Ankle pump motion multi-source signal adaptive feature modeling method oriented to individual differences

Abstract

The invention provides an individual difference-oriented ankle pump motion multi-source signal adaptive feature modeling method, which comprises the following steps of: acquiring an active ankle pump motion multi-source signal of a user without task constraint, and determining an individual safe and effective motion parameter interval; establishing a personalized task parameter mapping relationship associated with a preset training task based on the parameter interval; driving the user to execute a training task based on the mapping relation, synchronously collecting multi-source time sequence signals, and extracting basic motion characteristics representing the motion control ability of the user from the signals; fusing the basic motion features and the task performance data, and constructing an individual difference feature model for quantifying the matching degree between the user state and the task configuration; according to the method, the current task adaptation degree is evaluated based on the individual difference feature model, the training task is adaptively adjusted according to the evaluation result, the strong constraint influence of the task can be stripped from the active motion signal, the individual motion control difference of the user is truly reflected, and the adaptive accurate matching of the training task is realized accordingly.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the field of information processing technology and rehabilitation medicine engineering, and particularly to a method for adaptive feature modeling of multi-source signals of ankle pump exercise oriented to individual differences. BACKGROUND

[0002] Ankle pump exercise is a basic and important training method in lower limb rehabilitation, which plays a key role in maintaining joint mobility, enhancing muscle strength and promoting blood circulation. In order to improve patient participation and training compliance, existing technologies often design training tasks in the form of game interaction, guiding users to complete actions through pre-set movement patterns, rhythms and angle requirements. The system collects movement signals generated by the user during the completion of these set tasks through sensors, and makes a simple completion judgment or difficulty adjustment accordingly.

[0003] However, in the early stage of rehabilitation or when the user's ability is limited, the training process is dominated by the system's pre-set standardized tasks, and the user is often in a passive following state. At this time, the collected movement signals mainly reflect the timing, angle and frequency constraints imposed by external tasks, rather than the user's ankle joint's real biomechanical response characteristics and movement control ability. If the movement trajectory results under such task constraints, such as angle amplitude, frequency compliance, are directly used as the basis for constructing the user's ability baseline or individual difference model, the model is essentially a passive mapping of external task parameters, and cannot effectively distinguish whether the user's response lag or amplitude attenuation is due to insufficient active control ability, or is only mechanically following the instructions. This leads to the system being able to only judge whether the task is completed, but it is difficult to identify the quality of the user's action and the real functional state behind it, so that the subsequent training task matching and stimulation intensity regulation lack precise and personalized decision basis, affecting the effectiveness and safety of rehabilitation training.

[0004] Therefore, there is an urgent need for a method that can separate the influence of task constraints from active movement signals, truly reflect the individual movement control differences of users, and achieve adaptive and precise matching of training tasks, in order to solve the problem of inaccurate personalized modeling and poor training adaptability in the prior art. SUMMARY

[0005] Therefore, in order to solve the problems brought by the prior art, the present application provides a method for adaptive feature modeling of multi-source signals of ankle pump exercise oriented to individual differences.

[0006] In a first aspect, the present disclosure provides a method for adaptive feature modeling of multi-source signals of ankle pump exercise oriented to individual differences, comprising:

[0007] S1, collect multi-source signals of active ankle pump movement of the user without task constraints, determine an individual safe and effective movement parameter interval based on the signals, and establish a personalized task parameter mapping relationship associated with a preset training task based on the parameter interval;

[0008] S2, drive the user to perform a training task based on the mapping relationship, synchronously collect multi-source time series signals, and extract basic movement features representing movement control ability of the user from the signals;

[0009] S3, fuse the basic movement features and task performance data to construct an individual difference feature model for quantifying matching degree of user state and task configuration;

[0010] S4, evaluate current task adaptation degree based on the individual difference feature model, and adaptively adjust the training task according to the evaluation result.

[0011] Optionally, the S1 comprises:

[0012] Under the guidance of no external task, multi-source signals generated by active ankle pump movement of the user are collected, including kinematic signals, muscle force signals and hemodynamic signals;

[0013] Based on the multi-source signals, an individual safe and effective movement parameter interval is determined through comprehensive stimulation evaluation index analysis, and the parameter interval at least includes an angle safety range and a frequency effective interval;

[0014] The individual safe and effective movement parameter interval is associated and bound with different preset training tasks to form the personalized task parameter mapping relationship.

[0015] Optionally, the preset training task contains multiple task categories, and each task category is preset with a functional target label corresponding to its training target.

[0016] Optionally, the S2 comprises:

[0017] Under the constraint of the personalized task parameter mapping relationship, the training task is performed, time series signals of ankle joint depression angle and rotation angle are collected in real time, and instantaneous movement frequency is calculated;

[0018] Based on the time series signals and the instantaneous movement frequency, basic movement features reflecting movement control quality of the user are extracted.

[0019] Optionally, the extraction of the basic movement features reflecting movement control quality of the user comprises:

[0020] The deviation degree and stability index between the actual movement angle and the personalized target angle interval are calculated;

[0021] Calculate the consistency between the actual movement rhythm and the personalized target frequency interval, and the rhythm stability index;

[0022] Integrate the indicators of the angle dimension and the rhythm dimension to form a standardized active ankle pump basic movement feature set.

[0023] Optionally, the S3 comprises:

[0024] Based on the stability index in the basic movement feature, filter the effective movement period, and calculate the basic movement stability ability index;

[0025] In the effective movement period, calculate the average training accuracy and task completion degree;

[0026] Analyze the movement amplitude change trend in the effective movement period, and calculate the fatigue index in combination with the basic movement stability ability index;

[0027] Weighted fusion is performed on the average training accuracy, task completion degree, basic movement stability ability index and fatigue index to output an individual difference feature model value.

[0028] Optionally, the S4 comprises:

[0029] Continuously monitor the individual difference feature model value, the basic movement stability ability index and the fatigue index, and compare them with the preset threshold to determine the task mismatch state;

[0030] When determining the task mismatch, adjust the type and task configuration of the training task;

[0031] Under the new task configuration after adjustment, perform verification to confirm the improvement of the adaptation degree, and form a closed-loop adjustment.

[0032] Optionally, the adjustment of the type and task configuration of the training task comprises:

[0033] If the current basic movement stability ability index is lower than the lower limit of the basic movement stability ability, the training task type is switched to a task type with reduced stability requirement, otherwise, the current training task is maintained;

[0034] From the user's historical training records, backtrack the individual optimal movement parameter interval corresponding to the high adaptation state, adjust the current task parameters in a gradual manner according to the individual optimal movement parameter interval, and adjust the training duration according to the fatigue index.

