Ankle joint assisting method fusing myoelectricity and motion intention recognition

Abstract

The invention discloses an ankle joint assistance method fusing myoelectricity and motion intention recognition, and relates to the technical field of auxiliary medical treatment, and the method mainly comprises the steps: synchronously collecting a myoelectricity signal and an ankle joint kinematics signal of a calf muscle group of a user; recognizing a motion intention and an expected motion amplitude of the user before the user generates an actual motion based on signal fusion analysis; generating a self-adaptive assistance strategy according to an identification result in combination with the user capability level; the power assisting action is executed through the motor, and output is dynamically adjusted according to the real-time motion amplitude; in the exercise process, the exercise effectiveness is evaluated by analyzing the matching degree of myoelectricity activation and the exercise amplitude, and correction is conducted by adjusting a power assisting strategy or providing feedback when invalid exercise is recognized; and finally storing the motion data and continuously optimizing the evaluation model and the assistance parameters based on historical data. Through deep fusion of electromyographic signal real-time sensing and an intelligent control algorithm, a major breakthrough of ankle pump movement from traditional passive execution to intelligent active cooperation is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of auxiliary medical technology, in particular to an ankle joint assisting method fusing electromyography and motion intention recognition. BACKGROUND

[0002] In the field of clinical medicine and rehabilitation, ankle pump exercise is widely recognized as a basic and effective means to prevent deep vein thrombosis (DVT) of lower extremities. However, its actual preventive effect largely depends on the standardization, continuity and effectiveness of the exercise. At present, the ankle pump exercise assisting devices on the market can be mainly divided into two categories, but they all have obvious limitations. The first category is the passive exercise device driven by electricity or air. Although this type of device can provide passive ankle joint flexion and extension movement for patients who cannot move autonomously, its movement mode is usually fixed and rigid, and it cannot perceive and respond to the residual muscle strength or motion intention of the patient. This "one-size-fits-all" passive exercise mode not only may lead to disuse atrophy of muscles, but more importantly, due to the lack of motivation and monitoring of the patient's active participation, its pumping effect to promote blood circulation may be lower than that of active exercise, and it cannot ensure that each exercise reaches the physiologically effective amplitude required for thrombus prevention. The second category is mechanical or simple assisting tools, such as elastic bands or inclined pedal, which encourage active exercise but completely rely on the subjective initiative and physical condition of the patient. For postoperative patients who are weak, in pain or have cognitive impairment, it is difficult to ensure the frequency, amplitude and persistence of their exercise, and they are prone to abandon the exercise halfway due to fatigue or negligence, making the exercise a mere formality, i.e. "ineffective exercise". The more common defect is that the existing technology generally lacks a closed-loop, quantitative effectiveness evaluation and real-time feedback mechanism. Neither the electric device nor the simple tool can determine in real time whether the current exercise effectively activates the target muscle group, whether it reaches the preset safe activity range, and whether the muscle force is consistent with the joint movement. Therefore, neither the medical staff nor the patients themselves can know whether the exercise has truly reached the physiological effect of preventing thrombosis, which undoubtedly brings great uncertainty to the prevention of DVT. In summary, the existing technology either cannot solve the problem of lack or deficiency of patient's active exercise ability, or cannot ensure and verify the effectiveness and standardization of exercise, which is the core pain point that needs to be solved in the current ankle pump exercise assisting technology field. SUMMARY

[0003] To solve the problem of lack or deficiency of patient's active exercise ability and ensure and verify the effectiveness and standardization of exercise, the present application proposes an ankle joint assisting method fusing electromyography and motion intention recognition, comprising the steps of: S1: collecting real-time electromyography signals of target muscle groups of user's lower leg through electromyography sensors, and collecting kinematic signals of ankle joint through motion sensors; S2: Extracting electromyography signal features based on real-time electromyography signals and fusing kinematic signals to identify motion intention and expected motion amplitude before substantial physical motion of the user's ankle joint; S3: According to the identified motion intention and expected motion amplitude, the assistive force ratio is determined in combination with the user's ability level to generate an adaptive assistive force strategy; S4: According to the adaptive assistive force strategy, the motor is driven to execute the assistive action, and the motor output is dynamically adjusted based on the deviation between the actual motion amplitude identified based on the kinematic signal and the expected motion amplitude; S5: During the motion, the muscle activation level and the matching degree of the achieved motion amplitude are quantified by analyzing the real-time electromyography signal and the motion amplitude to evaluate the effectiveness of the motion, and when the motion is ineffective, the motion pattern is corrected by adjusting the assistive force strategy and / or providing feedback to the user; S6: Store motion data including motion amplitude and real-time electromyography signal, and optimize user ability evaluation model and assistive force strategy parameters based on historical data.

