset trajectory of finger pressure distribution, grip fatigue accumulation index, and sudden abnormal rotation factors, and further combines them into a comprehensive training feature vector. This overcomes the limitations of single-index analysis, achieving a unified representation of movement quality, fatigue risk, and coordination. The electronic control module compares the comprehensive training feature vector with a pre-set rehabilitation label library, automatically outputting movement quality level labels, fatigue risk labels, and coordination level labels based on the matching results, and can add confidence level prompts to form a structured rehabilitation assessment result. Compared to traditional curve analysis, this method can transform complex, multi-dimensional training data into intuitive rehabilitation labels, making it easier for patients and doctors to understand and use. Through combined analysis of cross-module features and label results, the electronic control module can identify abnormal movements and fatigue trends during training and generate personalized rehabilitation training suggestions for patients. Attached Figure Description

[0034] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:

[0035] Figure 1 This is a three-dimensional structural diagram of a multifunctional finger rehabilitation training device provided in an embodiment of the present invention.

[0036] In the diagram, 10 is the anti-slip base; 11 is the column; 20 is the perforated plate training module; 21 is the through hole; 30 is the grip strength training module; 31 is the upper pressure plate; 32 is the lower pressure plate; 33 is the force plate; 34 is the guide rail; 35 is the spring; 36 is the displacement sensor; 40 is the rotating disk; and 41 is the rotating shaft. Detailed Implementation

[0037] Exemplary embodiments of the present application will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0038] See Figure 1 As shown, an embodiment of the present invention provides a multifunctional finger rehabilitation training device, comprising:

[0039] The frame includes an anti-slip base 10 and a column 11 fixed to the upper surface of the anti-slip base 10.

[0040] The orifice plate training module 20 is mounted on the frame. Through holes 21 are opened on the surface of the orifice plate training module 20. A pressure sensor array 22 is installed inside the orifice plate training module 20.

[0041] Specifically, the cross-section of the through hole 21 can be a regular shape or an irregular image. In this embodiment, there are four through holes 21, and the cross-sectional shapes are triangular, square and circular, respectively.

[0042] The grip strength training module 30 is mounted on the frame. The grip strength training module 30 includes an upper pressure plate 31, a lower pressure plate 32, and a double spring 35 structure. The bottom surface of the lower pressure plate 32 is connected to the anti-slip base 10, and the top surface is connected to the double spring 35 structure. The bottom surface of the upper pressure plate 31 is connected to both the double spring 35 structure and the lower pressure plate 32. The double spring 35 structure includes a force plate 33, a guide rail 34, and springs 35. The force plate 33 and the upper pressure plate 31 are connected by two parallel springs 35, and the force plate 33 and the lower pressure plate 32 are connected by two guide rails 34. A displacement sensor 36 is fixed at the connection between the force plate 33 and the guide rail 34.

[0043] Specifically, the guide rail 34 consists of two support rods, one of which is fitted onto the other. Both support rods have equally spaced small holes on their side walls. When the holes on the two support rods overlap, they are fixed by inserting screws into the holes, thus fixing the length of the guide rail 34. The grip strength training module 30 adjusts the required grip strength by changing the length of the guide rail 34.

[0044] A rotation training module is mounted on the frame. The rotation training module includes a textured rotating disk 40 and a rotating shaft 41, on which a photoelectric encoder is mounted.

[0045] The electronic control module is located inside the anti-slip base 10 and is electrically connected to the pressure sensor array 22, the displacement sensor 36 and the photoelectric encoder respectively. The electronic control module includes a microcontroller unit, an analog-to-digital conversion interface, a storage unit and a communication interface. It is used to perform analog-to-digital conversion and calibration processing on the collected analog signals, generate a dataset for analysis, and store the analysis results or transmit them to an external terminal through the communication interface.

[0046] As can be understood from the above, the frame consists of an anti-slip base 10 and a column 11. The anti-slip base 10 is located at the bottom and uses a rubber pad or a textured surface to enhance stability, ensuring that the device does not slip when used on a table or floor. The column 11 is fixed to the upper surface of the anti-slip base 10 and provides mounting support for the perforated plate training module 20, grip strength training module 30, and rotation training module. The main body of the frame can be made of lightweight alloy or high-strength engineering plastic to balance portability and structural strength.

[0047] The perforated plate training module 20 is mounted on the frame, and its surface has multiple through holes 21. The shape and size of the through holes 21 can be designed as regular or irregular geometric shapes according to the needs of rehabilitation training. In this embodiment, there are four through holes 21, and their cross-sections are designed as triangles, squares, and circles, respectively, to simulate fingertip adaptation training under different daily gripping environments. A pressure sensor array 22 is embedded inside the through holes 21. The pressure sensors are evenly distributed in a ring along the edge of the through hole 21, and each through hole 21 is equipped with at least three sensors to collect multi-directional pressure signals generated by the fingers during insertion or withdrawal training. The electronic control module performs spatial mapping on these signals, which can reconstruct the force distribution and dynamic changes of the fingers during training, and thus identify irregular sliding or compensatory movements.

[0048] The guide rail 34 consists of two support rods, one of which can be fitted onto the outside of the other. Both support rods have evenly spaced small holes on their sidewalls. When the holes of the two support rods coincide, the user can fix them by inserting screws, thus adjusting the length of the guide rail 34. Adjusting the length of the guide rail 34 directly changes the initial preload of the spring 35, thereby adjusting the resistance level of grip strength training to meet the training needs of patients at different rehabilitation stages. A displacement sensor 36 is fixed at the connection between the force plate 33 and the guide rail 34 to collect the displacement of the upper pressure plate 31 relative to the lower pressure plate 32 in real time. The electronic control module calculates the force response based on the displacement and spring 35 parameters, constructs a grip strength-displacement response matrix, and extracts the grip strength elastic recovery coefficient, grip strength fatigue accumulation index, and grip strength coordination index from it to evaluate muscle strength and endurance.

[0049] The rotation training module is mounted on a frame and includes a textured rotating disk 40 and a rotating shaft 41. An optical encoder is installed inside or at the end of the rotating shaft 41. The surface of the rotating disk 40 is designed with an ergonomic texture to increase friction and tactile feedback, helping patients better control their rotational movements during training. The optical encoder can acquire real-time changes in the angle and angular velocity of the rotating disk 40. The electronic control module periodically segments the angle signal to form a set of rhythmic segments of the rotational movement and calculates the rotation angle consistency coefficient, rotation rhythm deviation, and rotational abrupt abnormality factor, thereby identifying whether the patient's movements have unstable rhythms or abnormal tremors.

