Ankle pump exercise monitoring and intelligent guidance method and system based on internet of things

Abstract

The present application provides an ankle pump exercise monitoring and intelligent guidance method and system based on the Internet of Things, relating to the technical field of Internet of Things monitoring, comprising collecting multi-modal physiological signals and motion trajectory data through a distributed sensing unit, extracting features and identifying motion stages based on biomechanical constraint models, dynamically generating motion intensity control strategies according to rehabilitation goals, outputting synchronous guidance information through tactile or audio-visual devices, and using closed-loop correction input to iteratively modify individualized parameter matrices, achieving precise monitoring and personalized guidance, and improving rehabilitation effectiveness and compliance.

Description

Technical Field

[0001] This invention relates to Internet of Things (IoT) monitoring technology, and more particularly to an IoT-based method and system for ankle pump movement monitoring and intelligent guidance. Background Technology

[0002] Ankle pump exercises are a rehabilitation method that promotes lower limb blood circulation through active dorsiflexion and plantarflexion of the ankle joint. They are widely used in clinical rehabilitation fields such as preventing deep vein thrombosis in the lower limbs, improving venous return, and reducing lower limb edema. With the aging population and the prevalence of sedentary lifestyles, the incidence of diseases related to lower limb venous blood circulation disorders is constantly rising, and ankle pump exercises, as a simple and effective self-rehabilitation method, are receiving increasing attention.

[0003] In recent years, with the rapid development of IoT technology, wearable devices have been increasingly widely used in health monitoring and exercise guidance. Traditional ankle pump exercise guidance mainly relies on on-site guidance from medical staff or self-practice by patients, lacking objective quantitative assessment and real-time feedback mechanisms. Some researchers have attempted to use a single sensor to collect ankle joint movement data or record the number of movements using simple counting devices, but these methods are insufficient to comprehensively evaluate the quality and effectiveness of ankle pump exercises.

[0004] Most existing technologies employ single sensors or fixed-position sensing devices, failing to acquire comprehensive physiological information and precise movement trajectories during ankle pump exercises, resulting in inaccurate assessments of exercise quality. Current technologies lack biomechanical analysis models based on individual differences, making it difficult to dynamically adjust exercise parameters according to the user's physiological characteristics and rehabilitation stage, thus hindering personalized and targeted rehabilitation guidance programs. Furthermore, existing technologies generally lack real-time closed-loop feedback mechanisms, failing to provide synchronous guidance and timely error correction suggestions during exercise. Users struggle to perceive differences between their own movements and standard movements, impacting the effectiveness of rehabilitation training. Summary of the Invention

[0005] This invention provides an IoT-based method and system for ankle pump movement monitoring and intelligent guidance, which can solve the problems in the prior art.

[0006] A first aspect of the present invention provides an IoT-based method for ankle pump motion monitoring and intelligent guidance, comprising: Multimodal physiological signals and movement trajectory data of users during ankle pump exercises are collected through distributed sensing units; Based on the biomechanical constraint model, the spatiotemporal features of the multimodal physiological signals are analyzed to extract multidimensional feature vectors representing motion quality, and the boundaries and stage division markers of the motion cycle are identified based on the motion trajectory data. Based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, and combined with individual physiological adaptability parameters, a phased exercise intensity regulation strategy is dynamically generated; and a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions is generated according to the phased exercise intensity regulation strategy. After aligning the real-time feedback control signal with the boundary of the motion cycle, the tactile feedback device or audiovisual prompting device is driven to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input. The closed-loop correction input is used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

[0007] Based on a biomechanical constraint model, spatiotemporal feature analysis is performed on the multimodal physiological signals to extract multidimensional feature vectors characterizing motion quality. Furthermore, the motion trajectory data is used to identify the boundaries and phase division markers of the motion cycle, including: Based on a biomechanical constraint model, the multimodal physiological signals are decomposed in the time domain and transformed by spatial projection to obtain the joint motion trajectory deviation sequence and the muscle synergy sequence. Then, a multidimensional feature vector representing the quality of motion is extracted based on the joint motion trajectory deviation sequence and the muscle synergy sequence. The motion trajectory data is segmented based on the motion stability parameters in the multi-dimensional feature vector, the motion cycle boundaries are identified, and each motion cycle is divided into three stages: preparation, execution, and recovery, generating a stage division sequence.

[0008] Based on a biomechanical constraint model, temporal decomposition and spatial projection transformation are performed on multimodal physiological signals to obtain joint motion trajectory deviation sequences and muscle synergistic action sequences, including: Based on the biomechanical constraint model, wavelet basis functions are selected for multimodal physiological signals. The selected wavelet basis functions are then used to perform multi-scale decomposition in the time domain to obtain joint motion feature sequences. The joint motion feature sequences are then adaptively matched with the joint degree of freedom constraint parameters to generate feature sequences characterizing the deviation of joint motion trajectory. Based on the muscle contraction dynamics parameters in the biomechanical constraint model, the basis vectors of the principal component projection are determined. The multimodal physiological signals are projected onto the basis vectors in the spatial domain to obtain the muscle activation feature sequence. The muscle activation feature sequence is then matched with the muscle contraction dynamics parameters in a time sequence to generate a muscle synergistic effect sequence.