[0004] The present application realizes a major breakthrough from traditional passive execution to intelligent active cooperation of ankle pump exercise through deep integration of electromyography signal real-time perception and intelligent control algorithm. By predicting the user's motion intention and providing accurate and moderate assistive force before the action starts, not only the physiological effective amplitude required for ankle joint activity to prevent thrombosis is effectively guaranteed, but also the effectiveness of each motion is improved by intelligently identifying and correcting ineffective motion patterns through real-time monitoring of the matching degree of muscle activation state and motion trajectory. At the same time, through adaptive adjustment and continuous learning ability, the user's ability changes can be dynamically adapted, ensuring safety while converting ankle pump exercise into a safe and controllable, accurate and effective and comfortable active rehabilitation process, fundamentally solving the core pain points of traditional methods such as uncertain effectiveness, dependence on subjective initiative and inability to quantify evaluation.

[0005] Further, in the S2 step, identifying motion intention and expected motion amplitude specifically comprises: Extracting amplitude and slope features of real-time electromyography signals, when the amplitude exceeds the preset amplitude and the slope shows an upward trend, it is determined that there is an active motion intention; at the same time, the initial change trend of the kinematic signal is predicted to predict the expected motion amplitude.

[0006] Further, in the S3 step, the assistive force ratio is determined using an inverse proportional adjustment strategy based on the amplitude of the real-time electromyography signal: the higher the ratio of the amplitude of the real-time electromyography signal to the user's maximum voluntary contraction ability, the lower the set assistive force ratio; otherwise, the assistive force ratio is increased.

[0007] Further, in the S4 step, the dynamic adjustment of the motor output is specifically: taking the difference between the expected movement amplitude and the actual movement amplitude as the control input, using a PID control algorithm to calculate the motor torque output, and ensuring that the actual movement amplitude smoothly tracks the expected movement amplitude.

[0008] Further, in the S5 step, quantifying the effectiveness of the movement is specifically: calculating a muscle activation efficiency index, which is the normalized ratio of the actual movement amplitude to the real-time electromyographic signal amplitude, and determining that the movement is invalid when the index is lower than a preset index.

[0009] Further, in the S5 step, the feedback provided to the user includes at least one of the following ways: Guiding the movement amplitude adjustment through voice prompts; providing insufficient muscle activation feedback through vibration feedback; and displaying the deviation between the movement trajectory and the target trajectory through a visual interface.

[0010] Further, in the S3 step, the user's ability level is determined by the following method: based on the maximum achieved movement amplitude and the average muscle activation level in the historical movement data, the user's ability level is determined by a clustering algorithm.

[0011] Further, in the S6 step, the optimization of the user ability evaluation model uses an incremental learning algorithm: Based on newly collected movement data, the clustering centers of the user's ability level are regularly updated, and the incremental learning model is gradually optimized.

[0012] Further, in the S2 step, based on the real-time electromyographic signal, electromyographic signal features are extracted and kinematic signals are fused, which is specifically: The electromyographic signal features extracted based on the real-time electromyographic signal are spliced with the feature vectors of the kinematic signals into a fused feature vector, which is input into a trained classifier for movement intention recognition.