[0050] This invention also provides the configuration details of the electronic control module in a multifunctional finger rehabilitation trainer, including:

[0051] The spatial distribution model of the finger pressure signal output by the pressure sensor array of the perforated plate training module is performed. The center point position of the finger pressure distribution, the dispersion of the finger pressure distribution, and the dynamic offset trajectory of the finger pressure distribution are calculated. The dynamic offset trajectory is used to identify irregular sliding or compensation actions of the finger in the perforated plate training.

[0052] The finger opening and closing stroke signal output by the displacement sensor of the grip strength training module is coupled with the force response of the grip strength training module to generate a grip strength-displacement response matrix. The grip strength elastic recovery coefficient, grip strength fatigue accumulation index, and grip strength coordination index are extracted from the grip strength-displacement response matrix. The grip strength fatigue accumulation index is obtained by the energy decay ratio of the multi-cycle response matrix and is used to evaluate the rate of endurance decline. The force response is calculated based on the displacement signal and spring parameters.

[0053] The rotation angle signal output by the photoelectric encoder of the rotation training module is periodically segmented to establish a set of rhythmic segments of rotational motion. In each set of rhythmic segments, the rotation angle consistency coefficient, rotation rhythm deviation degree and rotation sudden abnormal factor are calculated. The rotation sudden abnormal factor is obtained by detecting the number of short-time angular velocity jumps within the segment and is used to identify abnormal tremors or involuntary shaking in rotational motion.

[0054] The dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation are combined to form a cross-module comprehensive training feature vector. The comprehensive training feature vector is matched with the corresponding numerical range conditions or feature combinations in the pre-set rehabilitation label library, and the action quality level label, fatigue risk label, and coordination level label are output according to the matching results.

[0055] It should be noted that, in this embodiment, the electronic control module, in addition to including a microcontroller unit, analog-to-digital converter interface, storage unit, and communication interface, also includes a data preprocessing unit, a feature filtering unit, a label comparison unit, and an anomaly feedback unit. The working process of each unit is as follows: The rehabilitation label library is jointly established using healthy control group data and patient rehabilitation data. The feature vector of the healthy control group is derived from perforated plate training, grip strength training, and rotation training movements completed by different age and gender groups under standard training conditions, forming a basic benchmark range. Patient rehabilitation data is continuously collected in the early, middle, and late stages of rehabilitation, and labeled by doctors with movement quality level, fatigue risk level, and coordination level, forming a labeled sample set. The electronic control module periodically clusters new samples with existing data and adjusts the numerical range, dynamically updating the label library so that the label library can reflect the rehabilitation differences of different individuals and different stages, avoiding the bias caused by fixed thresholds.

[0056] After the analog signals output from the pressure sensor array, displacement sensor, and photoelectric encoder are converted from analog to digital, the electronic control module performs the following steps: Filtering and denoising: A low-pass filter is used to eliminate high-frequency electromagnetic interference, and a median filter is used to remove spike noise caused by mechanical vibration, ensuring signal stability. Normalization: The outputs of different sensors are mapped to a unified numerical range, such as [0,1], to avoid deviations caused by differences in physical dimensions between feature dimensions. Time alignment: When the sampling frequencies of different sensors are inconsistent, the electronic control module interpolates and synchronizes the signals based on timestamps to ensure the comparability of features across modules. Segmentation and labeling: The training process is divided into several standardized time windows, each with a training action number and patient ID, ensuring data traceability.

[0057] After extracting the dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation, the electronic control module performs the following processing on the features: Correlation analysis: Highly correlated features are determined by Pearson correlation coefficient or mutual information, and redundant components are removed. Stability test: The fluctuation range of each feature within continuous training cycles is detected. When the fluctuation exceeds the tolerance range, it is automatically marked as a low-confidence feature to avoid its interference with the final result. Unified vector construction: Finally, an eight-dimensional comprehensive training feature vector is formed, and each component is standardized to ensure consistency in numerical distribution.

[0058] The electronic control module performs a component-by-component comparison between the comprehensive training feature vector and the rehabilitation label library. The comparison process includes: numerical range retrieval: determining the numerical range each component falls into and its corresponding rehabilitation label range; confidence assignment: assigning high confidence to components falling within the center of the range and low confidence to components near the range boundary; and weighted fusion: combining the confidence of each feature dimension with the label weights to ultimately output the movement quality level label, fatigue risk label, and coordination level label.

[0059] The electronic control module appends a timestamp to the output results and generates a rehabilitation training log. The rehabilitation training log includes: training date, training duration, comprehensive training feature vector, output label, and corresponding confidence level. Log data can be stored permanently in a storage unit or transmitted to a doctor's terminal via a communication interface, facilitating remote diagnosis and tracking of training effectiveness.

[0060] When the electronic control module identifies the following situations during data analysis, it will trigger the abnormal feedback mechanism: Orifice plate training module: When the cumulative offset of the dynamic offset trajectory of the finger pressure distribution exceeds the normal range, it will output a "sliding compensation abnormality prompt"; Grip strength training module: When the cumulative grip strength fatigue index exceeds the set threshold, it will output a "fatigue overload prompt"; Rotation training module: When a short-term abnormal jump in angular velocity is detected by a sudden abnormal rotational factor, it will output a "tremor abnormality prompt". These abnormal prompts are not only stored in the rehabilitation training log but can also be transmitted to an external terminal via the communication interface to trigger audible and visual alarms or automatically pause training to prevent secondary injury caused by continued training while in an adverse movement state.

[0061] In some embodiments of this application, the pressure sensor array is uniformly distributed in a ring along the edge of the through hole of the orifice plate training module, and each through hole corresponds to at least three pressure sensors for collecting contact reaction forces from different directions when a finger is inserted and pressure is applied.

[0062] The output of the pressure sensor array is spatially mapped to form a two-dimensional pressure distribution matrix.

[0063] The electronic control module is based on a two-dimensional pressure distribution matrix. It uses the weighted centroid method to calculate the position of the center point of the finger pressure distribution and the second-order center distance to calculate the dispersion of the finger pressure distribution.