[0009] Based on the deviation between the multi-dimensional feature vector and the preset rehabilitation target state, and combined with individual physiological adaptation parameters, a phased exercise intensity regulation strategy is dynamically generated, including: The distance difference between the motor ability index in the multi-dimensional feature vector and the target motor parameter in the rehabilitation target state is calculated. The distance difference between the physiological state index in the multi-dimensional feature vector and the target physiological parameter in the rehabilitation target state is calculated. The distance difference is weighted and fused with the individual physiological adaptation parameter to obtain a comprehensive deviation index characterizing rehabilitation progress. The rehabilitation stages are divided according to the temporal variation characteristics of the comprehensive deviation index, and exercise intensity constraints based on the individual physiological adaptability parameters are established for each rehabilitation stage. An adaptive optimization algorithm is used to solve for the exercise duration, exercise frequency and exercise amplitude parameters that meet the constraints, and a staged exercise intensity regulation strategy is generated.

[0010] The real-time feedback control signal generated according to the phased motion intensity regulation strategy includes motion rhythm guidance information and amplitude correction instructions, including: The exercise duration and frequency parameters in the phased exercise intensity regulation strategy are converted into movement rhythm guidance information; A baseline motion trajectory is generated based on the motion amplitude parameters in the phased motion intensity control strategy. The deviation between the real-time collected actual motion trajectory and the baseline motion trajectory is calculated to generate an amplitude correction command. An adaptive filtering algorithm is used to fuse the motion rhythm guidance information and the amplitude correction command to generate a real-time feedback control signal.

[0011] After aligning the real-time feedback control signal with the motion cycle boundary in time, the haptic feedback device or audiovisual prompting device is driven to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input, including: The real-time feedback control signal is time-aligned with the boundary of the motion cycle to generate a periodically synchronized feedback control sequence. Based on the temporal characteristics of the feedback control sequence, tactile stimulation instructions and audiovisual prompts are generated at important moments in each motion cycle to drive the tactile feedback device to output force feedback signals and the audiovisual prompt device to output matching audio and video guidance signals, forming multi-channel synchronous guidance information. The system collects the user's motion trajectory data and response timing data under the guidance of the multi-channel synchronous information in real time. It calculates the deviation between the motion trajectory data and the expected motion trajectory, and calculates the delay between the response timing data and the motion cycle boundary. The system compensates and corrects the feedback control sequence based on the deviation and the delay to form a closed-loop correction input.

[0012] A second aspect of the present invention provides an ankle pump movement monitoring and intelligent guidance system based on the Internet of Things, comprising: The first unit is used to collect multimodal physiological signals and motion trajectory data of the user during ankle pump exercise through a distributed sensing unit; The second unit is used to perform spatiotemporal domain feature analysis on the multimodal physiological signals based on the biomechanical constraint model, extract multidimensional feature vectors that characterize the quality of movement, and identify the boundary and stage division markers of the movement cycle based on the movement trajectory data. The third unit is used to dynamically generate a phased exercise intensity regulation strategy based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, combined with individual physiological adaptability parameters; and to generate a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions based on the phased exercise intensity regulation strategy. The fourth unit is used to align the real-time feedback control signal with the motion cycle boundary in time, drive the tactile feedback device or audiovisual prompting device to output synchronous guidance information, and collect the user's real-time motion response data to form a closed-loop correction input. The closed-loop correction input is then used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

[0013] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0015] The beneficial effects of this application are as follows: By collecting multimodal physiological signals and motion trajectory data through distributed sensing units, comprehensive monitoring of ankle pump movements is achieved, improving the accuracy and comprehensiveness of data acquisition. Spatiotemporal feature analysis based on a biomechanical constraint model enables the system to accurately extract multidimensional feature vectors representing motion quality and to precisely identify and segment the motion cycle, providing a scientific basis for subsequent personalized guidance.

[0016] Based on the degree of deviation between the feature vector and the preset rehabilitation goal, and combined with individual physiological adaptability parameters, a phased exercise intensity regulation strategy is dynamically generated, realizing personalized and precise ankle pump exercise guidance and avoiding the guidance mode in traditional methods.

[0017] By aligning the feedback control signal with the boundary of the motion cycle to drive the tactile or audiovisual prompting device, the precise and synchronous output of guidance information is achieved, which greatly improves the user's compliance with performing ankle pump exercises in the correct manner. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the ankle pump motion monitoring and intelligent guidance method based on the Internet of Things, as described in an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] The technical solution of the present invention will be described in detail below with reference to specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0021] Figure 1 This is a flowchart illustrating the ankle pump motion monitoring and intelligent guidance method based on the Internet of Things according to an embodiment of the present invention. Figure 1 As shown, the method includes: Multimodal physiological signals and movement trajectory data of users during ankle pump exercises are collected through distributed sensing units; Based on the biomechanical constraint model, the spatiotemporal features of the multimodal physiological signals are analyzed to extract multidimensional feature vectors representing motion quality, and the boundaries and stage division markers of the motion cycle are identified based on the motion trajectory data. Based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, and combined with individual physiological adaptability parameters, a phased exercise intensity regulation strategy is dynamically generated; and a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions is generated according to the phased exercise intensity regulation strategy. After aligning the real-time feedback control signal with the boundary of the motion cycle, the tactile feedback device or audiovisual prompting device is driven to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input. The closed-loop correction input is used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

[0022] In one optional implementation, spatiotemporal feature analysis of the multimodal physiological signals is performed based on a biomechanical constraint model to extract multidimensional feature vectors characterizing motion quality, and the boundaries and phase division markers of the motion cycle are identified based on the motion trajectory data, including: Based on a biomechanical constraint model, the multimodal physiological signals are decomposed in the time domain and transformed by spatial projection to obtain the joint motion trajectory deviation sequence and the muscle synergy sequence. Then, a multidimensional feature vector representing the quality of motion is extracted based on the joint motion trajectory deviation sequence and the muscle synergy sequence. The motion trajectory data is segmented based on the motion stability parameters in the multi-dimensional feature vector, the motion cycle boundaries are identified, and each motion cycle is divided into three stages: preparation, execution, and recovery, generating a stage division sequence.