[0035] In a second aspect, the present disclosure provides an electronic device, comprising a memory and at least one processor, the memory stores a computer program, and the processor is used to execute the computer program to realize the method of the first aspect.

[0036] In a third aspect, the present disclosure provides a computer storage medium storing a computer program, which, when executed, implements the method of the first aspect.

[0037] The present disclosure has the following advantages compared with the prior art:

[0038] 1) By constructing a personalized task-parameter mapping relationship based on active motion acquisition, the problem of training tasks being preset by the system to reflect external constraints rather than the user's real ability is solved, and personalized training task generation based on the user's own safe and effective ability interval is realized, ensuring that the training challenge always fits the user's real functional level and safety boundary, establishing an accurate starting point for precise rehabilitation.

[0039] 2) By collecting multi-source signals and constructing multi-dimensional basic features during personalized task execution, the problem that traditional methods cannot distinguish the quality of action completion and the real motion function state is solved, and fine evaluation of the user's active control precision, stability and rhythm consistency is realized, so as to effectively distinguish the state difference between active control and passive following.

[0040] 3) By constructing an individual difference feature model integrating multi-dimensional data and performing closed-loop adaptive adjustment based on the model, the problem of lack of accurate basis for training task matching and intensity regulation due to inaccurate personalized modeling is solved, and the training system can dynamically adjust task parameters, types and duration according to the user's real-time state and historical best performance, thereby continuously maintaining the safety and effectiveness of training stimulation, and completing personalized adaptation throughout the whole cycle. BRIEF DESCRIPTION OF DRAWINGS

[0041] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present disclosure and, together with the specification, serve to explain the principles of the present disclosure.

[0042] Figure 1 A schematic diagram of a multi-source signal acquisition system for ankle pump exercise is shown.

[0043] Figure 2 A flowchart of an individual difference-oriented ankle pump exercise multi-source signal adaptive feature modeling method is shown.

[0044] Figure 3 A task adaptive adjustment closed-loop flowchart is shown.

[0045] The accompanying drawings have illustrated specific embodiments of this disclosure, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concepts of this disclosure to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0046] The present disclosure will be further described below with reference to the accompanying drawings. The following embodiments are only used to illustrate the technical solutions of the present disclosure more clearly, and should not be used to limit the scope of protection of the present disclosure.

[0047] The components of the embodiments of the invention described and illustrated herein can typically be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0048] In the following, the terms “comprising,” “having,” and their cognates, which may be used in various embodiments of the invention, are intended only to indicate a particular feature, number, step, operation, element, component, or combination thereof, and should not be construed as excluding, firstly, the presence of one or more other features, numbers, steps, operations, elements, components, or combinations thereof, or adding the possibility of one or more features, numbers, steps, operations, elements, components, or combinations thereof.

[0049] Unless otherwise specified, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which the various embodiments of the invention pertain. Terms (such as those defined in commonly used dictionaries) shall be interpreted as having the same meaning as in their contextual meaning in the relevant technical field and shall not be interpreted as having an idealized or overly formal meaning, unless clearly defined in the various embodiments of the invention.

[0050] In ankle pump rehabilitation training, to monitor user movements, it is usually necessary to use a lower limb sensor module to collect multi-source signals, such as... Figure 1 As shown. However, in the early stages of rehabilitation or when users have limited abilities, training is often dominated by pre-set tasks, with users passively following them. The signals collected at this time primarily reflect the requirements of the pre-set tasks, rather than the user's actual active motor control ability. If an individual model is directly constructed using the motion parameters constrained by such tasks, the model easily degenerates into a passive mapping of external parameters, unable to distinguish whether the user's response lag is due to insufficient ability or merely a forced attempt to follow instructions. This makes it difficult for the system to assess the quality of movement completion and the true functional state, resulting in insufficient accuracy in subsequent training adaptation.

[0051] In view of this, this disclosure provides an adaptive feature modeling method for ankle pump motion multi-source signals oriented towards individual differences. The method aims to extract real individual ability parameters based on active motion signals without task constraints, and to achieve adaptive generation and dynamic adjustment of training tasks, thereby solving the aforementioned problems of inaccurate personalized modeling and insufficient training adaptability. The method will be described in detail below with reference to the accompanying drawings and several embodiments.

[0052] Figure 2 A flowchart of the adaptive feature modeling method for multi-source signals of ankle pump movement oriented towards individual differences, as provided in the embodiments of this disclosure, is shown below. Figure 2 As shown, the process may include the following steps:

[0053] S1: Collect multi-source signals of active ankle pumping motion of the user without task constraints, determine the individual safe and effective motion parameter range based on the signals, and establish a personalized task parameter mapping relationship associated with the preset training task based on the parameter range.

[0054] We establish a personalized athletic ability benchmark for each user and precisely bind this benchmark to various ankle pump training tasks preset by the system, forming a personalized task-parameter mapping relationship. This mapping relationship is the cornerstone for subsequent personalized adaptive training, ensuring that training parameters such as angle and frequency settings always closely match the user's actual ability and safety range. This is achieved through the following sub-steps.

[0055] S1.1: Initialize a set of ankle pump training tasks containing multiple functional dimensions, and label each task with a rehabilitation functional goal.

[0056] During the initialization phase, the system pre-sets a set of ankle pump exercise training tasks containing various categories, which can be represented as follows: These tasks are designed to correspond to different functional dimensions of ankle rehabilitation. To ensure that subsequent parameter settings and adjustments are targeted, the system provides specific training tasks for each task. Label a specific functional training objective. The function training target label is a discrete classification label, for example, for identifying different training directions such as "rhythm control ability training", "angle fine adjustment ability training", "continuous output endurance training", or "lower limb coordination ability training". By pre-solidifying the function target at the task level, it avoids the equalization processing of all ankle pump movements, and provides a clear premise and basis for subsequent differentiated parameter configuration and effect evaluation according to different training targets. The training tasks are presented in the form of game interaction on the user terminal, for example, including "boat crossing" task for training rhythm control and endurance, "balloon stepping" task for training angle fine control, "rhythm striking" task for training coordination, and "garden watering" task for training endurance, etc. The system synchronously records the interaction data of the user during the execution of these game tasks, including accuracy data reflecting the matching degree of action and target, fatigue data reflecting the change of continuous movement ability, and completion data reflecting the level of task target achievement, providing multi-dimensional input for subsequent comprehensive evaluation.

[0057] S1.2: Collect kinematics, muscle strength and hemodynamic signals of user's active ankle pump movement without task guidance, and calculate the natural movement frequency.

