Rehabilitation training method and system integrating rehabilitation robot and cloud platform

Abstract

The invention relates to the technical field of rehabilitation training, in particular to a rehabilitation training method and system integrating a rehabilitation robot and a cloud platform. Comprising the following steps: constructing a standardized motion track data set; joint time sequence features are extracted, coordination deviation in the multi-joint coordination process is analyzed, time sequence coordination deviation distribution is determined, and when deviation exceeds a preset range, an abnormal deviation track and a compensation behavior mode are recognized; generating a personalized training protocol scheme in combination with the compensatory behavior pattern and patient personalized data, and obtaining an optimized motion guidance path through multi-joint collaborative path simulation and parameter correction; training feedback indexes are extracted based on the motion guidance path, the training effect is judged, rehabilitation robot control parameters are dynamically updated, and a continuously optimized function recovery path is formed. The problems that in existing rehabilitation training, multi-joint collaborative evaluation is insufficient, and a training scheme is difficult to dynamically adapt are solved, and intelligentization, individualization and continuous optimization of the rehabilitation training process are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rehabilitation training, and in particular to a rehabilitation training method and system fusing a rehabilitation robot and a cloud platform. BACKGROUND

[0002] With the acceleration of population aging and the continuous rise in the incidence of disabling diseases such as stroke, bone and joint diseases, and sports injuries, rehabilitation training has become an important medical means to restore the motor function of patients and improve their quality of life. In particular, in the field of upper limb function rehabilitation, patients often need to complete functional movements involving the coordination of multiple joints such as shoulders, elbows, and wrists in their daily activities. The movement recovery process not only depends on the improvement of single joint strength, but also depends on the coordination of multiple joints in time and space.

[0003] Although some existing rehabilitation robot systems have introduced sensors to collect patient movement data, the way they use the data is mostly limited to the position, angle, or simple speed level, making it difficult to identify the compensatory behavior of patients in the training process due to insufficient function in a timely manner. In addition, the existing systems generally lack a long-term and continuous evaluation mechanism for training effectiveness, and the adjustment of training parameters relies mostly on manual intervention, making it difficult to form a closed-loop optimization process based on data feedback. At the same time, with the development of cloud computing and Internet of Things technology, uploading rehabilitation training data to a cloud platform for centralized storage and analysis has become a trend. However, existing cloud-based rehabilitation platforms mostly focus on training record management and remote monitoring, and fail to achieve real-time or quasi-real-time adjustment of local training paths and control parameters based on cloud analysis results, making it difficult to fully leverage the advantages of cloud platforms in big data analysis and continuous optimization.

[0004] The present application is proposed based on the above problems. By fine collection and analysis of the multi-joint movement trajectory of the patient, the joint timing coordination deviation and compensatory behavior pattern are accurately identified, and the training scheme is dynamically generated and optimized in combination with the individualized characteristics of the patient. At the same time, the cloud platform is used to realize the continuous accumulation and feedback regulation of training data, thereby building a scientific, quantifiable, and continuously optimized rehabilitation training system to improve the safety, effectiveness, and individual adaptation of rehabilitation training. SUMMARY

[0005] The present application provides a rehabilitation training method and system fusing a rehabilitation robot and a cloud platform, which is used to accurately identify the joint timing coordination deviation and compensatory behavior pattern by fine collection and analysis of the multi-joint movement trajectory of the patient, and dynamically generate and optimize the training scheme in combination with the individualized characteristics of the patient.

[0006] In a first aspect, the present application provides a rehabilitation training method fusing a rehabilitation robot and a cloud platform, which comprises: Step S1: collecting joint movement trajectory data of a patient in a functional action through a sensor to generate original three-dimensional trajectory information; preprocessing the original three-dimensional trajectory information to obtain a standardized movement trajectory data set; Step S2: extracting joint time sequence features according to the standardized movement trajectory data set, analyzing coordination deviation and determining a time sequence coordination deviation distribution according to the joint time sequence features; if the time sequence coordination deviation distribution exceeds a preset range, separating an abnormal offset trajectory and judging a compensatory behavior mode; Step S3: generating a personalized training protocol scheme according to the compensatory behavior mode and patient individualized data; simulating a multi-joint collaborative path and performing collaborative parameter correction according to the personalized training protocol scheme to determine an optimized movement guidance path; Step S4: extracting feedback indicators according to the optimized movement guidance path, judging training effect and updating control parameters of a rehabilitation robot to generate a continuously optimized functional recovery path.