[0023] Based on the biomechanical constraint model, spatiotemporal features of multimodal physiological signals are analyzed to extract multidimensional feature vectors representing motion quality, and motion cycle boundaries and stage division markers are identified based on motion trajectory data.

[0024] The acquired multimodal physiological signals were preprocessed, including signal filtering, baseline correction, and normalization. Bandpass filters were used with a frequency range of 0.5-100Hz to remove environmental noise and physiological interference. Baseline correction was achieved by subtracting the signal baseline value under resting conditions. Normalization normalized the signal amplitude to the range of [-1, 1] for easier subsequent analysis and comparison.

[0025] Based on a biomechanical constraint model, the processed multimodal physiological signals were decomposed in the time domain. The time domain decomposition employed wavelet transform, selecting the Daubechies wavelet basis function to perform a five-level decomposition of the signal, extracting features from different frequency components. In rehabilitation training scenarios, the focus was on physiological signal changes within the 1-30Hz frequency band, which contains key information about muscle contraction and joint movement.

[0026] When performing spatial projection transformation, a human skeletal joint coordinate system is established, with the torso defined as the reference coordinate system, projecting limb movements into three-dimensional space. Taking upper limb rehabilitation training as an example, the origin of the shoulder joint coordinate system is defined as the acromion, with the X-axis pointing forward, the Y-axis pointing upward, and the Z-axis pointing laterally, forming a right-handed coordinate system. Through a rigid body transformation matrix, the motion data collected by the sensors is converted to a unified coordinate system, achieving spatial position standardization.

[0027] When acquiring the joint motion trajectory deviation sequence, the Euclidean distance between the actual motion trajectory and the standard trajectory is calculated based on a standard motion template. For upper limb forward extension movements, the standard trajectory is defined as the hand extending straight forward from the body midline, with the trajectory being a straight line. The deviation between the actual motion trajectory and the standard trajectory is calculated using the Euclidean distance time series, with a sampling frequency of 100Hz. When the deviation value exceeds a preset threshold (e.g., 3 cm), it is marked as an abnormal deviation point.

[0028] Simultaneously, muscle synergy sequences were acquired, and muscle activation patterns were analyzed using surface electromyography (EMG) signals. The activity of major upper limb muscle groups (including the deltoid, biceps, and triceps) was monitored, and the root mean square (RMS) value of the EMG signals was calculated as an indicator of muscle activation. The sampling window was 200 milliseconds, and the window sliding step was 50 milliseconds. Principal component analysis was used to extract muscle synergy patterns, retaining the top three principal components that explained more than 85% of the total variance, thus forming a time series of muscle synergy.

[0029] Based on the joint motion trajectory deviation sequence and muscle synergy sequence, a multi-dimensional feature vector characterizing motion quality is extracted. The feature vector includes the following dimensions: motion smoothness, trajectory accuracy, muscle coordination, motion efficiency, and motion consistency. Motion smoothness is quantified by calculating the root mean square jitter value of the velocity profile; trajectory accuracy is represented by the mean and standard deviation of the trajectory deviation sequence; muscle coordination is quantified by the weight distribution characteristics of the principal components of muscle synergy; motion efficiency is calculated by the ratio of the time required to complete the motion to the theoretical minimum time; and motion consistency is assessed by the similarity of trajectories between repeated motions.

[0030] The motion trajectory data is segmented based on the motion stability parameters extracted from the multi-dimensional feature vectors. The motion stability parameter is the jitter value of the velocity profile, calculated as the square integral of the third derivative of the velocity. A threshold is set for the motion stability parameter; when the parameter value exceeds the threshold from low to high, it is marked as the start point of the motion; when the parameter value falls below the threshold from high to low, it is marked as the end point of the motion, thus determining a complete motion cycle. In practical applications, the threshold is set to three standard deviations of the stability parameter in the resting state to ensure accurate capture of the start and end points of the motion.

[0031] After identifying the boundaries of the movement cycle, each movement cycle is divided into three phases: preparation, execution, and recovery. The preparation phase is characterized by an increase in muscle activity but minimal changes in joint displacement. The execution phase is characterized by the reaching of peak velocity and significant changes in joint displacement. The recovery phase is characterized by a gradual decrease in velocity until rest. Specifically, based on the velocity profile, the interval from zero to 20% of the peak velocity is defined as the preparation phase; the interval from 20% of the peak velocity to the peak velocity and then back to 20% of the peak velocity is defined as the execution phase; and the interval from 20% of the peak velocity to zero is defined as the recovery phase.

[0032] The final generated phase segmentation sequence includes the start and end timestamps of each exercise cycle and the boundary markers of each phase. The phase segmentation sequence is stored in the form of timestamps and corresponding phase identifiers (0 represents the non-exercise state, 1 represents the preparation phase, 2 represents the execution phase, and 3 represents the recovery phase), providing basic data support for subsequent exercise quality assessment and personalized rehabilitation program development.

[0033] In clinical applications, taking upper limb rehabilitation training for stroke patients as an example, this method can accurately identify each cycle and stage of the patient's upper limb extension-retraction movement, assess the quality of their movement, provide objective basis for rehabilitation therapists to adjust training programs, help patients establish correct movement patterns, and improve rehabilitation outcomes.