[0064] The electronic control module continuously records the changing trajectory of the center point of the finger pressure distribution during the training cycle, and represents the dynamic offset trajectory of the finger pressure distribution with the cumulative offset and the instantaneous offset speed.

[0065] It should be noted that the perforated plate training module includes multiple through holes on its surface. The cross-section of these through holes can be triangular, square, circular, or other irregular shapes to accommodate the different movement needs of patients during rehabilitation training. To ensure accurate acquisition of the force applied to the finger during insertion or withdrawal training, a sensor mounting slot is pre-drilled along the circumferential direction on the inner wall edge of each through hole. The pressure sensor array is fixed within the mounting slot, with the probe end face slightly protruding from the inner wall surface of the through hole to ensure direct contact between the finger and the sensor when inserted or subjected to force.

[0066] The pressure sensor array is arranged in a ring-shaped distribution. Preferably, four to six pressure sensors are distributed circumferentially in each through-hole, with equal intervals, thus forming a closed ring-shaped sensing zone. Each through-hole is equipped with at least three pressure sensors to ensure that contact reaction force signals applied by the finger can be collected from different directions. The sensor mounting depth and preload are calibrated to maintain zero-point stability when not in use and to output an electrical signal proportional to the contact pressure when force is applied.

[0067] After the output signal of the pressure sensor array enters the electronic control module, it is first converted into a digital signal through an analog-to-digital converter. The electronic control module then establishes a two-dimensional pressure distribution matrix based on the physical position coordinates of the sensors and the corresponding pressure values. Specifically, assuming that the circumferential coordinates of a certain through-hole are discretized into N sensor nodes, the rows of the matrix represent angular positions, and the columns represent the measured pressure values, thus forming a pressure spatial distribution map.

[0068] The electronic control module performs the following calculations based on the two-dimensional pressure distribution matrix: Center point position calculation: The weighted centroid method is adopted, that is, the coordinates of each sensor are used as independent variables and the output pressure value is used as weights to calculate the coordinate position of the resultant force point, which is used to characterize the overall force direction of the finger in the perforated plate training.

[0069] Dispersion calculation: The matrix is ​​operated on using the second-order center distance to obtain the distribution variance of each pressure value relative to the center point, thus obtaining the dispersion of finger pressure distribution, which reflects the uniformity of force. A small dispersion value indicates uniform force and stable training movements; a large dispersion value indicates concentrated or skewed force, indicating biased force application or compensatory movements.

[0070] Dynamic offset trajectory calculation: During the training cycle, the electronic control module continuously collects the center point position coordinates at multiple moments, forming a trajectory curve that changes over time. The electronic control module further calculates the cumulative offset, which is the overall offset distance of the center point position relative to its initial position during the training cycle; simultaneously, it calculates the instantaneous offset velocity, which is the rate of change of the center point position between adjacent moments. The cumulative offset characterizes the overall stability of the finger throughout the entire training cycle, while the instantaneous offset velocity reveals whether there is sudden slippage or rapid adjustment in a single movement.

[0071] In the training data analysis phase, the electronic control module interprets the above three types of features together: when the dispersion value is in the normal range and the cumulative offset is small and the instantaneous offset speed is stable, it is determined that the patient's finger movements are stable and the force is uniform; when the dispersion is too large, the cumulative offset increases significantly or the instantaneous offset speed has an abnormal peak, it is determined that the patient has compensatory movements, irregular sliding or insufficient muscle coordination, and the "unstable movement" prompt is recorded in the output label.

[0072] Furthermore, the electronic control module can perform time-series comparisons of the dispersion, cumulative offset, and instantaneous offset velocity obtained by the same patient in different training cycles to observe the changing trends of force uniformity and movement stability during the patient's rehabilitation process, thereby providing a basis for rehabilitation progress assessment and the formulation of personalized training programs.

[0073] Through the above implementation method, the pressure sensor array can not only acquire the magnitude of the pressure in a single direction, but also form a complete spatial distribution model. Combined with the calculation of the center point position, dispersion, and dynamic offset trajectory, it can comprehensively reflect the patient's movement stability and fine control ability during perforated plate training. This design significantly improves the accuracy of rehabilitation training devices in movement quality recognition and abnormality detection.

[0074] In some embodiments of this application, when the electronic control module performs coupled calculations on the finger opening and closing stroke signal output by the displacement sensor of the grip strength training module and the force response of the grip strength training module, the displacement sensor collects the displacement of the upper pressure plate relative to the lower pressure plate in real time.

[0075] The electronic control module calculates the force response of the grip strength training module based on the displacement and the spring parameters of the double spring structure, and maps the displacement and force response into a grip strength-displacement response matrix.

[0076] The stiffness difference between the loading and unloading segments is extracted from the gripping force-displacement response matrix and defined as the gripping force elastic recovery coefficient.

[0077] The ratio sequence of single-cycle energy to initial cycle energy is calculated in the grip force-displacement response matrix over multiple consecutive cycles, and the cumulative grip force fatigue index is obtained by accumulating the results.

[0078] In the grip force-displacement response matrix of the same period, the ratio of the actual curve area to the ideal rectangular envelope area is used as the grip force coordination index.

[0079] It should be noted that when processing the data output by the grip strength training module, the electronic control module first acquires the real-time signal from the displacement sensor. The displacement sensor, fixed at the connection between the force plate and the guide rail, can accurately measure the displacement of the upper pressure plate relative to the lower pressure plate. Since the force characteristics of the double-spring structure are known, and its spring constant can be obtained through calibration during assembly, the electronic control module can calculate the corresponding force response value based on the real-time displacement and spring parameters.

[0080] During data modeling, the electronic control module uses displacement as the horizontal axis and force response as the vertical axis, mapping their relationship into a grip force-displacement response matrix. This matrix reflects the biomechanical characteristics of the patient during a complete grip and relaxation process.

[0081] To quantify the functional state of a patient's fingers and hand, the electronic control module extracts the following indicators from the grip force-displacement response matrix:

[0082] Grip strength elastic recovery coefficient: In the matrix, the loading segment represents the process of the patient applying force to grip, and the unloading segment represents the process of the patient relaxing their fingers. The electronic control module calculates the stiffness difference between the loading and unloading segments, i.e., the difference in the slope of the two curves, and defines this difference as the grip strength elastic recovery coefficient. This coefficient reflects the ability of the muscle to recover to its initial state after completing one contraction; the closer the value is to zero, the better the elastic recovery performance of the patient's fingers.