[0007] As a preferred technical solution of the present application, in step S1, the original three-dimensional trajectory information is generated, including: A multi-angle sensor is used to collect data of the shoulder, elbow and wrist joints of the patient in a specific functional action, and the movement parameters of each joint are recorded; trajectory data containing three-dimensional spatial position information is generated according to the movement parameters; the trajectory data is preliminarily integrated to form the original three-dimensional trajectory information; The original three-dimensional trajectory information is subjected to integrity detection through a data verification mechanism to meet the requirement that the data cover the movement range of all target joints.

[0008] As a preferred technical solution of the present application, in step S1, the standardized movement trajectory data set is obtained, including: For discontinuous points in the original three-dimensional trajectory information, interpolation technology is used for data completion; a smoothing processing algorithm is used to optimize the curve of the original three-dimensional trajectory information to eliminate jitter interference in the collection process; the time axis and the space axis of the original three-dimensional trajectory information are subjected to normalization processing according to a preset standardization rule; the weight parameters of the original three-dimensional trajectory information are adjusted according to the movement characteristics of different joints to generate a standardized movement trajectory data set.

[0009] As a preferred technical solution of the present application, in step S2, the time sequence coordination deviation distribution is determined, including: Based on the standardized motion trajectory dataset, joint motion data for different training stages are divided; for each training stage, feature parameters of each joint in the time dimension are extracted; by using the delay parameter adjustment technique, the temporal differences between joints are calculated; the coordination of joint motion is analyzed based on the temporal differences to identify potential deviation patterns; and the deviation patterns are integrated to generate the temporal coordination deviation distribution.

[0010] As a preferred embodiment of the present invention, in step S2, if the timing coordination deviation distribution exceeds a preset range, the abnormal offset trajectory is separated and the compensatory behavior pattern is determined, including: By comparing the temporal coordination deviation distribution with the preset natural motion reference range, the joints with deviations exceeding the limit are identified; abnormal offset trajectory data of non-target joints are separated from the original three-dimensional trajectory information; and the causes of the abnormal offset trajectory data are analyzed in conjunction with the neural conduction velocity adaptation mechanism. Based on the analysis results, determine whether there is an abnormal pattern of compensatory behavior; generate feature labels corresponding to the abnormal pattern.

[0011] As a preferred embodiment of the present invention, step S3, generating a personalized training protocol scheme, includes: Based on the judgment results of the compensatory behavior pattern, key limiting factors of the patient's motor function are extracted; combined with the patient's individualized data, the patient's exercise habits and physical condition are analyzed; through staggered interval dynamic configuration technology, the functional movements are decomposed and the training rhythm is adjusted; based on the decomposed movement units, targeted training modules are designed; and the training modules and individualized data are integrated to generate the personalized training protocol scheme.

[0012] As a preferred embodiment of the present invention, step S3, determining the optimized motion guidance path, includes: Based on the personalized training protocol, a coordinated motion model of the shoulder, elbow, and wrist joints is constructed; the parameters of the coordinated motion model are adjusted using motion path simulation optimization technology; the motion trajectory parameters of each joint are corrected to address deviations during the simulation process; cross-device data synchronization is achieved through a time-series data cloud storage mechanism; and an optimized motion guidance path is generated based on the corrected parameters.

[0013] As a preferred embodiment of the present invention, step S4, forming a continuously optimized function recovery path, includes: Based on the optimized exercise guidance path, muscle function feedback indicators are extracted; the changing trends of the feedback indicators are analyzed through a real-time feedback loop update mechanism; if the feedback indicators meet the preset neural recovery threshold assessment criteria, the degree of improvement in the training effect is determined; based on the determination of the training effect, the parameters of the rehabilitation robot control module are dynamically adjusted; a closed-loop regulation is formed through the feedback data stored in the cloud to generate a continuously optimized functional recovery path.