[0058] Before the user starts any game training with a clear target, the system needs to obtain the baseline of the user's real movement ability in the natural state. For this purpose, the system guides the user to complete a series of active ankle pump movements completely relying on self-awareness control without specific game picture guidance or only with weak rhythm prompt. In this process, the system synchronously and continuously collects three types of key signals through the sensing module worn on the user's lower limbs: the first type is kinematics signals, including the trajectory of ankle joint depression angle changing with time , and the trajectory of ankle joint rotation angle changing with time , both in degrees; the second type is muscle strength representation parameter M(t) based on electromyography signal or force sensor signal, for reflecting the muscle output level in movement; the third type is local hemodynamic response parameter B(t) based on photoplethysmography signal, for evaluating the local blood circulation stimulation brought by movement. These signals together constitute a multi-source signal set representing the individual's original active ability.

[0059] In order to quantitatively analyze the user's movement rhythm from a dynamic perspective, the system analyzes the continuous angle trajectory within a set collection time window . By detecting the number of complete reciprocating motion cycles in the angle change waveform, the frequency of depression movement and the frequency of rotation movement are calculated respectively. The specific calculation method is as follows: and , wherein, represents the number of complete reciprocations of the depression angle accomplished within a time window , represents the number of complete reciprocations of the rotation angle accomplished. The time window for initial capacity assessment is usually set between 5 to 15 seconds to obtain a relatively stable frequency evaluation value. The data obtained in this step reflect the user's true movement output limit and physiological response characteristics under the condition of no external task constraints and interference, and are the basis for the calculation of all subsequent personalized parameters.

[0060] S1.3: Analyze multi-source signals to determine the safe and effective movement parameter interval of the individual through a comprehensive stimulation evaluation index.

[0061] After obtaining the original multi-source signals, the system needs to analyze and fuse them. The purpose is to determine a movement parameter interval that can produce effective rehabilitation stimulation for the current user, and will not cause excessive load or safety risks. The system constructs a comprehensive stimulation evaluation index E to quantitatively evaluate the effectiveness and safety of a certain movement state. The index is determined by three types of parameters: angle, muscle strength, and blood flow. The calculation formula is: , wherein, is the average value of the depression angle in the statistical period; is the maximum safe depression angle allowed by the system according to anatomical safety common sense, which is usually set to be between 25 degrees and 45 degrees; is the average value of the muscle strength parameter M(t) in the statistical period; is a reference muscle strength threshold value, which can be derived from the standard value of healthy people or the user's historical best data; is the average amplitude of the blood flow response parameter B(t) in the statistical period; is the blood flow safety reference threshold value. The ratio , , are all considered to be normalized and dimensionless factors reflecting the sufficiency of each dimension of stimulation, with a value range of around 0 to 1. This multiplication formula aims to reflect the synergistic effect of each dimension of stimulation. Significant deficiencies in any dimension will lead to a decrease in the value of the comprehensive index E, thereby identifying states that meet the requirements in certain aspects but are not balanced or safe overall.

[0062] The system determines through a large number of experiments and analysis that when the value of the comprehensive stimulation evaluation index E is greater than a preset stimulation effective threshold (for example, 0.3) and less than a safety threshold 1, it can be considered that the current movement state both produces sufficient physiological stimulation and is in a safe load range. Based on this criterion, the system analyzes the collected signals, reversely deduces and determines the movement parameter boundary that can make the value of E fall within this ideal interval.

[0063] Finally, the system outputs the set of safe and effective movement control parameter intervals of the individual user . The effective interval of the pressing frequency and the effective interval of the rotating frequency are determined based on the correlation between the comprehensive stimulation evaluation index E and the movement frequency, under the condition of ensuring that the value of E is within the effective stimulation range. The set specifically includes: a safe range of the pressing angle , a safe range of the rotating angle , an effective interval of the pressing frequency , and an effective interval of the rotating frequency . This process fundamentally avoids the risk of ineffective training or movement injury that may be caused by setting a training target only according to a single movement angle or frequency through the joint evaluation of multiple physiological signals.

[0064] S1.4: Associate the individual safe and effective movement parameter intervals with the training tasks of different functional targets, and establish the mapping relationship between the personalized tasks and the movement control parameters.

[0065] After obtaining the set of safe and effective movement parameter intervals of the individual user and the set of training tasks T of the calibrated functional targets, the system intelligently associates the two to generate the final personalized mapping relationship for the user. For each specific task in the set of training tasks T, the system extracts or fine-tunes a set of parameter constraints that are most suitable for the training target of the task from the set of individual parameter intervals based on the functional target label , and binds the set of parameters with the task .

[0066] This binding relationship constitutes a clear mapping M, such that Each parameter value in the mapping is rooted in the individual interval of the user. The system will intelligently fine-tune according to the type of task, for example, for a task with a functional target of "rhythm control", the system will ensure that the interval can effectively challenge the user's rhythm ability while possibly relaxing the limit of the angle range; while for a task with a functional target of "fine control", it will prioritize ensuring the The intervals are set more strictly and precisely to accurately control the training angle. The resulting complete mapping relationship M will serve as an unshakable individualized parameter benchmark for all subsequent gamified training tasks, ensuring that the difficulty and range of each training challenge presented to the user are precisely matched to their current real functional state, thereby achieving a fundamental shift from "humans adapting to machine presets" to "machines adapting to human capabilities".

[0067] Optionally, the mapping relationship M is not static but can be dynamically updated along with the user's rehabilitation process. The system can periodically (e.g., after each training phase or weekly) guide the user to repeat steps S1.2 and S1.3, collect their latest active motion signals, and recalculate the set of safe and effective motion parameter intervals. The system updates the task mapping relationship M accordingly. In this way, the system can capture the gradual improvement of the user's motor ability and automatically adjust the training benchmark to the new ability level, thereby achieving adaptive progression of rehabilitation difficulty and continuously maintaining the effectiveness of training stimulation.

[0068] In the technical solution of this disclosure, the user is guided to perform active ankle pump exercises under conditions without strong external task constraints, while simultaneously collecting multi-source signals of kinematics, muscle strength, and hemodynamics. Based on this, a safe and effective range of motion control parameters specific to each user is calculated. Furthermore, the system intelligently associates and binds this parameter range, derived from the user's actual abilities, with preset training tasks marked with clear functional rehabilitation goals, generating a personalized task-parameter mapping relationship. This method fundamentally changes the traditional rehabilitation system model where training parameters are preset by the system and the user passively follows, ensuring that the core difficulty of each training challenge is precisely anchored within the user's personal ability and safety boundaries. This lays a reliable and user-centered foundation for achieving truly personalized adaptive training.