[0013] Compared with the prior art, the present application has at least the following beneficial effects: (1) The ankle joint assistance method of the present application fuses electromyography and movement intention recognition, which intelligently predicts and applies appropriate assistance force in advance by recognizing the user's movement intention and expected amplitude before movement, which not only significantly reduces the user's active force burden, but more importantly ensures that each movement can reach the physiological effective amplitude required for thrombus prevention; (2) By analyzing the matching relationship between the electromyographic activation level and the actual movement amplitude in real time, the effectiveness of each movement can be accurately quantified, and invalid movement patterns such as "muscle strong contraction but joint movement insufficient" can be identified and corrected in real time, thereby fundamentally avoiding the reduction of prevention effect caused by non-standard or insufficient movement; (3) The ability to continuously learn and optimize based on historical data enables the auxiliary strategies to be dynamically adjusted along with the user's rehabilitation process, forming a constantly evolving personalized rehabilitation plan. Attached Figure Description

[0014] Figure 1 This is a step-by-step diagram of an ankle joint assist method that integrates electromyography and motor intention recognition. Detailed Implementation

[0015] The following are specific embodiments of the present invention, which are described in conjunction with the accompanying drawings. However, the present invention is not limited to these embodiments.

[0016] Existing ankle pump-assisted exercise techniques generally fall into two main categories: The first is passive exercise equipment centered on electric or pneumatic devices. While these provide basic ankle flexion and extension movements for patients unable to move independently, they rely on fixed movement trajectories and preset parameters, lacking the ability to respond to individual patient differences and real-time physiological states. Essentially, they remain a "one-way output" mechanical movement. The second is simple assistive tools that rely on the patient's active effort, such as elastic bands or incline pedals. Their effectiveness is entirely limited by the patient's subjective initiative and physical condition, making it difficult to guarantee the standardization and continuity of the exercise. Both approaches have significant limitations—the former cannot sense the patient's movement intention and muscle strength level, potentially leading to a mismatch between passive exercise and the patient's residual muscle strength, or even causing discomfort or injury; the latter cannot ensure the quality and effectiveness of the exercise, especially when the patient is fatigued or in pain, as the range and frequency of movement often fail to meet the physiological requirements for preventing thrombosis. More significantly, existing technologies generally lack effective real-time monitoring and feedback mechanisms, making it impossible to quantitatively assess the effectiveness of the exercise. Both medical staff and patients find it difficult to determine whether each exercise session truly achieves the physiological effect of preventing thrombosis. Based on these inherent shortcomings of existing technologies, this invention proposes an ankle joint assist method that integrates electromyography and motor intention recognition. By fusing multimodal signal perception and intelligent decision-making algorithms, it achieves a technological leap from passive execution to intelligent collaboration. Figure 1 As shown, the specific steps include: S1: Real-time electromyographic signals of the target muscle group in the user's calf are collected through an electromyographic sensor, while kinematic signals of the ankle joint are collected through a motion sensor. S2: Based on real-time electromyography (EMG) signals, extract EMG signal features and fuse kinematic signals to identify movement intentions and expected range of motion before the user's ankle joint produces substantial physical movement. S3: Based on the identified movement intention and expected movement range, combined with the user's ability level, determine the assistance ratio and generate an adaptive assistance strategy; S4: Drive the motor to perform assist action according to the adaptive assist strategy, and dynamically adjust the motor output based on the deviation between the actual motion amplitude and the expected motion amplitude identified by the kinematic signal; S5: During exercise, the effectiveness of the exercise is quantified by analyzing real-time electromyographic signals and the range of motion to assess the matching between muscle activation level and the achieved range of motion. When the exercise is deemed ineffective, the exercise pattern is corrected by adjusting the assist strategy and / or providing feedback to the user. S6: Stores motion data, including amplitude of motion and real-time electromyography signals, and optimizes user ability assessment models and assist strategy parameters based on historical data.

[0017] In a preferred embodiment of the present invention, the implementation of the technical solution begins with the acquisition of electrophysiological signals from the target muscle groups of the user's lower leg. Raw electromyographic signals are continuously acquired at a high sampling frequency using surface electrodes attached to the belly of the tibialis anterior and gastrocnemius muscles. These signals then enter a preprocessing stage, first passing through an appropriate bandpass filter to eliminate common interference signals, then undergoing rectification, and finally extracting the signal envelope using a moving average window. This process effectively preserves the temporal characteristics of the electromyographic signals while suppressing noise interference. Simultaneously, a motion sensor fixed to the foot synchronously acquires kinematic data of the ankle joint in multidimensional space, including parameters such as angle, angular velocity, and acceleration. This parallel acquisition method ensures the temporal synchronization of electrophysiological and kinematic signals, providing a reliable data foundation for subsequent multimodal information fusion analysis.