[0014] Secondly, the present invention also provides a rehabilitation training system integrating a rehabilitation robot and a cloud platform for implementing the above-mentioned method, the system comprising: The joint trajectory acquisition unit is used to acquire joint motion trajectory data of the patient during functional movements through sensors, generate raw three-dimensional trajectory information, and preprocess the raw three-dimensional trajectory information to obtain a standardized motion trajectory dataset. The timing deviation analysis unit is used to extract joint timing features based on the standardized motion trajectory dataset, analyze coordination deviations based on the joint timing features, and determine the timing coordination deviation distribution. The compensation pattern recognition unit is used to separate abnormal offset trajectories and determine compensation behavior patterns when the timing coordination deviation distribution exceeds a preset range. The training path generation unit is used to generate a personalized training protocol scheme based on the compensatory behavior pattern and individualized patient data; simulate multi-joint collaborative paths and perform collaborative parameter corrections based on the personalized training protocol scheme to determine the optimized motion guidance path; The feedback control and optimization unit is used to extract feedback indicators based on the optimized motion guidance path, judge the training effect, update the control parameters of the rehabilitation robot, and generate a continuously optimized functional recovery path.

[0015] Thirdly, the present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0016] The beneficial effects of this invention are as follows: This invention continuously collects and preprocesses the patient's joint movement trajectories using multi-angle sensors to construct a high-quality, standardized motion trajectory dataset. This accurately depicts the collaborative relationships of multiple joints over time, avoiding the limitations of traditional methods that rely solely on single-joint or instantaneous indicators. By analyzing joint temporal features and comparing them with natural motion benchmarks, it identifies coordination deviations in different training stages and separates abnormal deviation trajectories. Combined with neural conduction velocity adaptation mechanisms, it determines compensatory behavior patterns, improving the accuracy of compensation judgment by incorporating physiological mechanisms in addition to trajectory morphology, thus avoiding misinterpreting normal individual differences as abnormalities. Based on compensatory behavior patterns and individualized patient data, it generates personalized training protocols and transforms these training strategies into executable movement instructions through multi-joint collaborative modeling and path simulation optimization. The guided training pathway effectively connects "problem identification" to "intervention design," ensuring that the training pathway matches the patient's current ability level in terms of time, space, and rhythm, reducing training risks and improving training efficiency. Finally, by collecting and analyzing the execution feedback of the optimized exercise guidance pathway in real time, the degree of improvement in training effect is judged, and the control parameters of the rehabilitation robot are dynamically adjusted accordingly. At the same time, the feedback data is uploaded to the cloud platform to form a closed-loop control mechanism. Through the cooperation of the above technical solutions, the training pathway can continuously evolve with the patient's recovery state, achieving long-term and stable functional recovery optimization. This constructs an intelligent rehabilitation training method with data-driven as the core, rehabilitation robot execution as the carrier, and continuous optimization by the cloud platform as the support, effectively improving the safety, individual adaptability, and long-term recovery effect of rehabilitation training. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a rehabilitation training method that integrates a rehabilitation robot and a cloud platform, as shown in the embodiment. Figure 2 This is a schematic diagram illustrating the correlation analysis between trajectory offset temporal lag and nerve conduction velocity in the embodiment; Figure 3 This is a structural diagram of a rehabilitation training system that integrates a rehabilitation robot and a cloud platform, as shown in the embodiment. Detailed Implementation

[0019] This invention provides a rehabilitation training method and system integrating a rehabilitation robot and a cloud platform. The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" or "having" and any variations thereof are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0020] For ease of understanding, the specific process of the embodiments of the present invention will be described below, such as... Figure 1 As shown, an embodiment of the present invention provides a rehabilitation training method integrating a rehabilitation robot and a cloud platform, comprising: Step S1: Collect joint motion trajectory data of the patient during functional movements using sensors to generate raw three-dimensional trajectory information; preprocess the raw three-dimensional trajectory information to obtain a standardized motion trajectory dataset; In step S1, generating the original three-dimensional trajectory information includes: Multi-angle sensors are used to collect data on the patient's shoulder, elbow, and wrist joints during specific functional movements, recording the motion parameters of each joint; trajectory data containing three-dimensional spatial position information is generated based on the motion parameters; the trajectory data is initially integrated to form the original three-dimensional trajectory information; The original three-dimensional trajectory information is checked for integrity through a data verification mechanism to ensure that the data covers the range of motion of all target joints.

[0021] Specifically, this invention is applicable to upper limb function recovery training. By collecting, preprocessing, and analyzing the motion trajectory data of joints such as the shoulder, elbow, and wrist, it provides data support for the generation and adjustment of personalized training programs, ensuring the efficiency and accuracy of the training process.