[0083] Grip strength fatigue accumulation index: During continuous multi-cycle training, the electronic control module calculates the energy of the grip strength-displacement response matrix for each cycle, with the energy equivalent to the area envelope of the curve. The energy decay sequence is obtained by calculating the ratio of the energy of a single cycle to the energy of the first cycle. The electronic control module then accumulates and averages this sequence to form the grip strength fatigue accumulation index. This index reflects the rate of endurance decline during prolonged repetitive training; a rapid decrease in the index indicates significant muscle fatigue.

[0084] Grip strength coordination index: Within the same training cycle, the electronic control module calculates the ratio of the actual area of ​​the grip strength-displacement response curve to the area of ​​the ideal rectangular envelope formed by the maximum displacement and maximum force, which is defined as the grip strength coordination index. This index is used to evaluate the coordination of force and displacement of the patient throughout the gripping process. The higher the consistency, the better the neuromuscular control ability; if the ratio is too low, it indicates that there is lag or imbalance in the patient's force exertion process.

[0085] By calculating the above indicators, the electronic control module can transform the grip strength training process into quantifiable parameter results and compare them with a rehabilitation label database. Compared with traditional trainers that simply record the maximum grip strength, this embodiment can provide multi-dimensional analysis of elastic recovery ability, endurance change trends, and coordination levels, thereby providing doctors with a more comprehensive basis for rehabilitation assessment.

[0086] In some embodiments of this application, when calculating the grip fatigue cumulative index, the electronic control module divides the grip force-displacement response matrix of the continuous training cycle into segments, calculates the ratio of the load-unload loop energy of each cycle to the load-unload loop energy of the first cycle, obtains the energy decay sequence, and uses the cumulative average value of the energy decay sequence as the grip fatigue cumulative index; when the grip fatigue cumulative index is higher than the first threshold, it is determined to be a high-risk fatigue state; when the grip fatigue cumulative index is between the first threshold and the second threshold, it is determined to be a medium-risk fatigue state; and when the grip fatigue cumulative index is lower than the second threshold, it is determined to be a low-risk fatigue state.

[0087] It should be noted that when calculating the cumulative grip strength fatigue index, the electronic control module first divides the continuous training data output by the grip strength training module into periods. The electronic control module uses the displacement of the upper pressure plate relative to the lower pressure plate collected by the displacement sensor as a reference. When the displacement gradually increases from zero to the maximum displacement and then returns to zero, it is defined as a complete training cycle.

[0088] The electronic control module divides the grip force-displacement response curve of each training cycle into a loading segment (force application phase) and an unloading segment (relaxation phase), and calculates the area of ​​the loading-unloading curve for each segment. The area of ​​the curve is equivalent to the energy consumed by the patient to complete the movement in that cycle.

[0089] In the initial training cycle, the electronic control module records the load-unload loop energy of the first cycle as an energy baseline. Subsequently, in subsequent cycles, the electronic control module calculates the ratio of the current cycle energy to the first cycle energy to obtain the energy decay value. As the number of training cycles increases, the electronic control module gradually forms an energy decay sequence.

[0090] To reflect the overall fatigue trend, the electronic control module performs cumulative averaging of the energy decay sequence to obtain the grip strength fatigue cumulative index. This index can characterize the rate of decline in the patient's energy expenditure capacity during multi-cycle training, i.e., the decline in muscle endurance.

[0091] In terms of judgment logic, the electronic control module presets a first threshold and a second threshold: when the cumulative grip strength fatigue index is higher than the first threshold, it indicates that the energy decay is significant and the patient's muscles rapidly lose endurance in a short period of time, which is judged as a high-risk fatigue state; when the cumulative grip strength fatigue index is between the first threshold and the second threshold, it indicates that the patient's endurance declines relatively slowly, which is judged as a medium-risk fatigue state; when the cumulative grip strength fatigue index is lower than the second threshold, it indicates that the energy decay is small and the patient maintains good endurance in multi-cycle training, which is judged as a low-risk fatigue state.

[0092] Based on the above, the grip strength fatigue accumulation index can not only quantify the degree of fatigue in a single training session, but also reflect the trend of declining endurance during long-term training. Compared with traditional single force value monitoring methods, this embodiment provides a dynamic evaluation index based on energy consumption characteristics, which can more accurately distinguish the fatigue state of different patients and provide a basis for adjusting the intensity of rehabilitation training.

[0093] For the first and second thresholds, after calculating the cumulative grip strength fatigue index, the electronic control module needs to determine whether the patient is in different levels of fatigue risk based on the magnitude of this index. To ensure the scientific validity of the determination, the first and second thresholds need to be obtained through a combination of clinical statistics and individual correction.

[0094] First, during the initial use phase, the electronic control module fully records the patient's energy consumption value in the first training cycle and defines this value as the baseline. Subsequently, in each subsequent training cycle, the module calculates a new energy consumption value and compares it with the baseline to obtain the percentage decrease in energy consumption over the cycle, thus forming an energy decrease sequence. The module continuously tracks this sequence to analyze the overall decline trend of the patient's energy consumption across multiple training cycles.

[0095] During threshold setting, the electronic control module first constructs a population control reference based on clinical statistical data. Healthy individuals typically exhibit slow and relatively small energy declines during multi-cycle grip strength training; the module uses the average rate of decline and fluctuation range of this group as a low-risk reference interval. Rehabilitation patients, especially those in the early stages of rehabilitation, often experience rapid energy declines within a short period; the module uses the average rate of decline and fluctuation range of this group as a high-risk reference interval. By comparing the distribution of the two groups, the module obtains an upper limit for determining high-risk fatigue and a lower limit for determining low-risk fatigue, which serve as the first and second thresholds, respectively.

[0096] To avoid misjudgments due to differences in baseline energy levels among patients, the electronic control module performs individualized corrections when determining thresholds. Specifically, it compares the patient's baseline energy value with the average baseline energy level in a population control database. If the patient's baseline energy level is significantly lower than the population average, it indicates insufficient overall strength reserves. In this case, the electronic control module lowers the first and second thresholds to make the judgment more consistent with the patient's actual situation. Conversely, if the patient's baseline energy level is significantly higher than the population average, it indicates good strength reserves. The electronic control module raises the first and second thresholds to avoid misjudging a normal range of energy decreases as abnormal fatigue.