[0034] In one optional implementation, based on a biomechanical constraint model, the multimodal physiological signals are decomposed in the temporal domain and transformed by spatial projection to obtain joint motion trajectory deviation sequences and muscle synergistic action sequences, including: Based on the biomechanical constraint model, wavelet basis functions are selected for multimodal physiological signals. The selected wavelet basis functions are then used to perform multi-scale decomposition in the time domain to obtain joint motion feature sequences. The joint motion feature sequences are then adaptively matched with the joint degree of freedom constraint parameters to generate feature sequences characterizing the deviation of joint motion trajectory. Based on the muscle contraction dynamics parameters in the biomechanical constraint model, the basis vectors of the principal component projection are determined. The multimodal physiological signals are projected onto the basis vectors in the spatial domain to obtain the muscle activation feature sequence. The muscle activation feature sequence is then matched with the muscle contraction dynamics parameters in a time sequence to generate a muscle synergistic effect sequence.

[0035] When performing time-domain decomposition of multimodal physiological signals, suitable wavelet basis functions are selected based on the biomechanical constraint model. The physical characteristics of joint motion are considered when selecting wavelet basis functions. For upper limb motion, the db4 or db6 wavelet basis functions in the Daubechies wavelet family can be selected. These wavelet basis functions have good adaptability to capturing abrupt and smooth changes in joint motion. For lower limb motion, the Symlets wavelet family can be used, which has a good ability to characterize the characteristic changes in gait cycle.

[0036] In practice, the acquired joint angle signals are subjected to wavelet transform, with a decomposition scale of four levels. Detail coefficients and approximation coefficients for different frequency bands are obtained through recursive filtering. The obtained coefficients are then thresholded to remove noise and retain effective joint motion information. Finally, the signal is reconstructed using inverse wavelet transform to obtain the joint motion feature sequence.

[0037] The joint motion feature sequence is compared with a pre-established joint degree-of-freedom constraint parameter library, which contains parameters such as the angle range, velocity limit, and acceleration threshold of each joint in a normal human body. An adaptive matching algorithm is used to calculate the deviation between the feature sequence and the standard parameters. This matching process employs a dynamic programming algorithm, considering temporal characteristics and amplitude differences, and outputs a feature sequence representing the deviation of the joint motion trajectory.

[0038] In terms of spatial projection transformation, the basis vectors for principal component analysis are determined based on muscle contraction dynamics parameters, including muscle activation time constant, tendon elasticity coefficient, and muscle fiber force-length relationship. By analyzing the covariance matrix of these parameters, the main eigenvectors are extracted as basis vectors.

[0039] Multimodal physiological signals are projected onto basis vectors to generate muscle activation feature sequences. During the projection process, the signals are preprocessed, including bandpass filtering to remove power supply interference and baseline drift, followed by normalization. The projected feature sequences reflect the synergistic interaction patterns among different muscle groups.

[0040] The muscle activation feature sequence is time-series matched with muscle contraction kinetic parameters using a sliding window technique. The window size is dynamically adjusted based on the muscle activation-contraction delay time. Within each window, the cross-correlation coefficient between the muscle activation pattern and the theoretical contraction response is calculated to generate a muscle synergistic sequence.

[0041] In practical applications, such as rehabilitation training, the above methods can be used to analyze patients' movement patterns in real time. For example, when performing upper limb rehabilitation training for stroke patients, by analyzing the movement trajectory deviation sequence of the elbow joint, coordination disorders that occur during the extension process can be found. At the same time, through muscle synergy sequence analysis, abnormal co-activation phenomena between the biceps and triceps can be identified, providing a basis for adjusting the rehabilitation training program.

[0042] For motor assessment applications, a scoring system based on the above-mentioned characteristic sequences can be established. By using the Euclidean distance between the joint movement trajectory deviation sequence and the standard template as the scoring basis, and combining the spatiotemporal pattern differences in the muscle synergy sequence, a comprehensive assessment result can be given to help physicians and therapists objectively evaluate the degree of recovery of patients' motor function.

[0043] Furthermore, this method can also be used for gesture recognition in human-computer interaction systems. By analyzing the temporal characteristics of the wrist joint motion trajectory deviation sequence and combining it with the pattern characteristics of the forearm muscle synergy sequence, it can achieve accurate recognition of complex gestures and improve the naturalness and accuracy of human-computer interaction systems.

[0044] The joint motion trajectory deviation sequence and muscle synergy sequence obtained by the above methods not only reflect the dynamic characteristics of human movement, but also reveal the intrinsic mechanism of the neuromuscular system's control strategy, providing quantitative technical support for the assessment and rehabilitation of movement disorders.

[0045] In one optional implementation, a phased exercise intensity regulation strategy is dynamically generated based on the deviation between the multi-dimensional feature vector and the preset rehabilitation target state, combined with individual physiological adaptation parameters, including: The distance difference between the motor ability index in the multi-dimensional feature vector and the target motor parameter in the rehabilitation target state is calculated. The distance difference between the physiological state index in the multi-dimensional feature vector and the target physiological parameter in the rehabilitation target state is calculated. The distance difference is weighted and fused with the individual physiological adaptation parameter to obtain a comprehensive deviation index characterizing rehabilitation progress. The rehabilitation stages are divided according to the temporal variation characteristics of the comprehensive deviation index, and exercise intensity constraints based on the individual physiological adaptability parameters are established for each rehabilitation stage. An adaptive optimization algorithm is used to solve for the exercise duration, exercise frequency and exercise amplitude parameters that meet the constraints, and a staged exercise intensity regulation strategy is generated.