[0022] In this embodiment, multi-angle sensors are used to collect motion data of the shoulder, elbow, and wrist joints when the patient performs functional movements, such as reaching for an object. These sensors are positioned around the patient's shoulder, elbow, and wrist to ensure coverage of the entire range of motion of the functional movement. The sensors record real-time joint position and angle data, generating raw trajectory data containing three-dimensional spatial position information. The raw trajectory data is presented as a point sequence in a Cartesian coordinate system, where the coordinates of each point reflect the position and motion changes of the joint in space. To ensure the accuracy and reliability of the data, the raw trajectory data is preprocessed, including preliminary integration, noise reduction, interpolation, and normalization.

[0023] The data preprocessing process begins by merging the original data points to form a continuous trajectory curve, ensuring the consistency and continuity of the trajectory. Next, a data verification mechanism checks the integrity of the trajectory information to ensure that the range of motion of all target joints is covered. This is achieved by verifying the number of sampling points for each joint; if the number of sampling points does not reach a preset threshold (e.g., 80% from the start to the end of the movement), the joint is marked as incomplete, and missing trajectory points are filled using interpolation. The interpolation employs a linear averaging method based on adjacent points to ensure a smooth transition of the trajectory and data consistency. If data from a missing joint is detected, a re-acquisition of the signal is triggered to ensure data coverage of the shoulder, joint, and other joints. The complete trajectories of the three target joints—elbow and wrist—are verified. Based on this, the integrity of the range of motion is confirmed, and the boundary values ​​of the trajectory are assessed to ensure they meet the benchmarks of natural movement. For example, the shoulder joint rotation angle should be at least 90 degrees. If the actual trajectory fails to meet this benchmark, the reasons are analyzed, such as sensor position deviation, and automatic correction is performed. By analyzing the trends of adjacent trajectories, the reasonable positions of the trajectory endpoints are inferred, ensuring the accuracy and recoverability of the data. These technical solutions lay the foundation for the rehabilitation robot to adjust its training program in real time based on accurate motion trajectory data, thereby improving the personalized adaptation during patient training and effectively promoting functional recovery.

[0024] Further, in step S1, a standardized motion trajectory dataset is obtained, including: For discontinuities in the original 3D trajectory information, interpolation techniques are used to complete the data; a smoothing algorithm is used to optimize the curve of the original 3D trajectory information to eliminate jitter interference during the acquisition process; the time axis and spatial axis of the original 3D trajectory information are normalized according to preset standardization rules; and the weight parameters of the original 3D trajectory information are adjusted according to the motion characteristics of different joints to generate a standardized motion trajectory dataset.

[0025] Specifically, to obtain high-quality input data that can support subsequent multi-joint temporal analysis, coordination deviation identification, and personalized training protocol generation, the raw three-dimensional trajectory information acquired by sensors is preprocessed to form a standardized motion trajectory dataset. This standardized motion trajectory dataset serves as the foundational data object for the collaborative analysis between the local control of the rehabilitation robot and the cloud platform, playing a crucial role in the entire technical solution.

[0026] During implementation, due to factors such as short-term sensor occlusion, communication jitter, or discontinuous patient movements during actual acquisition, the original 3D trajectory information may exhibit discontinuities. This embodiment addresses this by performing interpolation to complete the original 3D trajectory information. Specifically, it automatically identifies the location of discontinuities in the time series dimension and locates adjacent valid sampled data points before and after these discontinuities. By calculating the coordinate difference between adjacent data points in 3D space, corresponding compensation coordinate values ​​are generated based on a linear interpolation formula and inserted into the time position of the discontinuity point. This restores the continuity of the trajectory along the time axis, thus reconstructing the originally missing or broken 3D trajectory into a continuous trajectory sequence. This allows the same joint to form a complete data representation throughout the entire functional movement cycle. After completing the trajectory connection... Following continuous repair, the interpolated and completed 3D trajectory information is further smoothed to eliminate unavoidable random jitter interference during the acquisition process. Specifically, a Gaussian filtering algorithm based on normal distribution is introduced into the complete trajectory sequence. Each sampling point in the trajectory is assigned a different weight within its time neighborhood, with the weight of nearby sampling points being higher than that of distant sampling points. This suppresses high-frequency noise through weighted averaging. The standard deviation parameter of the Gaussian filter is set according to the rhythmic characteristics of the rehabilitation training movements, thus achieving a balance between smoothing effect and preservation of trajectory details. After this smoothing process, the 3D trajectory exhibits a smoother curve shape in space, with a significant reduction in jitter amplitude, allowing the trajectory changes to truly reflect the actual movement trend of the patient's joints, rather than sensor noise.