[0097] During the threshold application process, the electronic control module compares the patient's current grip fatigue accumulation index with the corrected threshold range: when the grip fatigue accumulation index exceeds the first threshold, the electronic control module outputs a high-risk fatigue state, indicating that the patient may experience significant endurance loss in a short period of time; when the grip fatigue accumulation index is between the first and second thresholds, the electronic control module outputs a medium-risk fatigue state, indicating that the patient has some fatigue accumulation, but it is still within a controllable range; when the grip fatigue accumulation index is below the second threshold, the electronic control module outputs a low-risk fatigue state, indicating that the patient's endurance level remains stable.

[0098] By combining the above methods with dynamic correction based on individual patient differences, the misjudgment problems that may be caused by using fixed thresholds in traditional methods are avoided. This can more accurately reflect the patient's actual fatigue level and provide a scientific basis for adjusting the intensity of rehabilitation training.

[0099] In some embodiments of this application, when establishing a set of rhythmic segments of rotational motion, the electronic control module divides the rotation angle signal output by the photoelectric encoder of the rotation training module into multiple equal-length windows according to the time sequence of the training session, and extracts the complete rotation angle cycle as a rhythmic segment within each time window to form a set of rhythmic segments of rotational motion; the electronic control module uses the average angular velocity, standard deviation of angular velocity and duration of each rhythmic segment as segment parameters for calculating the rotation angle consistency coefficient and rotation rhythm deviation.

[0100] A training session refers to a continuous training process completed by a patient using a multifunctional finger rehabilitation trainer. This process is typically set by a doctor or rehabilitation system with start and end times. Within this timeframe, the patient repeatedly performs the same or similar rehabilitation exercises, such as repeatedly rotating a disc for several minutes.

[0101] It should be noted that when processing the photoelectric encoder signal output by the rotation training module, the electronic control module first continuously acquires the rotation angle signal according to the time sequence of the training session. The photoelectric encoder can output high-resolution angular displacement signals, and the electronic control module sorts the signals according to the timestamps to obtain the complete original curve of the rotation angle changing over time.

[0102] To facilitate segmentation and analysis of the rotational motion, the electronic control module divides the entire training session into multiple equal-length windows. The length of each window can be set according to training requirements, typically covering multiple complete rotation cycles to ensure the stability of the calculation results. Within each window, the electronic control module uses peak detection and cycle labeling methods to extract a series of complete rotation angle cycles, defining this set of cycles as a rhythmic segment. As the training session progresses, the rhythmic segments from all windows converge to form a set of rhythmic segments representing the rotational motion.

[0103] In the feature extraction stage, the electronic control module calculates the following parameters for each rhythmic segment: Average angular velocity: obtained by dividing the total angular displacement by the segment duration, reflecting the overall speed of rotation in that segment. Angular velocity standard deviation: obtained by statistically analyzing the fluctuation of angular velocity within the segment over time, reflecting the uniformity and stability of the rotational movement. Segment duration: recording the time from the start to the end of the segment, reflecting the completeness and regularity of the movement rhythm. Based on these parameters, the electronic control module can further calculate the rotation angle consistency coefficient and rotation rhythm deviation. Specifically, the rotation angle consistency coefficient measures the closeness of the average angular velocities between rhythmic segments. When the difference in average angular velocities between different segments is small, it indicates that the movement rhythm is well maintained; conversely, it indicates insufficient uniformity of the patient's rotational movement. The rotation rhythm deviation measures the fluctuation of the duration of each segment. When the durations are basically consistent, the deviation is small, indicating that the rotational movement has a stable rhythm; when the durations differ significantly, the deviation is large, suggesting that the patient has rhythmic instability or insufficient control during the execution of the movement.

[0104] In this way, the electronic control module can not only identify the patient's immediate performance in a single rotational movement, but also reflect the changing trends in the patient's movement stability and rhythm control ability throughout the entire training session through a holistic analysis of rhythmic segments. This provides data support for the comprehensive evaluation of finger dexterity, neural control ability, and coordination in rehabilitation training.

[0105] In some embodiments of this application, when calculating the rotational sudden anomaly factor, the electronic control module divides each segment in the set of rhythmic segments of the rotational motion into multiple short-time windows, calculates the rate of change of the rotational angular velocity within each short-time window, and records an anomaly event when the instantaneous rate of change of the rotational angular velocity exceeds an anomaly threshold set based on the mean and standard deviation of the angular velocity of that segment. The electronic control module weights and sums the number and amplitude of all anomalies as the rotational sudden anomaly factor.

[0106] It should be noted that the multifunctional finger rehabilitation trainer includes a photoelectric encoder mounted on the rotating shaft of the rotation training module. The photoelectric encoder outputs a continuous signal showing the rotation angle changing over time during rotation training; the electronic control module receives this signal and processes it accordingly.

[0107] In a complete rotational motion cycle, the electronic control module first divides the angle signal into several short-time windows. The short-time windows are set based on the average cycle length of the rotational training motion. For example, in clinical statistics, the average cycle for a patient to complete one full rotation is usually around one second. Therefore, the length of the short-time window is selected between tens of milliseconds and hundreds of milliseconds to ensure that a complete cycle contains several short-time windows, thereby effectively capturing local dynamic changes.

[0108] Within each short-term window, the electronic control module calculates the trend of rotational angular velocity changes over time. To determine whether an abnormal event has occurred, the module determines an adaptive threshold based on the average angular velocity and fluctuation range of the rhythmic segment. Specifically, at the beginning of each rhythmic segment, the module first calculates the average level and fluctuation amplitude of the angular velocity within that segment, and then sets a threshold range based on the statistical results: if the angular velocity change within the short-term window exceeds this range, it is considered an abnormal event. This approach is based on the fact that the range and speed of movement vary significantly between different patients or even the same patient at different training stages. Using a uniform fixed value could easily lead to misjudgment, while a dynamic threshold based on segment data better reflects the patient's current state.

[0109] After analyzing a single rhythmic segment, the electronic control module counts all abnormal events and assigns weights based on the amplitude of each event. The amplitude weights are based on the clinical practice that the severity of abnormal tremors is usually related to the amplitude of angular velocity mutations; therefore, the larger the amplitude, the higher its contribution to the overall result. Finally, the electronic control module sums the weighted results of these abnormal events to obtain the rotational sudden abnormality factor corresponding to that rhythmic segment.