[0046] A multi-dimensional feature vector containing both motor ability and physiological state indicators is obtained. Motor ability indicators may include joint range of motion, muscle strength rating, balance score, and coordination measurement. For example, in lower limb rehabilitation, joint range of motion includes the range of motion of the three main joints: hip, knee, and ankle; muscle strength rating is obtained using a 0-5 scale based on manual muscle strength testing; balance can be determined using the Berg Balance Scale; and coordination is measured by evaluating performance in completing a specific sequence of movements under timed conditions. Physiological state indicators include parameters such as heart rate, blood pressure, respiratory rate, blood oxygen saturation, and fatigue level.

[0047] Define the rehabilitation target state, including target exercise parameters and target physiological parameters. Target exercise parameters can be set as standard values ​​for the exercise capacity of healthy individuals in the same age group. For example, the target range of knee flexion and extension is 0-135 degrees, and the target muscle strength of the major lower limb muscle groups is grade 4-5. Target physiological parameters are set as ideal values ​​within the individual's safe exercise range, such as heart rate not exceeding 85% of maximum heart rate during exercise, and blood pressure fluctuations controlled within ±15% of the baseline value.

[0048] Individual physiological adaptation parameters were obtained, including fatigue recovery coefficient, post-exercise heart rate recovery rate, and muscle tone regulation capacity. The fatigue recovery coefficient was calculated by continuously measuring the rate of change in subjective fatigue scores before and after exercise; the heart rate recovery rate was recorded as the time to recover to the baseline heart rate after exercise cessation; and muscle tone regulation capacity was determined by measuring the duration of muscle contraction under a given load.

[0049] The distance difference between the multi-dimensional feature vector and the rehabilitation target state is calculated. For the motor ability index, the calculation formula is: Motor ability distance difference = ∑(Current motor ability index value - Target motor parameter value). 2 For example, if a patient's knee joint range of motion is 0-90 degrees and the target range is 0-135 degrees, then the difference in this dimension is 45 degrees, which is squared and included in the summation calculation. For physiological state indicators, a similar method is used to calculate the physiological state distance difference: Physiological state indicator difference = ∑(Current physiological state indicator value - Target physiological parameter value). 2 .

[0050] The distance difference is weighted and fused with individual physiological adaptation parameters to obtain a comprehensive deviation index. The weighted fusion formula is: Comprehensive Deviation = (Motor Capacity Distance Difference × w1 + Physiological State Distance Difference × w2) / (Fatigue Recovery Coefficient × a + Heart Rate Recovery Rate × b + Muscle Tone Regulation Ability × c), where w1, w2, a, b, and c are weight coefficients that are dynamically adjusted according to the type and stage of rehabilitation. For example, in the early rehabilitation stage, w2 has a higher weight to ensure physiological safety; as rehabilitation progresses, the weight of w1 gradually increases to enhance functional recovery.

[0051] Rehabilitation stages are defined based on the temporal variation characteristics of the comprehensive deviation index. By observing the changing trend of the comprehensive deviation index during continuous rehabilitation training, a rehabilitation stage is identified as completed when the value decreases stepwise and remains stable at a certain level for a period of time. In practice, it can be set that when the change rate of the comprehensive deviation index is less than 5% for three consecutive assessments, the current stage is considered stable and the next stage can be started.

[0052] For each rehabilitation stage, exercise intensity constraints are established based on individual physiological adaptation parameters. For example, for individuals with low fatigue recovery coefficients, the upper limit of exercise duration is set to T1 minutes; for individuals with slow heart rate recovery rate, the interval between two exercise sessions is set to be no less than T2 minutes; for individuals with limited muscle tone regulation ability, the range of motion is limited to no more than P of the current joint range of motion.

[0053] An adaptive optimization algorithm is used to solve for the motion parameters under constraints. A genetic algorithm is used to construct the objective function: maximizing (rehabilitation efficiency × safety factor), where rehabilitation efficiency is positively correlated with exercise duration, frequency, and amplitude, and the safety factor is negatively correlated with the deviation of physiological indicators. Through iterative calculation, the optimal combination of exercise duration, frequency, and amplitude parameters is found.

[0054] The final stage generates a phased exercise intensity regulation strategy, which includes a specific training program table. For example, for the initial stage of lower limb functional rehabilitation, the following program can be generated: train twice a day, each session lasting 15 minutes, with an exercise frequency of 10 times / minute and an exercise range of 70% of the current activity level; rest intervals of no less than 30 minutes; if the heart rate exceeds the safety threshold or the fatigue score reaches moderate during training, immediately adjust to a low-intensity mode or pause.

[0055] As rehabilitation progresses, the system dynamically updates the exercise intensity control strategy based on the latest comprehensive deviation calculation results, ensuring that the rehabilitation process is both efficient and safe.

[0056] In one optional implementation, generating a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions according to the phased motion intensity control strategy includes: The exercise duration and frequency parameters in the phased exercise intensity regulation strategy are converted into movement rhythm guidance information; A baseline motion trajectory is generated based on the motion amplitude parameters in the phased motion intensity control strategy. The deviation between the real-time collected actual motion trajectory and the baseline motion trajectory is calculated to generate an amplitude correction command. An adaptive filtering algorithm is used to fuse the motion rhythm guidance information and the amplitude correction command to generate a real-time feedback control signal.