[0018] After signal acquisition and preprocessing, the technical solution of this invention enters the crucial stage of motion intention recognition. The core of this stage lies in predicting the user's motion intention before they actually initiate significant joint movement by analyzing the preprocessed electromyographic (EMG) signals in real time. Specifically, by continuously monitoring the EMG signal envelope value of the target muscle group, when the signal amplitude exceeds a certain multiple of the resting baseline level and the signal slope maintains a specific trend over a continuous time period, it can be preliminarily determined that the user has a motion intention. To further improve recognition accuracy, this invention also incorporates kinematic signal-assisted judgment, analyzing the initial movement trend measured by motion sensors to form a complementary verification with the EMG signals. This multimodal recognition mechanism can accurately predict the desired direction and expected amplitude of movement shortly before the user begins to perform an action, providing ample preparation time for subsequent assisted intervention.

[0019] Based on the identified movement intention information, the technical solution of this invention enters the adaptive assistance strategy generation stage. This stage first requires assessing the user's motor ability by analyzing historical data such as electromyographic activity levels and joint range of motion, classifying users into different ability levels. Subsequently, the system dynamically calculates the required assistance ratio based on a comparison of the amplitude of real-time collected electromyographic signals with the user's individual baseline values. Specifically, when strong electromyographic activity is detected, the system correspondingly reduces the assistance level to encourage active exertion; conversely, when the user's muscle strength is insufficient, the assistance ratio is increased to ensure the movement range reaches the expected target. This intelligent adjustment mechanism effectively realizes the principle of on-demand assistance allocation, maintaining the user's active participation while ensuring the effectiveness of exercise training.

[0020] After the assist strategy is determined, the technical solution of this invention enters the motion execution and real-time control stage. This stage uses a sophisticated control algorithm to translate the preset assist strategy into specific joint movements. The control process adopts a hierarchical architecture: the upper-level controller generates an ideal target motion trajectory based on the assist strategy, while the lower-level controller adjusts the output torque in real time through the control algorithm to ensure that the actual motion trajectory can smoothly and accurately track the target value.

[0021] Specifically, the upper layer of the controller uses a fuzzy adaptive PID controller. This controller takes the deviation between the expected motion trajectory and the actual motion trajectory as its main input. Unlike traditional PID controllers, its control parameter (proportional coefficient) Integral coefficient Differential coefficients The PID controller is not fixed but rather self-tunes online through a fuzzy inference system. This system uses trajectory deviation and its rate of change as input variables and adjusts the PID controller parameters in real time based on a pre-defined fuzzy rule base. For example, when a large trajectory deviation is detected, the system automatically increases the proportional gain to improve response speed; when the motion approaches the target position, it appropriately enhances the derivative action to suppress overshoot. This adaptive mechanism effectively overcomes the problem of unstable control performance in traditional PID controllers when dealing with different users and different motion stages.

[0022] At the lower level of the control system, impedance control based on electromyography (EMG) feedforward is introduced. The unique feature of this controller is that it uses real-time acquired EMG signals as feedforward compensation to predict the user's force application trend in advance. Here, the assist system converts the processed EMG signal amplitude into the expected active torque through an established EMG-torque mapping model, and directly injects this torque value as a feedforward signal into the control loop. Simultaneously, the impedance controller calculates the auxiliary torque required to achieve the target motion trajectory based on preset virtual impedance parameters (including stiffness, damping, and mass). The combination of the feedforward channel and impedance control allows the system to provide corresponding assistance at the initial stage of the user's force application, reducing the inherent time delay of traditional feedback control and achieving true human-machine collaborative movement.

[0023] Throughout the exercise, the assistive system continuously monitors the actual movement of the ankle joint and compares it with a preset safety range. If the system detects that the movement is approaching the safety boundary, it immediately adjusts the control parameters, limiting the output torque to prevent the joint movement from exceeding the safe range. This dynamic safety protection mechanism significantly improves the safety of the training process, making it particularly suitable for users in the early stages of rehabilitation.