[0027] After interpolation completion and smoothing optimization, the time and spatial axes of the trajectory data are uniformly normalized according to preset standardization rules. Specifically, in the time dimension, the start and end times of the current functional movement trajectory are obtained, and a minimum-maximum normalization method is used to linearly map the original timestamps to a standard interval of 0 to 1, thereby eliminating the influence of differences in movement duration between different patients and different training rounds on the analysis results. In the spatial dimension, the same normalization processing is performed on the x, y, and z components of the three-dimensional coordinate axes, compressing the range of motion of each joint in space to a uniform scale, so that the trajectory data of different individuals and different joints are uniform. The data is numerically comparable. Through this dual temporal and spatial normalization process, the original trajectory data is transformed from a physical quantity scale into a standardized feature space representation, facilitating subsequent temporal feature extraction and cross-sample analysis. Furthermore, considering the significant differences in the biomechanical roles of different joints in functional movements, a joint weight adjustment mechanism is introduced to perform weighted fusion processing on the normalized trajectory information. Specifically, the motion characteristic indicators of each joint are first quantified, such as by calculating the proportion of angular displacement range, average angular velocity, or amplitude of motion of each joint in the normalized trajectory, to assess its contribution to the overall movement. Subsequently, based on the above quantification results, different joints are assigned... Corresponding weight parameters, for example in upper limb rehabilitation training scenarios, assign higher weights to the shoulder joint, which has a large degree of rotational freedom and is responsible for the main movement, and relatively lower weights to the elbow and wrist joints, which bear the transmission and auxiliary functions. Simultaneously, in individualized scenarios with specific injury histories or functional limitations, the weight parameters can be dynamically adjusted according to the patient's condition. By multiplying the joint weights by the corresponding joint's normalized trajectory data, a standardized data representation that balances joint specificity and overall synergy is generated, allowing the temporal changes of key joints to be fully reflected in subsequent coordination analysis. Finally, after interpolation completion, smoothing, normalization, and weight adjustment, the multi-... Joint trajectory data is integrated and stored as a standardized motion trajectory dataset. This dataset is organized according to joint dimensions and time series structure, and stored in a local or cloud-based time series database through data serialization, thereby realizing data synchronization and sharing between the rehabilitation robot control module and the cloud platform analysis module. Through the above technical solution, the standardized motion trajectory dataset can not only be directly used to extract joint time series features and analyze coordination deviations, but also serve as a unified data foundation for subsequent compensatory behavior recognition, multi-joint collaborative path simulation, and training effect feedback evaluation, achieving closed-loop connection and consistent support of data processing logic in the overall technical solution.

[0028] Step S2: Extract joint temporal features based on the standardized motion trajectory dataset; analyze coordination deviation and determine temporal coordination deviation distribution based on the joint temporal features; if the temporal coordination deviation distribution exceeds a preset range, separate abnormal offset trajectories and determine compensatory behavior patterns. In step S2, determining the timing coordination deviation distribution includes: Based on the standardized motion trajectory dataset, joint motion data for different training stages are divided; for each training stage, feature parameters of each joint in the time dimension are extracted; by using the delay parameter adjustment technique, the temporal differences between joints are calculated; the coordination of joint motion is analyzed based on the temporal differences to identify potential deviation patterns; and the deviation patterns are integrated to generate the temporal coordination deviation distribution.

[0029] Specifically, in order to achieve a refined assessment of the patient's multi-joint motion coordination status and provide a quantitative basis for subsequent judgment of compensatory behavior and generation of personalized training protocols, joint temporal features are extracted and coordination analysis is performed on the above-mentioned standardized motion trajectory dataset to determine the temporal coordination deviation distribution that reflects the patient's motor control ability. This process is implemented in the application scenario of integrating rehabilitation robots and cloud platforms. The analysis results can be used for local real-time control and can also be used for long-term trend modeling and cross-stage comparison through the cloud.