[0057] The generation process of real-time feedback control signals first requires processing the time and frequency parameters in the phased motion intensity control strategy. The motion duration is discretized using 100ms as the basic sampling unit to generate a timestamp sequence; simultaneously, the motion frequency parameters are periodically mapped to calculate the time interval between adjacent motion cycles. The construction of the motion rhythm guidance information adopts a multi-level structure: a 200ms continuous high-intensity cue signal with a signal strength of 1.0 is used to indicate the start of the motion; the rhythm control signal during the motion uses a rectangular wave with a 40% duty cycle, a peak value of 0.8, and a trough value of 0.2, guiding the motion rhythm through changes in signal strength; a 300ms gradual pulse is used at the end of the motion, with the signal strength linearly decreasing from 0.8 to 0, indicating the end of the motion. The rhythm guidance information also includes signal envelope modulation, adaptively adjusting the signal strength according to the motion progress to avoid user fatigue adaptation.

[0058] For the processing of motion amplitude parameters, a baseline motion trajectory is constructed based on the target motion amplitude value. A cubic spline interpolation method is used, with control points set at key nodes such as the start, peak, and end points of the motion. The positions and tangent directions of the control points are determined by the biomechanical characteristics of the motion. The interpolation interval is divided non-uniformly, increasing the density of control points in areas with significant changes in motion speed to ensure trajectory smoothness. When acquiring the user's motion trajectory in real time, the sampling frequency is set to 100Hz, and a Butterworth low-pass filter is used for signal preprocessing with a cutoff frequency of 10Hz to filter out high-frequency jitter and measurement noise.

[0059] The filtered actual trajectory and the reference trajectory are dynamically time-warped, and the Euclidean distance at corresponding time points is calculated to obtain the deviation sequence. The deviation threshold is set to 15% of the target amplitude; when the deviation exceeds the threshold, an amplitude correction command is triggered. A piecewise nonlinear mapping is used between the correction command strength and the deviation value: when the deviation value is near the threshold, the correction strength increases slowly; when the deviation value significantly exceeds the threshold, the correction strength increases rapidly, with the maximum correction strength limited to within 0.6.

[0060] The fusion of rhythm guidance information and amplitude correction commands employs an adaptive Kalman filter algorithm. The state vector contains signal amplitude, phase, and rate of change information, and the state transition equation is constructed after dimensional unification. The measurement noise covariance matrix is ​​initialized as a diagonal matrix, with diagonal elements set to 0.01 for the rhythm signal measurement variance and 0.02 for the correction command measurement variance. The process noise covariance matrix is ​​adaptively estimated using a sliding window, with a window length of 64 sampling points, and the covariance value is dynamically updated based on the statistical characteristics of the innovation sequence.

[0061] A phase compensation term is introduced into the state prediction model, and the compensation coefficient is determined by minimizing the recent phase error. The filter's gain matrix is ​​adaptively adjusted according to changes in signal characteristics, suppressing signal abrupt changes while ensuring tracking performance. The fused real-time feedback control signal uses 16-bit quantization precision, and the signal amplitude is normalized and mapped to the [-1,1] interval.

[0062] Practical Application Case: Taking hip flexion and extension exercise rehabilitation training as an example, the exercise duration is set to 20 seconds, the exercise frequency to 0.5Hz, and the target movement amplitude to 90 degrees. In the movement rhythm guidance information, the initial prompt signal lasts for 200ms with an amplitude of 1.0; the rhythm control signal has a period of 2 seconds, a high level lasts for 800ms, and an amplitude that varies in the range of [0.6, 0.8]; the completion prompt signal linearly decays from 0.8 to 0 within 300ms. The real-time acquired movement trajectory shows a maximum deviation of 16 degrees at the 8th second, exceeding the threshold of 13.5 degrees, triggering an amplitude correction command. The correction command strength at this moment is calculated to be 0.4, which is then superimposed on the rhythm guidance signal after adaptive filtering. The root mean square error of the fused signal is controlled within 0.05, the phase delay is less than 50ms, the signal transmission delay is less than 20ms, and the overall delay meets the real-time feedback requirements.

[0063] In one optional implementation, the real-time feedback control signal is time-aligned with the motion cycle boundary to drive the haptic feedback device or audiovisual prompting device to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input, including: The real-time feedback control signal is time-aligned with the boundary of the motion cycle to generate a periodically synchronized feedback control sequence. Based on the temporal characteristics of the feedback control sequence, tactile stimulation instructions and audiovisual prompts are generated at important moments in each motion cycle to drive the tactile feedback device to output force feedback signals and the audiovisual prompt device to output matching audio and video guidance signals, forming multi-channel synchronous guidance information. The system collects the user's motion trajectory data and response timing data under the guidance of the multi-channel synchronous information in real time. It calculates the deviation between the motion trajectory data and the expected motion trajectory, and calculates the delay between the response timing data and the motion cycle boundary. The system compensates and corrects the feedback control sequence based on the deviation and the delay to form a closed-loop correction input.

[0064] The system acquires real-time feedback control signals and motion cycle boundary data. Real-time feedback control signals are typically generated by the motion capture system and contain parameters of the user's current motion state, such as joint angles, acceleration, and position coordinates. Motion cycle boundary data defines the periodic feature points of a specific motion pattern. For example, in gait training, the moment the heel strikes the ground can be defined as the starting boundary of the gait cycle.

[0065] The real-time feedback control signal is time-aligned with the boundary of the motion cycle to generate a cycle-synchronized feedback control sequence. During this process, the Dynamic Time Warping (DTW) algorithm is used to calculate the optimal matching path between the real-time feedback signal and the standard motion cycle template. Based on this matching result, the time axis of the feedback signal is adjusted to match the cycle boundary. For example, in a rehabilitation training scenario, if the patient is detected to be at 25% of the motion cycle, the corresponding feedback control signal should also precisely correspond to that time point to ensure the accurate timing of subsequent tactile and audiovisual feedback.