[0024] As the exercise progresses, the technical solution of this invention enters the effectiveness evaluation and feedback adjustment stage. This stage involves a comprehensive analysis of electromyographic signals and kinematic data to quantitatively evaluate the effectiveness of each exercise session. The core evaluation indicator is the muscle activation efficiency index, which is calculated by correlating the actual exercise effect with the corresponding electromyographic activity level. When the evaluation result is lower than the preset index, the system determines that the current exercise needs adjustment and analyzes the possible reasons for poor results: if it is due to insufficient muscle coordination, the system will appropriately adjust the assistance strategy to promote correct muscle activation patterns; if it is due to abnormal movement trajectory, the system will apply guiding torque to help the user correct the movement path; if it is due to fatigue, the assistance system will suggest rest or reduce exercise intensity through appropriate prompts. This feedback adjustment based on real-time evaluation ensures continuous optimization of training quality.

[0025] After each training session, all training data, including electromyographic signals, motor parameters, assistance strategies, and effectiveness assessment results, are properly stored and a complete training profile is established. The assistance system employs an intelligent learning algorithm to periodically update the user's ability assessment model and assistance strategy parameters based on newly acquired data. By analyzing the trends in electromyographic activity and improvements in motor ability in recent training data, the system dynamically adjusts the user's ability assessment and optimizes the calculation method of the assistance strategy accordingly. This self-optimization capability allows the method to adapt to the dynamic changes in the user's rehabilitation process, maintaining a consistently personalized training intensity.

[0026] In practical applications, users first need to complete the system calibration procedure. This process includes basic electromyography (EMG) testing and joint range of motion testing. By guiding users to perform specific movements, the system records corresponding EMG signal characteristics and joint movement characteristics, establishing personalized baseline parameters. During daily training, users simply place their feet on the training device, and the assistive system automatically begins monitoring EMG signals, recognizing movement intentions, and providing appropriate assistance. During training, the system displays the movement status and training effect in real time through a human-computer interaction interface, providing users with necessary feedback. When the system detects signs of fatigue or a decline in exercise quality, it will suggest appropriate rest through suitable prompts to avoid overtraining. After each training session, the system automatically generates a detailed training report, including key training indicators and effect evaluations, helping users and relevant professionals to fully understand the training situation.

[0027] In summary, this invention proposes an ankle joint assistance method that integrates electromyography (EMG) and movement intention recognition. Through deep integration of real-time EMG signal perception and intelligent control algorithms, it achieves a significant breakthrough in ankle pump exercise, moving from traditional passive execution to intelligent active coordination. By predicting the user's movement intention and providing precise and appropriate assistance before the movement begins, it not only effectively ensures that ankle joint activity reaches the physiologically effective range required to prevent thrombosis, but also intelligently identifies and corrects ineffective movement patterns by monitoring the matching degree between muscle activation state and movement trajectory in real time, thereby improving the actual effect of each exercise. Simultaneously, the system's personalized adaptive adjustment and continuous learning capabilities can dynamically adapt to changes in the user's abilities. Under the premise of ensuring safety, it transforms ankle pump exercise into a safe, controllable, precise, effective, and comfortable active rehabilitation process, fundamentally solving the core pain points of traditional methods, such as uncertain effects, reliance on subjective initiative, and inability to quantify and evaluate results.

[0028] It should be noted that all directional indications (such as up, down, left, right, front, back, etc.) in the embodiments of the present invention are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indication will also change accordingly.

[0029] Furthermore, in this invention, descriptions involving terms such as "first," "second," and "a" are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0030] In this invention, unless otherwise explicitly specified and limited, the terms "connection," "fixed," etc., should be interpreted broadly. For example, "fixed" can mean a fixed connection, a detachable connection, or an integral part; it can mean a mechanical connection or an electrical connection; it can mean a direct connection or an indirect connection through an intermediate medium; it can mean the internal communication of two components or the interaction between two components, unless otherwise explicitly limited. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0031] Furthermore, the technical solutions of the various embodiments of the present invention can be combined with each other, but only if they are feasible for those skilled in the art. If the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such combination of technical solutions does not exist and is not within the scope of protection claimed by the present invention.