[0110] Throughout the training process, the electronic control module sequentially processes multiple rhythmic segments, generating a series of rotational aberration abnormalities. Through continuous analysis of these factors, the module can determine changes in the patient's rotational stability over different time periods. If the abnormal factors show elevated levels in multiple consecutive segments, it indicates that the patient may have persistent tremor, shaking, or sudden loss of control. The module stores the curves of these abnormal factors in a storage unit and can transmit them to an external terminal via a communication interface for further analysis by the physician.

[0111] In some embodiments of this application, when the electronic control module forms a cross-module comprehensive training feature vector, it treats the dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation as independent components to construct an eight-dimensional comprehensive training feature vector. The eight-dimensional comprehensive training feature vector is then normalized to ensure that each component is in a uniform numerical range, thereby guaranteeing the comparability of cross-module features.

[0112] It should be noted that, firstly, the electronic control module treats the following eight indicators as independent components: **Finger Pressure Distribution Dynamic Offset Trajectory:** Obtained by continuous time-series tracking of the two-dimensional pressure distribution matrix acquired by the pressure sensor array, used to reflect whether there is offset, sliding, or compensating movement of the fingers during perforated plate training. **Finger Pressure Distribution Dispersion:** Calculated from the pressure distribution matrix using the second-order center distance, used to measure the concentration or dispersion of finger force application. High concentration indicates stable finger force application, while high dispersion suggests unstable control. **Grip Force Elastic Recovery Coefficient:** Obtained from the stiffness difference between the loading and unloading segments in the grip force-displacement response matrix, used to reflect the muscle's recovery ability after force application. **Grip Force Fatigue Cumulative Index:** Obtained from the cumulative average of the continuous multi-cycle grip force-displacement response energy ratio sequence, used to evaluate the rate of endurance decline over time during training. **Grip Force Coordination Index:** Obtained from the ratio of the area of ​​the single-cycle response curve to the area of ​​the ideal rectangular envelope, used to reflect the level of coordinated control of the fingers during force application and release. Rotational Angle Consistency Coefficient: Calculated from the similarity of average angular velocities between rotational segments, reflecting the uniformity and repeatability of rotational movements. Rotational Rhythm Deviation: Calculated from the differences in duration between rhythmic segments, measuring the stability of the movement rhythm. Rotational Sudden Abnormality Factor: Obtained by weighting the number and amplitude of abrupt angular velocity changes within a short time window, used to identify tremors or involuntary shaking in rotational movements.

[0113] The eight components mentioned above differ significantly in their numerical form, dimensions, and range of values. For example, the pressure distribution offset trajectory is expressed in millimeters or displacement, the grip force-related components are expressed as energy ratios or stiffness differences, and the rotational-related components are expressed as angular velocity fluctuations or time deviations. Directly combining these components may lead to a loss of comparability between different components due to inconsistencies in numerical scales.

[0114] Therefore, the electronic control module normalizes the eight components before constructing the comprehensive feature vector. The normalization process is based on statistical data from the rehabilitation database. Specifically, the electronic control module first calls historical training data from different population groups in the rehabilitation database to determine the common value range of each component in both healthy and rehabilitated groups. The upper and lower limits of this common value range are then used as mapping boundaries, and the current patient's component values ​​are linearly mapped to a unified standard interval (e.g., the interval between 0 and 1).

[0115] When a component exceeds the common range, the electronic control module marks it as an out-of-bounds value and retains its out-of-limit information after mapping, so that doctors can identify abnormalities during result analysis. Through this normalization method, the electronic control module can ensure that all eight components are within the same numerical range, avoiding excessive amplification or reduction of a component in the comprehensive calculation due to differences in dimensions.

[0116] Finally, the electronic control module combines the eight normalized components in a fixed order to form an eight-dimensional comprehensive training feature vector. This feature vector includes finger force control, muscle recovery ability, endurance changes, and motor coordination, as well as the stability and abnormal fluctuations of rotational movements, thus comprehensively depicting the patient's overall performance in multi-module training. The electronic control module can store this comprehensive training feature vector as a standardized data structure and compare it with a rehabilitation label library to output movement quality level labels, fatigue risk labels, and coordination level labels.

[0117] In some embodiments of this application, when the electronic control module matches the comprehensive training feature vector with the rehabilitation label library, it compares each component of the comprehensive training feature vector with the numerical range conditions in the rehabilitation label library one by one. During the matching process, a confidence level calculation method is used. When a component of the comprehensive training feature vector is located in the central region of the numerical range conditions, a high confidence level is assigned. When a component is located in the edge region of the numerical range conditions, a low confidence level is assigned. Finally, the movement quality level label, fatigue risk label, and coordination level label are output based on the weighted result of the confidence level of each label.

[0118] It should be noted that the numerical ranges in the rehabilitation label library are determined based on a combination of clinical statistical data and individualized calibration.

[0119] First, the initial establishment of the rehabilitation label database was based on clinical controlled trials. The study subjects included healthy individuals and patients at different stages of rehabilitation, collecting multi-dimensional parameters from the perforated plate training module, grip strength training module, and rotation training module. For example, large-sample measurements were conducted on the dynamic offset trajectory of finger pressure distribution, grip strength elastic recovery coefficient, and rotation angle consistency coefficient in healthy adults, and their means, standard deviations, and percentile distributions were statistically analyzed to form a reference range for the healthy population.

[0120] Secondly, for the rehabilitation population, the same parameters were collected in the early, middle, and late stages of rehabilitation, and the distribution characteristics of each stage were statistically analyzed. By comparing the differences in various parameters between the healthy and rehabilitation populations, the electronic control module can divide the numerical range into several level intervals, such as "near-healthy level interval", "mildly abnormal interval", and "significantly abnormal interval".

[0121] When defining specific numerical values, the rehabilitation label database does not use a single threshold, but rather a statistical distribution interval. For example, the mean of the healthy population plus or minus one standard deviation is used as the central reference interval, and the mean plus or minus two standard deviations is used as the marginal interval, thus ensuring that the central region represents a high-confidence match and the marginal region represents a low-confidence match. For the rehabilitation patient population, numerical intervals are gradually set to adapt to the rehabilitation process based on the actual distribution characteristics at different stages.