[0066] Based on the temporal characteristics of the feedback control sequence, key moments within each movement cycle are identified. These key moments typically include the cycle start point, key turning point, and cycle end point. For example, in upper limb rehabilitation training, the start of arm raising, reaching the highest point, and returning to the initial position can be defined as key moments. In gait training, moments such as heel strike, full foot contact, and forefoot lift can be defined as key moments.

[0067] For each identified key moment, tactile stimulation commands and audiovisual cues are generated. Tactile stimulation commands include driving parameters such as vibration intensity, frequency, and duration, while audiovisual cues include audio signal parameters (such as pitch, volume, and rhythm) and visual display content (such as trajectory guide lines, target position markers, and completion indicators). Different command patterns are designed for different types of key moments. For example, for the start of a cycle, a short, strong tactile vibration and a crisp cue sound can be generated; for the point where the target is achieved, a softer, continuous vibration and a success sound effect can be generated.

[0068] The generated tactile stimulation commands are transmitted to a tactile feedback device, which outputs a force feedback signal. The tactile feedback device can take the form of a vibrating motor array, a pressure airbag, or a linear resonator, modulating the output tactile stimulation according to the parameters of the tactile stimulation command. Simultaneously, audiovisual prompts are transmitted to an audiovisual prompting device, which outputs corresponding audio-visual guidance signals. The audiovisual prompting device includes a display screen and speakers, responsible for presenting visual guidance and audio prompts, respectively. In this way, multimodal guidance information is generated synchronously across tactile, visual, and auditory channels.

[0069] The system collects real-time user response data under multi-channel synchronous guidance, including motion trajectory data and response timing data. Motion trajectory data is acquired through a motion capture system, recording the positional changes of the user's joints or limbs in three-dimensional space; response timing data records the user's reaction time to prompts at key moments. The collected motion trajectory data is compared with a preset expected motion trajectory to calculate the spatial position deviation. The deviation can be calculated using metrics such as Euclidean distance or Manhattan distance. Simultaneously, the response timing data is compared with the standard timing of the motion cycle boundary to calculate the time delay, i.e., the difference between the user's actual response time and the ideal response time.

[0070] Based on the calculated deviation and delay, the feedback control sequence is compensated and corrected. The spatial deviation is used to adjust the intensity and position of tactile and visual feedback, while the temporal delay is used to adjust the pre-trigger time of subsequent feedback. Specifically, if a systematic delay is detected in the user's response to a key point, the prompt signal at the corresponding position in the next cycle can be triggered earlier; if the user's movement trajectory deviates from the expected direction, the intensity of tactile feedback on the opposite side of the deviation is increased, and the display brightness of the correct path is enhanced in visual guidance. This dynamic adjustment based on the user's immediate response characteristics forms a closed-loop correction input, used to optimize the feedback control sequence in the next cycle.

[0071] In practical applications, such as upper limb rehabilitation training, when a patient performs repetitive arm-raising exercises, the system first identifies the start and end points of each arm-raising cycle. Then, at key turning points (such as raising the arm to a predetermined height), it triggers tactile vibration feedback, simultaneously displaying motion trajectory guidance on the monitor and playing rhythmic prompts. If the system detects insufficient arm-raising height or delayed movement, it will increase the intensity of tactile feedback at the corresponding location in the next cycle and trigger a prompt in advance, guiding the patient to adjust their movement performance, thereby achieving continuous optimization of the closed-loop training effect.

[0072] A second aspect of the present invention provides an ankle pump movement monitoring and intelligent guidance system based on the Internet of Things, comprising: The first unit is used to collect multimodal physiological signals and motion trajectory data of the user during ankle pump exercise through a distributed sensing unit; The second unit is used to perform spatiotemporal domain feature analysis on the multimodal physiological signals based on the biomechanical constraint model, extract multidimensional feature vectors that characterize the quality of movement, and identify the boundary and stage division markers of the movement cycle based on the movement trajectory data. The third unit is used to dynamically generate a phased exercise intensity regulation strategy based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, combined with individual physiological adaptability parameters; and to generate a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions based on the phased exercise intensity regulation strategy. The fourth unit is used to align the real-time feedback control signal with the motion cycle boundary in time, drive the tactile feedback device or audiovisual prompting device to output synchronous guidance information, and collect the user's real-time motion response data to form a closed-loop correction input. The closed-loop correction input is then used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

[0073] A third aspect of the present invention provides an electronic device, comprising: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the aforementioned method.

[0074] A fourth aspect of the present invention provides a computer-readable storage medium having stored thereon computer program instructions that, when executed by a processor, implement the aforementioned method.

[0075] This invention can be a method, apparatus, system, and / or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of the invention.

[0076] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An ankle pump exercise monitoring and intelligent guidance method based on the Internet of Things, characterized in that, include: Multimodal physiological signals and movement trajectory data of users during ankle pump exercises are collected through distributed sensing units; Based on the biomechanical constraint model, the spatiotemporal features of the multimodal physiological signals are analyzed to extract multidimensional feature vectors representing motion quality, and the boundaries and stage division markers of the motion cycle are identified based on the motion trajectory data. Based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, and combined with individual physiological adaptability parameters, a phased exercise intensity regulation strategy is dynamically generated; and a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions is generated according to the phased exercise intensity regulation strategy. After aligning the real-time feedback control signal with the boundary of the motion cycle, the tactile feedback device or audiovisual prompting device is driven to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input. The closed-loop correction input is used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

2. The method according to claim 1, characterized in that, Based on a biomechanical constraint model, spatiotemporal feature analysis is performed on the multimodal physiological signals to extract multidimensional feature vectors characterizing motion quality. Furthermore, the motion trajectory data is used to identify the boundaries and phase division markers of the motion cycle, including: Based on a biomechanical constraint model, the multimodal physiological signals are decomposed in the time domain and transformed by spatial projection to obtain the joint motion trajectory deviation sequence and the muscle synergy sequence. Then, a multidimensional feature vector representing the quality of motion is extracted based on the joint motion trajectory deviation sequence and the muscle synergy sequence. The motion trajectory data is segmented based on the motion stability parameters in the multi-dimensional feature vector, the motion cycle boundaries are identified, and each motion cycle is divided into three stages: preparation, execution, and recovery, generating a stage division sequence.