[0069] S2: Based on the mapping relationship, the user is driven to perform training tasks, and multi-source time series signals are collected simultaneously to extract basic motion features that characterize the user's motion control ability.

[0070] During the training task execution under personalized parameter constraints, multi-source motion signals from users are collected and refined into a set of fundamental features characterizing their true motion control capabilities. These features are stripped of the influence of the absolute difficulty of the task, purely reflecting the user's performance quality, and providing standardized and quantifiable input for subsequent individual difference modeling. This is achieved through the following sub-steps.

[0071] S2.1: Call the training task and load the corresponding personalized motion control parameters to generate game interaction content.

[0072] The system calls the specific training task to be performed from the personalized mapping relationship M generated in step S1 according to the training plan . At the same time, the system synchronously loads from the relationship M a complete set of personalized motion control parameters closely bound to the task . These parameters are the safe and effective intervals previously customized for the user in the task-free state, specifically including the target range of the depression angle , the target range of the rotation angle , the target interval of the depression frequency , and the target interval of the rotation frequency .

[0073] After loading, the system uses this set of personalized parameters as the golden criteria for the running of this gamified task . The game engine will dynamically generate or adjust the interactive content according to these parameters. The game engine dynamically adjusts the game content according to the loaded personalized parameters. For example:

[0074] When performing the “balloon-popping” task, the height range of the virtual balloon appearing is strictly limited within the rotation angle of the ankle pump corresponding to the target range.

[0075] When performing the “boat crossing” task, the ideal navigation rhythm (i.e., the target ankle pump frequency) required for the boat to maintain is set within the interval .

[0076] When performing the “garden watering” task, the minimum ankle pump frequency required to maintain the water flow is , and the task base duration is .

[0077] Through this mechanism, the system ensures that the difficulty core of each gamified challenge is precisely anchored on the personal ability boundary of the user, providing a unified personalized task background for subsequent signal collection and evaluation, and making the motion performance of different users and different tasks comparable.

[0078] S2.2: Collect the time series of the ankle joint depression angle and rotation angle during task execution, and calculate the instantaneous motion frequency in real time.

[0079] With the start of the personalized game task, the user begins to actively perform the ankle pump motion under the guidance and motivation of the task target. During this period, the system continuously collects kinematic signals at a constant high sampling frequency Figure 1 through the sensing module worn on the user's lower limbs, as shown in . The main collected raw signals are the motion trajectories of the ankle joint in the sagittal plane and the coronal plane, i.e., the depression angle time series and the rotation angle time series , in degrees. The sampling frequency is usually set between 20 Hz and 100 Hz, which is sufficient to accurately capture all the dynamic details of ankle pump movement without signal distortion.

[0080] It is understood that Figure 1 the patch in the

[0081] In order to evaluate the rhythmic characteristics of movement in real time, the system uses a sliding time window for online frequency analysis. The system defines an analysis window with a short time length , usually selected between 3 seconds and 10 seconds to adapt to the rhythm changes that may occur during training. In each sliding window, the system performs waveform analysis on the angle sequence and collected in real time, identifies and counts the number of complete movement cycles in the window. Based on this, the instantaneous average depression frequency and the instantaneous average rotation frequency are calculated, which are consistent with step S1.2.

[0082] Through the above process, the system not only obtains the high-fidelity original angle-time sequence and , but also synchronously generates the frequency sequence and characterizing the movement rhythm. The four sets of data together constitute a multi-source time series signal set during the task execution. This set completely and objectively records the active movement output process of the user under the guidance of a specific task goal, which integrates subjective intention and objective ability, and is the raw data cornerstone for all subsequent advanced feature analysis and modeling.

[0083] S2.3: Calculate the deviation and stability of the actual movement angle from the personalized target angle interval.

[0084] After obtaining the original signal, the degree of agreement between the user's actual movement trajectory and the personalized target requirements is quantified, i.e., to evaluate how well it is done.

[0085] First, the angle dimension analysis is performed. The system calculates the center value of the target angle interval as the ideal target point according to the loaded target angle interval. For example, the ideal target value of the depression angle is given by the following formula: .

[0086] Then the system calculates the absolute deviation of each ankle angle sample value from the ideal target value during the analysis period, for example a complete game round or a fixed time slice , and takes the average of all sample points during the period to obtain the ankle angle deviation index , whose formula is: , where T represents the total number of sample points during the analysis period. The smaller the value of the ankle angle deviation index, the closer the user maintains the ankle angle to the ideal target during the period, and the higher the overall accuracy of the angle control. The same process is applied to the rotation angle, and the rotation angle deviation index can be calculated.

[0087] However, the average deviation alone is not enough to describe the stability of the control. A user may sometimes hit the target accurately and sometimes deviate greatly, and his average deviation may be similar to another user who always deviates slightly, but their control patterns are completely different. Therefore, the system further calculates the angle stability index. The following ankle angle stability is taken as an example, whose formula is: , where is the standard deviation of the ankle angle sequence during the current analysis period, which quantifies the degree of fluctuation of the angle value. Divide by the target center value to normalize and eliminate the impact of different target amplitudes. is a dimensionless index representing stability, whose value is between 0 and 1. The larger the value, the smaller the relative fluctuation of the angle output, and the more stable and repeatable the movement control. Similarly, the rotation angle stability index can be calculated. Through these two sets of indexes of deviation and stability, the system can clearly distinguish whether the user is in a stable and accurate control state or in a fluctuating and unstable following state.

[0088] S2.4: Calculate the consistency and stability index of the actual movement rhythm and the personalized target frequency interval.

[0089] Rhythm control in the time dimension is another core of movement function. This step aims to evaluate the matching quality of the user's movement rhythm and the task requirement rhythm.

[0090] First, the system calculates the center frequency of the loaded target frequency interval as the ideal target of rhythm control. For example, the target center frequency of the ankle depression is: .

[0091] Next, the system calculates the average of the down-pressing frequency in the current analysis period as the frequency value for evaluation . Then the down-pressing rhythm consistency index is calculated, which measures the matching degree of the frequency value to the target , and the formula is: , The index aims to map the matching degree into a reference range with 1 as the best value and 0 as the lower limit. When equals 1, it indicates perfect matching; the smaller the value, the greater the deviation. Similarly, the rotating rhythm consistency index can be obtained. The consistency index reflects the user's ability to aim at the target rhythm.