[0122] Finally, to account for individual differences, the electronic control module compares the patient's baseline level (such as grip strength-displacement energy in the first cycle, initial rotational rhythm deviation, etc.) with the group data in the label library during the patient's first training, and proportionally adjusts the numerical range conditions based on the patient's baseline value. For example, if the patient's baseline grip strength level is significantly lower than the group average, the interval boundary of the corresponding numerical range condition is lowered during label matching, making the judgment result closer to the patient's individual reality.

[0123] After constructing and normalizing the cross-module integrated training feature vector, the electronic control module compares and analyzes this integrated training feature vector with the rehabilitation label library. During the actual matching process, the electronic control module compares each component of the integrated training feature vector with the corresponding numerical range condition in the rehabilitation label library. When the value of a component falls within the central region of the label library's numerical range, the electronic control module determines that the component highly matches the label condition and assigns it a high confidence level. When the value of a component falls within the edge region of the numerical range, the electronic control module determines that its match with the label condition is unstable and assigns it a low confidence level. If the value of a component exceeds the label library's range, the electronic control module records it as an anomaly and generates a separate warning label.

[0124] After multi-component matching is completed, the electronic control module weights the labeling results for each component. The weighting is based on an assessment by rehabilitation experts of the importance of each component in rehabilitation training, and this assessment is stored in a rehabilitation label database. For example, the grip strength fatigue accumulation index has a higher weight in endurance assessment, while the finger pressure distribution dispersion has a higher weight in coordination assessment. Based on this, the electronic control module weights and sums the confidence levels of each component to form a comprehensive score.

[0125] Finally, the electronic control module outputs three main labels based on the weighted results: movement quality level label, fatigue risk label, and coordination level label. The movement quality level label reflects the overall standardization of the patient's movements; the fatigue risk label reflects the risk level of endurance decline during training; and the coordination level label reflects the patient's coordination and rhythmic stability during multi-module training.

[0126] When outputting labels, the electronic control module provides a label level and a corresponding confidence level to indicate the reliability of the label results. A high confidence level indicates that the judgment result is stable; a low confidence level suggests that the doctor needs to further confirm the result by combining it with other assessment methods. In this way, the electronic control module realizes personalized rehabilitation assessment based on big data statistics, which can provide more accurate and interpretable references for clinical rehabilitation training.

[0127] In some embodiments of this application, when the electronic control module outputs the labeling results, it arranges the motion quality level label, fatigue risk label, and coordination level label in hierarchical order and adds a confidence level to each label. When the confidence level of at least one label is lower than the lower confidence level limit, a confirmation prompt label is added. When the confidence level of all labels is higher than the upper confidence level limit, a stability label is added. The combination of labels with stability labels is transmitted to an external terminal through the communication interface in the form of rehabilitation label codes.

[0128] It should be noted that after matching the comprehensive training feature vector with the rehabilitation label library, the electronic control module generates three core labels: movement quality level label, fatigue risk label, and coordination level label. To facilitate intuitive understanding of the results by doctors or the rehabilitation system, the electronic control module adopts a hierarchical sequential approach in its output: first, the movement quality level label is output; then, the fatigue risk label is output; and finally, the coordination level label is output. This ensures that the label results have a fixed logical order and avoids information confusion.

[0129] After each label, the electronic control module adds a confidence level. The confidence level is derived from the confidence value calculated during the matching process and is divided into multiple intervals, such as high confidence, medium confidence, and low confidence. In this way, doctors can not only see the specific level when receiving the label, but also intuitively judge the reliability of the judgment result.

[0130] Under specific circumstances, the electronic control module adds supplementary warning labels. When the confidence level of at least one label is lower than the lower confidence limit, the electronic control module generates a "pending confirmation warning label," indicating that the result of that label may be uncertain and requires manual review or confirmation using other detection methods. When the confidence levels of all labels are higher than the upper confidence limit, the electronic control module generates a "stable label," indicating that the current training state is stable and the judgment result is reliable.

[0131] To facilitate subsequent analysis and cross-system sharing, the electronic control module combines the tag results with the prompt tags into a unified data encoding structure, namely the rehabilitation tag code. This rehabilitation tag code encapsulates three core tags, prompt tags, and confidence levels into a set of data units and assigns them a unique identifier. This rehabilitation tag code can be transmitted to external terminals, including doctor workstations, mobile applications, or remote rehabilitation platforms, via a communication interface.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A multifunctional finger rehabilitation training device, characterized in that, include: The frame includes an anti-slip base and uprights fixed to the upper surface of the anti-slip base; An orifice plate training module is mounted on the frame. The surface of the orifice plate training module is provided with through holes, and a pressure sensor array is installed inside the orifice plate training module. A grip strength training module is installed on the frame. The grip strength training module includes an upper pressure plate, a lower pressure plate, and a double spring structure. The bottom surface of the lower pressure plate is connected to the anti-slip base, and the top surface is connected to the double spring structure. The bottom surface of the upper pressure plate is connected to both the double spring structure and the lower pressure plate. The double spring structure includes a force-bearing plate, guide rails, and springs. The force-bearing plate and the upper pressure plate are connected by two parallel springs, and the force-bearing plate and the lower pressure plate are connected by two guide rails. A displacement sensor is fixed at the connection between the force-bearing plate and the guide rails. A rotation training module is mounted on the frame. The rotation training module includes a textured rotating disk and a rotating shaft, and a photoelectric encoder is provided on the rotating shaft. An electronic control module is located inside the anti-slip base and is electrically connected to the pressure sensor array, displacement sensor, and photoelectric encoder. The electronic control module includes a microcontroller unit, an analog-to-digital conversion interface, a storage unit, and a communication interface. It is used to perform analog-to-digital conversion and calibration processing on the acquired analog signals, generate a dataset for analysis, and store the analysis results or transmit them to an external terminal through the communication interface. The electronic control module is configured to: model the spatial distribution of the finger pressure signal output by the pressure sensor array of the perforated plate training module, calculate the center point position of the finger pressure distribution, the dispersion of the finger pressure distribution, and the dynamic offset trajectory of the finger pressure distribution, wherein the dynamic offset trajectory is used to identify irregular sliding or compensation actions of the finger in the perforated plate training. The finger opening and closing stroke signal output by the displacement sensor of the grip strength training module is coupled with the force response of the grip strength training module to generate a grip strength-displacement response matrix. The grip strength elastic recovery coefficient, grip strength fatigue accumulation index, and grip strength coordination index are extracted from the grip strength-displacement response matrix. The grip strength fatigue accumulation index is obtained by the energy decay ratio of the multi-cycle response matrix and is used to evaluate the rate of endurance decline. The force response is calculated based on the displacement signal and spring parameters. The rotation angle signal output by the photoelectric encoder of the rotation training module is periodically segmented to establish a set of rhythmic segments of rotational motion. In each set of rhythmic segments, the rotation angle consistency coefficient, the rotation rhythm deviation degree, and the rotation sudden abnormality factor are calculated. The rotation sudden abnormality factor is obtained by detecting the number of short-time angular velocity jumps within the segment and is used to identify abnormal tremors or involuntary shaking in rotational motion. The dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation are combined to form a cross-module comprehensive training feature vector. The comprehensive training feature vector is matched with the corresponding numerical range conditions or feature combinations in the preset rehabilitation label library. Based on the matching results, the action quality level label, fatigue risk label, and coordination level label are output. When matching the comprehensive training feature vector with the rehabilitation label library, the electronic control module compares each component of the comprehensive training feature vector with the numerical range conditions in the rehabilitation label library one by one. During the matching process, a confidence level calculation method is used. When a component of the comprehensive training feature vector is located in the central region of the numerical range conditions, a high confidence level is assigned. When a component is located in the edge region of the numerical range conditions, a low confidence level is assigned. Finally, based on the weighted result of the confidence level of each label, the movement quality level label, fatigue risk label, and coordination level label are output. When outputting the label results, the electronic control module arranges the action quality level label, fatigue risk label, and coordination level label in hierarchical order and adds a confidence level to each label. When the confidence level of at least one label is lower than the lower confidence level, a confirmation prompt label is added. When the confidence level of all labels is higher than the upper confidence level, a stability label is added. The combination of labels with stability labels is transmitted to the external terminal through the communication interface in the form of rehabilitation label code.