[0030] In the specific implementation process, the joint movement data of patients during rehabilitation training is first divided into training stages based on the time-series information contained in the standardized motion trajectory dataset. This stage division is not a simple time truncation, but rather combines the repetition cycle of functional movements to divide the complete training process into multiple continuous intervals such as the initial stage, reinforcement stage, and stabilization stage. Each stage corresponds to the patient's adaptation period, reinforcement practice period, and relatively stable execution period, respectively. For example, in upper limb rehabilitation training, the initial stage can cover the first few repetitions of functional movements to reflect the patient's true coordination state before fully adapting to the training rhythm, while subsequent stages are used to observe the trend of coordination changes as training progresses. After completing the training stage division... For each training phase, characteristic parameters of each joint in the time dimension are extracted from the standardized motion trajectory data. These characteristic parameters are not single indicators, but are selected specifically according to the functional differences of different joints in biomechanics. For example, for the shoulder joint, which plays a dominant driving role, parameters reflecting initiation and driving ability, such as peak velocity and peak acceleration, are extracted; for the elbow joint, parameters reflecting transmission and coordination ability, such as motion duration and displacement amplitude, are extracted; and for the wrist joint, parameters related to fine motor functions, such as time delay and amplitude variation, are extracted. All of these characteristic parameters are associated with corresponding timestamps to form a temporal feature sequence arranged in chronological order, so that the motion performance of each joint in the same training phase can be aligned and analyzed on a unified time axis.

[0031] Building upon this foundation, a start-up delay parameter adjustment technique is further introduced to quantitatively model the temporal relationships between multiple joints. Specifically, the start-up time point of each joint is first identified from the aforementioned temporal feature sequence, and the original start-up delay value between joints is calculated accordingly. For example, the start-up time difference between the shoulder and elbow joints, and between the elbow and wrist joints, is calculated. Subsequently, the original delay value is calibrated by combining the patient's nerve conduction velocity parameters. The nerve conduction velocity can be estimated from clinical experience or historical data and used to deduce the reasonable delay range between joints under ideal conditions. When the original delay value exceeds this reasonable range, the delay parameter is normalized and calibrated using a preset adjustment formula, thereby generating an adjusted temporal difference sequence that reflects the patient's true nerve control ability. This allows the differences in joint start-up among different patients, at different ages, or with different degrees of injury to be mapped to a unified analytical scale, avoiding misjudgments caused by individual physiological differences.

[0032] Subsequently, based on the adjusted temporal difference sequence, a comprehensive analysis of joint coordination is performed. Specifically, this is done by comparing the temporal differences with a pre-constructed baseline coordination template. This template, derived from statistical analysis of joint temporal data from healthy individuals performing the same functional movements, represents typical patterns of natural coordinated movement. When the actual temporal difference falls within the allowable range of the baseline coordination template, the joint is considered coordinated. When the temporal difference exceeds this range, the corresponding deviation pattern is identified based on the direction and amplitude of the deviation, such as a delayed initiation deviation pattern or an premature initiation deviation pattern. For complex movement scenarios involving multiple joints, the deviation patterns between multiple joint pairs are combined and analyzed to identify complex coordination deviations, avoiding information omissions caused by single-joint analysis. After completing the deviation pattern identification within a single training phase, the deviation patterns identified in each training phase are integrated and statistically analyzed to generate... The temporal coordination deviation distribution, reflecting the type and frequency of deviations, not only describes the proportion of different deviation patterns in each training stage but also reflects the dynamic trend of coordination changes with the training process. For example, it can be used to determine whether delay deviations gradually decrease with training or whether new abnormal patterns appear in a specific stage. As a structured analysis result, the temporal coordination deviation distribution can be used locally by the rehabilitation robot to trigger anomaly judgment and adjust control parameters. On the other hand, it can also be stored and analyzed long-term through a cloud platform, providing a basis for subsequent personalized training strategy optimization and group data modeling. The above technical solution realizes the systematic transformation from standardized motion trajectory data to temporal coordination deviation distribution, transforming joint motion coordination from an abstract concept into quantifiable, comparable, and traceable data results. This provides key technical support for the closed-loop control of compensatory behavior recognition and personalized rehabilitation training in the overall technical solution.

[0033] Further, in step S2, if the timing coordination deviation distribution exceeds a preset range, the abnormal offset trajectory is separated and the compensatory behavior pattern is determined, including: By comparing the temporal coordination deviation distribution with the preset natural motion reference range, the joints with deviations exceeding the limit are identified; abnormal offset trajectory data of non-target joints are separated from the original three-dimensional trajectory information; and the causes of the abnormal offset trajectory data are analyzed in conjunction with the neural conduction velocity adaptation mechanism. Based on the analysis results, determine whether there is an abnormal pattern of compensatory behavior; generate feature labels corresponding to the abnormal pattern.