[0092] However, the system also needs to evaluate the user's endurance in maintaining a stable rhythm. A user may be able to keep up with the rhythm in a short time, but may not be able to maintain a stable frequency. Therefore, the system introduces the rhythm stability index. Take the down-pressing rhythm stability index as an example, and the calculation formula is: , where is the standard deviation of the down-pressing frequency sequence (generated by step S2.2) in the analysis period, which quantifies the size of the frequency fluctuation over time. Divided by to achieve normalization. The value range of is also between 0 and 1. The larger the value, the more constant the frequency output is in the entire analysis period, and the stronger the self-stabilization ability of rhythm control. Similarly, the rotating rhythm stability index can be calculated. By combining the consistency index and the stability index, the system can comprehensively evaluate the user's rhythm control ability and distinguish whether the user can continuously and stably output the correct rhythm or only occasionally hit or fluctuate in rhythm. In the subsequent steps, when the rhythm consistency needs to be calculated in a single valid period k, the same calculation method will be used, and it is denoted as and .

[0093] S2.5: Integrate angle and rhythm features, output active ankle pump basic movement feature set.

[0094] After the above series of refined calculations and analyses, the system aggregates and encapsulates all the generated underlying indicators to form a comprehensive active ankle pump basic movement feature set . The feature set is a data structure that contains eight core features, namely , among which, and characterize spatial accuracy; and characterize spatial stability; and characterize temporal accuracy; and characterize temporal stability.

[0095] feature set Each feature has a clear biomechanical or motor control interpretation, and their calculation is completely decoupled from the absolute difficulty of the task preset. For example, a task requiring 30 degrees and a task requiring 40 degrees, after personalized interval normalization processing, the output feature values can be directly compared. Therefore, the basic motor feature set purely depicts the level of intrinsic control ability of the user when calling his own motor system to achieve the given goal in the controlled training environment, provides a solid, reliable and directly operable data basis for step S3 to build a more complex individual difference model, and step S4 to make scientific adaptive decisions. It marks the system's key leap from perceiving movement to understanding motor ability.

[0096] In the technical solution of the embodiments of the present disclosure, under the constraint of the personalized task-parameter mapping relationship established in step S1, multi-source motor signals generated by the user when performing the training task are collected. By real-time alignment and comparative analysis of the actual movement trajectory, rhythm generated by the user and the target parameter interval customized for the user, a series of basic features of movement control accuracy, stability and rhythm consistency are quantitatively calculated. The calculation of these features decouples the influence of the absolute difficulty of the task preset, and purely reflects the level of intrinsic control ability of the user when calling his own motor system to achieve the given personalized goal, thereby converting the high-dimensional, complex original time series signal into a set of standardized, highly interpretable and directly usable for advanced modeling. Quantitative feature set, realizing the key transformation from perceiving movement to understanding motor ability.

[0097] S3: Fusion of basic motor features and task performance data to construct an individual difference feature model for quantifying the matching degree of user state and task configuration.

[0098] Fusion of basic motor features and task performance data to construct an individual difference feature model that can comprehensively evaluate the user's motor state, fatigue resistance and task adaptation. The task performance data, i.e., the accuracy, fatigue and completion data collected synchronously during the execution of the gamified task in step S1.1, provides a macroscopic performance basis for the user interaction level for the model. The model outputs a high-level decision indicator for quantifying the user's ability to maintain an effective training state under the current task configuration. This is achieved through the following sub-steps.

[0099] S3.1: Extract stability indicators from the active ankle pump base movement feature set, and calculate the base movement stability capability index.

[0100] Receiving the active ankle pump base movement feature set from step S2.5 , which contains indicators in multiple dimensions such as angle stability, rhythm stability, etc. In order to have a global quantitative evaluation of the user's movement control stability, the system selects the feature components that can directly reflect the movement repeatability and output smoothness, i.e., the down pressure angle stability , the rotation angle stability , the down pressure rhythm stability index , and the rotation rhythm stability index . The value range of these four indicators is between 0 and 1, and the larger the value, the more stable it is in that dimension.

[0101] The system integrates these four stability indicators and calculates their arithmetic mean to obtain a comprehensive base movement stability capability index , whose calculation formula is: ,

[0102] The calculated is also a scalar value between 0 and 1. This index overall depicts the individual's internal ability level to maintain the output smoothness and repeatability of active movement under the current specific task constraints. The higher the value, the more stable and controllable the user's movement system is in executing the current task, which is the physiological basis for effective and safe training.

[0103] S3.2: Screen effective movement cycles according to stability threshold.

[0104] In the continuous training process, not every action cycle has equal analysis value. The user may have occasional mistakes, brief distractions, or unclear intentions, and if these data are not distinguished and included in the model calculation, it will introduce noise and affect the judgment of true ability. Therefore, the system needs to screen the continuous task cycles generated in the training process according to the base movement stability capability.

[0105] The system uses the base stability features calculated in step S2 as screening conditions. For a given task cycle, when it simultaneously satisfies the following four conditions, the cycle is determined as an effective movement cycle: the down pressure angle stability is not lower than the preset angle stability lower threshold ; the rotation angle stability is not lower than ; downstroke rhythm stability index not less than a preset lower limit threshold of rhythm stability ; rotation rhythm stability index not less than . Among them, the threshold value and The value range is usually set between 0.5 and 0.7, representing the lowest acceptable stability level.

[0106] All cycles screened by the above conditions are summarized to form a set of valid movement cycles This screening mechanism only adopts those movement cycle data that users show sufficient stability and control. In this way, the system ensures that all subsequent calculations on training performance such as accuracy and fatigue are based on high-quality, high-intention movement data, so that the model can more accurately reflect the user's true functional state rather than their momentary errors or fluctuations.

[0107] S3.3: Calculate the average training accuracy and task completion rate in the valid movement cycle.

[0108] After obtaining a high-quality set of valid movement cycles , the system can evaluate the user's training performance within these cycles. This step mainly calculates two key performance indicators: training accuracy and task completion rate.

[0109] Training accuracy is used to quantify the accuracy of user actions within the valid cycle. The system first calculates the cycle accuracy Q(k) in each valid cycle k. Q(k) is derived from the matching degree of user actions and task goals, which is calculated by the average of the downstroke rhythm consistency and rotation rhythm consistency in the cycle, and the calculation formula is: , wherein, and are the downstroke rhythm consistency index and rotation rhythm consistency index calculated according to the method described in step S2.4 in the valid cycle k. Then, the system takes the average of the accuracy of all valid cycles to obtain the valid cycle average training accuracy index , and the calculation formula is: , wherein, represents the total number of cycles in the valid cycle set. The higher the value, the better the user's actions match the task goals within the period of stable control.

[0110] Task completion degree evaluates the user's achievement of the overall training goal from a more macro perspective. It is defined as the proportion of the number of valid periods to the total number of periods in the current training phase, and the calculation formula is: , wherein, is the total number of training periods in the current phase. The higher, the greater the proportion of time that the user can maintain stable and effective movement output in this training, and the higher the overall training efficiency. These two indicators jointly describe the user's training performance under personalized difficulty from two dimensions of quality and quantity.