Claims

1. A method for ankle joint assistance by fusing myoelectricity and motion intention recognition, characterized in that, The method comprises the steps of: S1: collecting real-time electromyography signals of a user's target muscle group in the calf through electromyography sensors, and collecting kinematic signals of the ankle joint through motion sensors; S2: extracting electromyography signal features based on the real-time electromyography signals and fusing the kinematic signals to identify a motion intention and an expected motion amplitude before the user's ankle joint produces substantial physical motion; S3: determining a power assistance ratio and generating an adaptive power assistance strategy according to the identified motion intention and expected motion amplitude, in combination with the user's ability level; S4: driving a motor to perform power assistance actions according to the adaptive power assistance strategy, and dynamically adjusting the motor output based on the deviation between the actual motion amplitude identified based on the kinematic signals and the expected motion amplitude; S5: during the motion, evaluating the muscle activation level and the matching of the achieved motion amplitude to quantify the motion effectiveness, and guiding the motion pattern correction by adjusting the power assistance strategy and / or providing feedback to the user when the motion is ineffective; S6: storing motion data including the motion amplitude and real-time electromyography signals, and optimizing the user ability evaluation model and the power assistance strategy parameters based on historical data.

2. The ankle joint assistance method of claim 1, wherein, In the S2 step, the identification of the motion intention and the expected motion amplitude is specifically: extracting the amplitude and slope features of the real-time electromyography signals, determining that there is an active motion intention when the amplitude exceeds a preset amplitude and the slope shows an upward trend, and predicting the expected motion amplitude according to the initial change trend of the kinematic signals.

3. The ankle joint assistance method of claim 1, wherein, In the S3 step, the determination of the power assistance ratio adopts an inverse ratio adjustment strategy based on the amplitude of the real-time electromyography signals: the higher the ratio of the amplitude of the real-time electromyography signals to the maximum autonomous contraction ability of the user, the lower the set power assistance ratio; otherwise, the power assistance ratio is increased.

4. The ankle joint assisting method of claim 1, wherein In the S4 step, the dynamic adjustment of the motor output is specifically: taking the difference between the expected motion amplitude and the actual motion amplitude as the control input, calculating the motor torque output using a PID control algorithm to ensure that the actual motion amplitude smoothly tracks the expected motion amplitude.

5. The ankle assist method of fusing myoelectricity and motion intention recognition according to claim 1, wherein, In the S5 step, the quantification of the motion effectiveness is specifically: calculating a muscle activation efficiency index, which is the normalized ratio of the actual motion amplitude to the amplitude of the real-time electromyography signals, and determining that the motion is ineffective when the index is lower than a preset index.

6. The ankle assist method of fusing myoelectricity and motion intention recognition according to claim 1, wherein, In the S5 step, the feedback provided to the user includes at least one of the following ways: guiding the motion amplitude adjustment through voice prompts; prompting muscle activation deficiency through vibration feedback; displaying the deviation between the motion trajectory and the target trajectory through a visual interface.

7. The ankle assist method of fusing myoelectricity and motion intention recognition according to claim 1, wherein, In the S3 step, the user's ability level is determined by the following method: based on the maximum achieved motion amplitude and the average muscle activation level in the historical motion data, the user's ability level is determined through a clustering algorithm.

8. The ankle assist method of fusing myoelectricity and motion intention recognition according to claim 7, wherein, In the S6 step, the optimization of the user ability evaluation model adopts an incremental learning algorithm: based on newly collected motion data, periodically updating the clustering centers of the user's ability level to gradually optimize the incremental learning model.

9. The ankle assist method of fusing myoelectricity and motion intention recognition according to claim 1, wherein, In the S2 step, the extraction of electromyography signal features based on real-time electromyography signals and the fusion of kinematic signals are specifically: Based on real-time electromyographic signal extraction, electromyographic signal features are extracted, and the feature vectors of the kinematic signals are spliced into a fusion feature vector, which is input into a trained classifier for movement intention recognition.

Images