2. The multifunctional finger rehabilitation trainer according to claim 1, characterized in that, The pressure sensor array is evenly distributed in a ring along the edge of the through hole of the orifice plate training module. Each through hole corresponds to at least three pressure sensors, which are used to collect contact reaction forces from different directions when a finger is inserted and pressure is applied. The output of the pressure sensor array is mapped to form a two-dimensional pressure distribution matrix; The electronic control module calculates the position of the center point of the finger pressure distribution based on the two-dimensional pressure distribution matrix using the weighted centroid method, and calculates the dispersion of the finger pressure distribution using the second-order center distance. The electronic control module continuously records the changing trajectory of the center point of the finger pressure distribution during the training cycle, and represents the dynamic offset trajectory of the finger pressure distribution with the cumulative offset and the instantaneous offset speed.

3. The multifunctional finger rehabilitation trainer according to claim 2, characterized in that, When the electronic control module couples and calculates the finger opening and closing stroke signal output by the displacement sensor of the grip strength training module with the force response of the grip strength training module, the displacement sensor collects the displacement of the upper pressure plate relative to the lower pressure plate in real time. The electronic control module calculates the force response of the grip strength training module based on the displacement and the spring parameters of the double spring structure, and maps the displacement and force response into a grip strength-displacement response matrix. The stiffness difference between the loading and unloading segments is extracted from the gripping force-displacement response matrix and defined as the gripping force elastic recovery coefficient. The ratio sequence of single-cycle energy to initial cycle energy is calculated in the grip force-displacement response matrix over multiple consecutive cycles, and the cumulative grip force fatigue index is obtained by accumulating the ratios. In the grip force-displacement response matrix of the same period, the ratio of the actual curve area to the ideal rectangular envelope area is used as the grip force coordination index.

4. The multifunctional finger rehabilitation trainer according to claim 3, characterized in that, When calculating the cumulative index of grip fatigue, the electronic control module divides the grip force-displacement response matrix in the continuous training cycle into segments, calculates the ratio of the loading-unloading loop energy of each cycle to the loading-unloading loop energy of the first cycle, obtains the energy decay sequence, and uses the cumulative average value of the energy decay sequence as the cumulative index of grip fatigue. When the cumulative grip strength fatigue index is higher than the first threshold, it is determined to be a high-risk fatigue state; when the cumulative grip strength fatigue index is between the first and second thresholds, it is determined to be a medium-risk fatigue state; and when the cumulative grip strength fatigue index is lower than the second threshold, it is determined to be a low-risk fatigue state.

5. The multifunctional finger rehabilitation trainer according to claim 4, characterized in that, When establishing a set of rhythmic segments for rotational motion, the electronic control module divides the rotation angle signal output by the photoelectric encoder of the rotation training module into multiple windows of equal duration according to the time sequence of the training session, and extracts the complete rotation angle cycle within each time window as a rhythmic segment to form a set of rhythmic segments for rotational motion. The electronic control module uses the average angular velocity, standard deviation of angular velocity, and duration of each rhythmic segment as segment parameters for calculating the rotation angle consistency coefficient and rotation rhythm deviation.

6. The multifunctional finger rehabilitation trainer according to claim 5, characterized in that, When calculating the rotational sudden anomaly factor, the electronic control module divides each segment in the set of rhythmic segments of the rotational motion into multiple short-time windows. Within each short-time window, it calculates the rate of change of the rotational angular velocity. When the instantaneous rate of change of the rotational angular velocity exceeds the anomaly threshold set based on the mean and standard deviation of the angular velocity of that segment, an anomaly event is recorded. The electronic control module weights and sums the number and amplitude of all anomaly events to obtain the rotational sudden anomaly factor.

7. The multifunctional finger rehabilitation trainer according to claim 6, characterized in that, When forming a cross-module comprehensive training feature vector, the electronic control module treats the dynamic offset trajectory of finger pressure distribution, the dispersion of finger pressure distribution, the elastic recovery coefficient of grip force, the cumulative index of grip force fatigue, the grip force coordination index, the consistency coefficient of rotation angle, the deviation of rotation rhythm, and the sudden abnormality factor of rotation as independent components to construct an eight-dimensional comprehensive training feature vector. The eight-dimensional comprehensive training feature vector is then normalized to ensure that each component is within a uniform numerical range, thereby guaranteeing the comparability of cross-module features.

Images