[0111] S3.4: Analyze the change of movement amplitude in the valid period, and calculate the fatigue index combined with the basic stability ability.

[0112] Fatigue is a key factor affecting the sustainability of training. The system needs to quantify the accumulation of fatigue of the user in the continuous valid movement process. However, pure amplitude decline can be caused by many reasons, and the system introduces the basic movement stability ability index as a key constraint to distinguish the output decay caused by insufficient ability from real physiological fatigue.

[0113] Specifically, the system analyzes the trend of the change of the average value of the depression angle between consecutive periods in the set of valid periods . When is negative, it means that the movement amplitude has decreased. The system sums all these decay amounts and divides by the number of valid periods to get the average decay amplitude. Finally, multiply this average decay amplitude by a correction factor related to the basic stability ability, so as to get the fatigue index under the ability constraint, and the calculation formula can be expressed as: .

[0114] The design logic of this model is that the basic stability ability of the user is low, the controllability of the movement system itself is poor, and even if the amplitude decreases, it may be mainly control fluctuation rather than deep fatigue; and the output of the user is usually stable, and once there is a sustained decrease in amplitude, it is more likely to reflect the real accumulation of fatigue. Therefore, the observed decay is weighted by , so that the fatigue index can more sensitively and specifically reflect the real functional fatigue level. The greater the value, the more significant the evaluated fatigue accumulation after considering the individual ability difference.

[0115] S3.5: Weighted fusion of accuracy, completion, stability index, and fatigue, output individual difference feature model value.

[0116] The system has obtained four core intermediate indicators: the basic movement stability index reflecting the intrinsic control ability , the effective accuracy reflecting the high-quality cycle action precision , the effective completion reflecting the overall training efficiency , and the fatigue index reflecting the degree of fatigue accumulation under the ability constraint . This step fuses these four indicators with different dimensions into a unified and comprehensive individual difference feature model.

[0117] The system constructs the model through a weighted linear combination formula, and defines its output as the individual difference feature model value , and the calculation formula is: , Wherein, is the effective cycle average training accuracy index, obtained by averaging the accuracy Q(k) of each effective cycle k; 、 、 、 is a pre-set weight coefficient, and is usually set greater than zero to ensure that the basic movement ability plays a fundamental constraint role in the model. Optionally, the weight coefficient 、 、 、 can be differentiated according to different rehabilitation stages or training emphases. For example, in the early stage of rehabilitation, a higher (basic stability ability weight) and a lower (fatigue weight) can be set to ensure safety and build confidence; in the middle and later stages of rehabilitation, the (accuracy weight) and (completion weight) can be appropriately increased to improve the training intensity and precision. The system can provide a configuration interface to allow therapists to pre-set or dynamically adjust the weight combination for different users according to clinical judgment, so as to seamlessly embed professional rehabilitation strategies into the automated training process.

[0118] Described is the comprehensive effective movement output state that the individual can maintain within the range allowed by his / her own ability under the current game control parameter configuration. The higher the value is, the better the current state of the user is, which means that the user performs accurately, with high training input and low fatigue under the premise of being able to control stably, that is, the current task parameters are highly matched with the state of the user. On the contrary, The decrease of the value indicates that the matching degree may have a problem. This model successfully condenses the discrete and multi-level movement physiological and performance data into a quantitative criterion that can be directly used for subsequent decision-making, realizing the sublimation from data features to decision-making basis.

[0119] In the technical solution of the embodiments of the present disclosure, the basic movement feature set generated in step S2 is deeply fused and analyzed, and a comprehensive individual difference feature model is constructed in combination with the macroscopic interaction data exhibited by the user in the gamified task. The model first filters the effectiveness of the training period according to the movement stability, ensuring that the analysis is based on high-quality movement data. Then, the model quantitatively combines the action accuracy, overall task completion efficiency of the user in the effective period, and introduces the basic stability ability as a constraint condition to evaluate the real functional fatigue accumulation. Finally, by weightedly fusing the above multi-dimensional indexes, a unified and quantitative individual difference feature model value is output. The model successfully condenses the discrete physiological signals and performance data into a high-level decision-making index that can comprehensively evaluate the matching degree of the current state of the user and the task configuration, providing a direct and scientific basis for the adaptive regulation of the system.

[0120] S4: Evaluate the current task adaptation degree based on the individual difference feature model, and adaptively adjust the training task according to the evaluation result.

[0121] Figure 3 A task adaptive adjustment closed-loop flowchart provided by the embodiments of the present disclosure is shown. As shown in Figure 3 , based on the individual difference feature model and the rehabilitation goal, the gamified training task is evaluated and adjusted in real time. When task mismatch is detected, the system automatically adjusts the task type or configuration, so that the training task continuously matches the real-time ability and rehabilitation needs of the user, thereby dynamically maintaining the effectiveness, safety and participation of the training stimulus. This is achieved through the following sub-steps.

[0122] S4.1: Monitor the individual difference feature model value, the basic movement stability ability index and the fatigue degree, and determine the task mismatch state.

[0123] The system takes the individual difference feature model value output in step S3 as a direct quantitative index of the current task adaptation degree, that is, defines the current task adaptation degree . The value has comprehensively reflected the basic stability ability , effective accuracy , effective completion degree and fatigue degree of the user Therefore, the degree of matching between the current task configuration and the user state can be fully characterized. The higher, the better the matching degree.

[0124] The system needs to set clear criteria to determine whether the task is mismatched. For this purpose, the system sets three judgment thresholds: the lower limit threshold of the adaptation degree , used to measure whether the overall matching degree is too low; the lower limit threshold of the basic motor ability , used to ensure that the user's basic control ability has not deteriorated; the fatigue trigger threshold , used to prevent excessive fatigue. The value range of these thresholds is usually: and between 0.5 and 0.7, corresponding to the meaning of the fatigue index in step S3.4.

[0125] The system monitors the above indicators over a continuous number of evaluation periods, for example, 2 to 5 periods. If in the continuous L periods, any of the following conditions is met, the system determines that the current task is no longer matched with the individual state: the current task adaptation degree is lower than the threshold ; or the basic motor stability ability index is lower than the threshold ; or the fatigue index reaches or exceeds the threshold . The determination result is assigned a task matching state flag , when , it means that the current task is mismatched and needs to be adjusted; when , it means that the task is still matched and the current training can continue.

[0126] S4.2: Extract the motion parameters corresponding to the high adaptation state from the historical training records, and backtrack the individual optimal motion output interval.