3. The method according to claim 2, characterized in that, Based on a biomechanical constraint model, temporal decomposition and spatial projection transformation are performed on multimodal physiological signals to obtain joint motion trajectory deviation sequences and muscle synergistic action sequences, including: Based on the biomechanical constraint model, wavelet basis functions are selected for multimodal physiological signals. The selected wavelet basis functions are then used to perform multi-scale decomposition in the time domain to obtain joint motion feature sequences. The joint motion feature sequences are then adaptively matched with the joint degree of freedom constraint parameters to generate feature sequences characterizing the deviation of joint motion trajectory. Based on the muscle contraction dynamics parameters in the biomechanical constraint model, the basis vectors of the principal component projection are determined. The multimodal physiological signals are projected onto the basis vectors in the spatial domain to obtain the muscle activation feature sequence. The muscle activation feature sequence is then matched with the muscle contraction dynamics parameters in a time sequence to generate a muscle synergistic effect sequence.

4. The method according to claim 1, characterized in that, Based on the deviation between the multi-dimensional feature vector and the preset rehabilitation target state, and combined with individual physiological adaptation parameters, a phased exercise intensity regulation strategy is dynamically generated, including: The distance difference between the motor ability index in the multi-dimensional feature vector and the target motor parameter in the rehabilitation target state is calculated. The distance difference between the physiological state index in the multi-dimensional feature vector and the target physiological parameter in the rehabilitation target state is calculated. The distance difference is weighted and fused with the individual physiological adaptation parameter to obtain a comprehensive deviation index characterizing rehabilitation progress. The rehabilitation stages are divided according to the temporal variation characteristics of the comprehensive deviation index, and exercise intensity constraints based on the individual physiological adaptability parameters are established for each rehabilitation stage. An adaptive optimization algorithm is used to solve for the exercise duration, exercise frequency and exercise amplitude parameters that meet the constraints, and a staged exercise intensity regulation strategy is generated.

5. The method according to claim 1, characterized in that, The real-time feedback control signal generated according to the phased motion intensity regulation strategy includes motion rhythm guidance information and amplitude correction instructions, including: The exercise duration and frequency parameters in the phased exercise intensity regulation strategy are converted into movement rhythm guidance information; A baseline motion trajectory is generated based on the motion amplitude parameters in the phased motion intensity control strategy. The deviation between the real-time collected actual motion trajectory and the baseline motion trajectory is calculated to generate an amplitude correction command. An adaptive filtering algorithm is used to fuse the motion rhythm guidance information and the amplitude correction command to generate a real-time feedback control signal.

6. The method according to claim 1, characterized in that, After aligning the real-time feedback control signal with the motion cycle boundary in time, the haptic feedback device or audiovisual prompting device is driven to output synchronous guidance information, and the user's real-time motion response data is collected to form a closed-loop correction input, including: The real-time feedback control signal is time-aligned with the boundary of the motion cycle to generate a periodically synchronized feedback control sequence. Based on the temporal characteristics of the feedback control sequence, tactile stimulation instructions and audiovisual prompts are generated at important moments in each motion cycle to drive the tactile feedback device to output force feedback signals and the audiovisual prompt device to output matching audio and video guidance signals, forming multi-channel synchronous guidance information. The system collects the user's motion trajectory data and response timing data under the guidance of the multi-channel synchronous information in real time. It calculates the deviation between the motion trajectory data and the expected motion trajectory, and calculates the delay between the response timing data and the motion cycle boundary. The system compensates and corrects the feedback control sequence based on the deviation and the delay to form a closed-loop correction input.

7. An ankle pump motion monitoring and intelligent guidance system based on the Internet of Things, used to implement the method of any one of claims 1-6, characterized in that, include: The first unit is used to collect multimodal physiological signals and motion trajectory data of the user during ankle pump exercise through a distributed sensing unit; The second unit is used to perform spatiotemporal domain feature analysis on the multimodal physiological signals based on the biomechanical constraint model, extract multidimensional feature vectors that characterize the quality of movement, and identify the boundary and stage division markers of the movement cycle based on the movement trajectory data. The third unit is used to dynamically generate a phased exercise intensity regulation strategy based on the degree of deviation between the multi-dimensional feature vector and the preset rehabilitation target state, combined with individual physiological adaptability parameters; and to generate a real-time feedback control signal containing movement rhythm guidance information and amplitude correction instructions based on the phased exercise intensity regulation strategy. The fourth unit is used to align the real-time feedback control signal with the motion cycle boundary in time, drive the tactile feedback device or audiovisual prompting device to output synchronous guidance information, and collect the user's real-time motion response data to form a closed-loop correction input. The closed-loop correction input is then used to iteratively correct the individualized parameter matrix in the biomechanical constraint model.

8. An electronic device, characterized in that, include: processor; Memory used to store processor-executable instructions; The processor is configured to invoke instructions stored in the memory to execute the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, When the computer program instructions are executed by the processor, they implement the method described in any one of claims 1 to 6.

Images