[0127] Once the system determines that the current task is mismatched, i.e. , the system does not blindly and randomly adjust the parameters, but finds the parameter configuration corresponding to the state when the user performed best from the user's historical training records as a reference benchmark for adjustment. The system maintains a historical training record set H, each record of which contains at least the individual difference feature model value , the basic motor stability ability index and the task parameter configuration G(j) at that time.

[0128] The system filters out those segments in which the user is in a high adaptation state from the historical records. The filtering conditions are: not lower than a higher high adaptation degree threshold , the threshold value is usually in the range of 0.8 to 0.9, and at the same time not less than the lower limit threshold of basic ability These selected segments represent the moments when the user felt relaxed, performed well and stable in the task or similar tasks in the past.

[0129] For the selected high adaptation segments, the system performs weighted statistical analysis on their corresponding task parameter intervals to determine the optimal motion output interval of the individual user. The weight is defined as the proportion of the segment to the sum of all selected segments , that is: .

[0130] Then, the system calculates the optimal interval center value of the depression angle, rotation angle, depression frequency and rotation frequency respectively. For example, the optimal center value of the depression angle is calculated as: , wherein, is the target center value of the depression angle corresponding to the jth high adaptation segment. Similarly, the optimal center value of the rotation angle , the optimal center value of the depression frequency and the optimal center value of the rotation frequency can be obtained. These optimal interval center values together constitute the individual optimal motion output interval set , which reflects the most comfortable and effective motion parameter range of the individual user in history.

[0131] S4.3: According to the current basic ability state, decide whether to switch the task type, and adjust the motion parameters to the historical optimal interval, and adjust the training time according to the fatigue degree.

[0132] When the task mismatch flag , the system starts the adaptive adjustment process. The adjustment decision takes into account the current user's basic ability state and the optimal motion interval obtained by backtracking . The adjustment involves three aspects: task type, motion control parameter and training time.

[0133] First, the system decides whether to switch the task type according to the current basic motion stability ability index . If is still not less than the threshold , it means that the user's basic control ability is acceptable, and the current game task type may still be suitable, so the system will keep the current task type and only adjust its motion control parameters. Otherwise, if is less than , it indicates that the stability of the user has decreased significantly, and the current game task type may have exceeded its current controllable range. The system will automatically switch to a task type with relatively lower stability requirements. For example, from the "small boat crossing" or "rhythm strike" task with high rhythm stability requirements to the "balloon stepping" task with high angle control requirements, to match the user's current actual ability state.

[0134] Secondly, for the motion control parameters, the system adopts a gradual pullback strategy to adjust the current parameter to the historical optimal interval center value. For each motion control parameter , such as the current target depression angle center value, its new value is calculated by the following formula: , wherein, is the optimal interval center value of the corresponding parameter, is the adjustment coefficient, usually ranging from 0.2 to 0.5, to ensure that the adjustment is smooth and gradual, avoiding sudden changes in parameters that may cause discomfort to the user. At the same time, the system will ensure that the adjusted new parameter value does not exceed the personal initial safe ability interval determined in step S1, to ensure safety.

[0135] Optionally, to improve the smoothness of the adjustment process and the user's sense of touch, the system can fine-tune the adjustment coefficient used when calculating the new parameter value according to the user's recent state. Specifically, the system analyzes the fluctuation of the individual difference feature model value in the last few training sessions: if continues to be stable at a high level, it indicates that the user is well adapted and the value of can be increased moderately to allow more aggressive parameter exploration; if fluctuates greatly or shows a downward trend, it indicates that the user's state is unstable, and the value of should be reduced to take a more conservative pullback adjustment. This coefficient fine-tuning based on recent state trends makes the adaptive process both sensitive and robust.

[0136] Finally, the system adjusts the training duration of a single task according to the current fatigue index . The new training duration is obtained by multiplying the base duration by a fatigue-related factor, and is guaranteed to be positive, with the calculation formula being: , wherein, is the preset minimum positive training duration. When the user's fatigue is high, the training duration will be shortened accordingly to prevent overtraining.

[0137] Based on the above adjustments, the system generates a new set of ankle pump game task combinations and parameter configurations , which includes the task type, updated motion control parameters, and adjusted training duration.

[0138] S4.4: Run a verification period under the new task configuration to confirm the effectiveness of the adjustment and form a closed-loop adjustment.

[0139] Generate a new task configuration After that, the system does not immediately use it for long-term training, but first performs a short verification period to confirm whether the adjustment effectively improves the user's matching state. The system runs for a short period of time under the new configuration and re-collects data, executes the processes of steps S2 and S3, calculates the basic motion stability ability index , effective accuracy , effective completion , and fatigue under the new configuration, and finally obtains the new individual difference feature model value .

[0140] The system verifies whether the adjustment is effective based on the following conditions: the new is not lower than the lower limit threshold of the adaptation degree , and the new is not lower than the lower limit threshold of the basic ability . If both conditions are met, it is confirmed that the adjustment is effective, and the system will formally determine it as the effective configuration for the next long-term training , and use it as the input for the next round of training. At the same time, the process and results of this adjustment are recorded in the historical training record set H, which is used to enrich future decision-making basis.

[0141] If the verification result does not meet the above conditions, it means that the adjustment may be insufficient or excessive, and the system will reduce the adjustment coefficient , for example, by half, and then re-execute the parameter back-pulling calculation of step S4.3 to generate a new configuration and verify it again until the effective condition is met or the upper limit of the preset number of adjustment attempts is reached. Through this verification closed loop, the system ensures that each adaptive adjustment is prudent and effective, and the ultimate goal is to enable the training program to intelligently follow the dynamic changes of the user's ability without relying on human intervention, always maintaining within the user's individual optimal training stimulation interval.

[0142] In the technical solution of the embodiments of the present disclosure, based on the individual difference feature model value output in step S3, the adaptation degree of the current training task to the user state is evaluated in real time and closed-loop decision is made. When the system determines that the task is mismatched, instead of blindly or randomly adjusting, the individual optimal exercise parameter interval of the user is traced back as the adjustment reference according to the performance of the user in the historical high adaptation state. The system intelligently decides whether to maintain or switch the task type according to the current basic ability state of the user, and uses a smooth and gradual way to pull the task parameters back to the historical optimal interval, while adjusting the training duration according to the real-time fatigue degree. The effectiveness of each adjustment is confirmed through short-time verification in a closed loop to ensure the correctness of the adjustment direction. Through a series of operations, dynamic and closed-loop adaptive adjustment of the training task parameters and types is realized, so that the training system can intelligently follow the fluctuations and evolution of the user's ability, and always maintain the training stimulus in a safe, effective and personalized optimal interval.

[0143] Optionally, in some embodiments, the system can also use efficacy as a training endpoint evaluation target. Specifically, after completing the closed-loop adaptive adjustment and forming an effective configuration , the efficacy evaluation phase can be entered: the system takes the active exercise signal data of the user collected in step S1.2 without task constraints as the baseline, and after completing N training sessions (N is 3 to 10 times), the user is guided to perform active ankle pump exercise without task constraints again, and the same source signal .

[0144] The system calculates: Kinematic improvement amount , Muscle strength improvement amount , Hemodynamic improvement amount , And the normalized , , . Based on this, the efficacy score : ,

[0145] Among them, , , is the weight coefficient, which can be 0.2 to 0.5 and satisfies .

[0146] When the efficacy score is not less than the preset efficacy threshold ( which can be 0.15 to 0.30), that is , and at the same time meets the actual exercise parameters (such as and Stable within its individual safe and effective range of motion parameters Internal (e.g., its basic motion stability index) Not lower than the lower threshold When ankle pump training is deemed effective, corresponding indicators (such as...) can be output to determine if the treatment is satisfactory. Otherwise, the treatment effect may be deemed unsatisfactory (e.g., And trigger the training task target label. A re-evaluation of the mapping relationship with the personalized task parameters (i.e., the mapping relationship M established in step S1.4) is conducted to make policy-level adjustments.

[0147] According to embodiments of this disclosure, an electronic device is also provided, which may include a processor, a communications interface, a memory, and a communication bus, wherein the processor, the communications interface, and the memory communicate with each other via the communication bus. The processor can invoke logical instructions stored in the memory to execute the methods provided in the above embodiments.

[0148] Furthermore, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0149] On the other hand, this disclosure also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the methods provided in the above embodiments.

[0150] The device embodiments described above are merely illustrative, wherein the units described as separate components can or can not be physically separate, and the components displayed as units can or can not be physical units, i.e., can be located in one place, or can be distributed to multiple network units. Part or all of the modules can be selected to achieve the purposes of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0151] Through the description of the above embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary universal hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of software products, and the computer software products can be stored in a computer readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and include a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0152] It should be understood that the above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than limit them; although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some technical features; and these modifications or replacements do not make the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A method for adaptive feature modeling of multi-source signals of ankle pump motion oriented towards individual differences, characterized in that, The method includes: S1. Collect multi-source signals of active ankle pumping motion of the user without task constraints, determine the individual safe and effective motion parameter range based on the signals, and establish a personalized task parameter mapping relationship associated with the preset training task based on the parameter range. S2. Based on the mapping relationship, drive the user to perform training tasks, and simultaneously collect multi-source time series signals to extract basic motion features that characterize the user's motion control ability. S3. Integrate the basic motion features and task performance data to construct an individual difference feature model for quantifying the matching degree between user status and task configuration; S4. Evaluate the current task fit based on the individual difference feature model, and adaptively adjust the training task according to the evaluation results.

2. The method for adaptive feature modeling of multi-source signals of ankle pump movement oriented towards individual differences according to claim 1, characterized in that, S1 includes: Without external task guidance, multi-source signals generated by the user's active ankle pump movement are collected, including kinematic signals, muscle force signals and hemodynamic signals; Based on the multi-source signals, the safe and effective range of motion parameters for an individual is determined through comprehensive stimulus evaluation index analysis. The parameter range includes at least the safe range of angles and the effective range of frequency. The individual safe and effective motion parameter range is associated and bound with different preset training tasks to form the personalized task parameter mapping relationship.

3. The adaptive feature modeling method for multi-source signals of ankle pump movement oriented towards individual differences according to claim 2, characterized in that, The preset training tasks include multiple task categories, and each task category has a preset functional target label corresponding to its training objective.

4. The adaptive feature modeling method for multi-source signals of ankle pump motion oriented towards individual differences according to claim 1, characterized in that, S2 includes: The training task is executed under the constraints of the personalized task parameter mapping relationship, and the time series signals of ankle joint compression angle and rotation angle are collected in real time, and the instantaneous motion frequency is calculated. Based on the time series signal and instantaneous motion frequency, basic motion features reflecting the quality of user motion control are extracted.

5. The adaptive feature modeling method for multi-source signals of ankle pump motion oriented towards individual differences according to claim 4, characterized in that, The extraction of basic motion features reflecting the quality of user motion control includes: Calculate the deviation and stability index between the actual motion angle and the personalized target angle range; Calculate the consistency and rhythm stability index between the actual exercise rhythm and the personalized target frequency range; By integrating indicators from both angular and rhythmic dimensions, a standardized set of basic movement characteristics for active ankle pumps is formed.

6. The method for adaptive feature modeling of multi-source signals of ankle pump motion oriented towards individual differences according to claim 1, characterized in that, S3 includes: Effective motion cycles are selected based on the stability index in the basic motion characteristics, and the basic motion stability index is calculated. Within the effective exercise cycle, the average training accuracy and task completion rate are calculated; Analyze the trend of motion amplitude change within the effective motion cycle, and calculate the fatigue index in combination with the basic motion stability index; The average training accuracy, task completion rate, basic motor stability index, and fatigue index are weighted and fused to output individual difference feature model values.

7. The adaptive feature modeling method for multi-source signals of ankle pump motion oriented towards individual differences according to claim 1, characterized in that, S4 includes: Continuously monitor the individual difference characteristic model values, basic motor stability index and fatigue index, and compare them with preset thresholds to determine the task mismatch status; When a task mismatch is determined, the type and configuration of the training task are adjusted. Execute verification under the adjusted new task configuration, and form a closed-loop adjustment after confirming the improved adaptability.

8. The method for adaptive feature modeling of multi-source signals of ankle pump motion oriented towards individual differences according to claim 7, characterized in that, The adjustment of the training task type and task configuration includes: If the current basic motor stability index is lower than the lower limit of basic motor stability, then switch the training task type to a task type with lower stability requirements; otherwise, maintain the current training task. The system retrieves the individual's optimal motion parameter range from the user's historical training records when the user was in a highly adapted state. Based on this range, the current task parameters are adjusted progressively, while the training duration is adjusted according to the fatigue index.

9. An electronic device, characterized in that, The electronic device includes a memory and at least one processor, the memory storing a computer program, and the processor executing the computer program to implement the adaptive feature modeling method for multi-source signals of ankle pump movement oriented towards individual differences, as described in any one of claims 1-8.

10. A computer storage medium, characterized in that, It stores a computer program, which, when executed, implements the adaptive feature modeling method for multi-source signals of ankle pump movement oriented towards individual differences, according to any one of claims 1-8.

Images