[0034] Specifically, based on the obtained temporal coordination deviation distribution, in order to further identify whether patients have abnormal compensatory behaviors caused by functional limitations during rehabilitation training, and to provide a clear basis for the dynamic adjustment of subsequent personalized training protocols, the temporal coordination deviation distribution and the original three-dimensional trajectory information are jointly analyzed, thereby realizing the separation of abnormal deviation trajectories and the determination of compensatory behavior patterns.

[0035] By comparing the aforementioned temporal coordination deviation distribution with a preset natural motion benchmark range, joints with excessive deviations are identified. This natural motion benchmark range is constructed based on the temporal coordination statistics of healthy individuals performing the same functional movements, representing the permissible time difference interval for normal joint synergistic movement. During the comparison, the deviation values ​​of each joint at different training stages are analyzed one by one. When the temporal deviation corresponding to a certain joint continuously or significantly exceeds the benchmark upper limit, that joint is marked as a deviation-exceeding joint, thus identifying the key target for subsequent abnormal trajectory analysis. After identifying the deviation-exceeding joints, the original three-dimensional trajectory information is reviewed, and the non-target joint trajectories associated with these deviation-exceeding joints are separated. Specifically, based on the joint identification information contained in the trajectory data, the original three-dimensional trajectory is first divided at the joint level. Then, within the trajectory subset corresponding to the non-target joints, statistical indicators such as the average displacement, displacement change rate, or angle change amplitude of the trajectory are used to screen trajectory segments with significantly higher deviation levels than the overall average. This allows for the extraction of suspected abnormal non-target joint deviation trajectory data from the complete multi-joint motion trajectory, enabling subsequent analysis to focus on joint movement behaviors that may play a compensatory role.

[0036] After obtaining abnormal trajectory data, the causes of the abnormal trajectory are analyzed using a neural conduction velocity adaptation mechanism to distinguish between true functional impairment caused by decreased neural control and trajectory abnormalities caused by active or passive compensatory behaviors. Specifically, based on the patient's individual neurophysiological parameters or historical training data, the neural conduction velocity of the corresponding joint is calculated. The neural conduction velocity is estimated by the propagation time of nerve impulses over a known anatomical distance and used as a reference threshold to judge the rationality of joint initiation and response. Subsequently, the time-series characteristics of the abnormal trajectory are matched with the aforementioned neural conduction velocity threshold. When the temporal lag shown in the trajectory significantly exceeds the expected range allowed by the neural conduction velocity, the abnormal deviation is more likely due to slowed neural conduction or insufficient control. Conversely, if the neural conduction velocity is within the normal or near-normal range, but the trajectory deviation is still large, further analysis of characteristics such as displacement peak value and changes in motion amplitude is used to determine that the abnormality is more likely caused by external compensatory behaviors, such as... Figure 2 As shown, for example, in shoulder joint analysis, if the propagation time is 0.02 seconds and the distance is 0.5 meters, the speed is 25 meters per second. Assuming the patient performs an arm-raising motion, the abnormal deviation trajectory shows a 0.1-second delay in elbow joint displacement. The speed calculated through the nerve conduction velocity adaptation mechanism is 20 meters per second, which is lower than the normal threshold of 30 meters per second. Therefore, the cause is analyzed as a deviation due to slowed nerve conduction. By introducing this nerve conduction velocity adaptation mechanism, the analysis of the cause of abnormal trajectories is upgraded from a simple judgment of trajectory morphology to a comprehensive judgment combining physiological mechanisms, thereby significantly improving the accuracy of the judgment.

[0037] Based on the causal analysis, the motion patterns reflected by the abnormal offset trajectories are further represented in a patterned manner, and the existence of compensatory behavior abnormal patterns is determined accordingly. Specifically, the above causal analysis results are transformed into structured deviation pattern vectors, where the deviation pattern vectors comprehensively represent information such as abnormal joint type, offset direction, offset amplitude, and temporal characteristics. Subsequently, the above deviation pattern vectors are matched and analyzed with a pre-constructed compensatory behavior pattern library, which contains a variety of typical non-target joint compensation for insufficient target joint function, such as the pattern of the elbow joint or trunk generating additional movement to complete the task when the shoulder joint function is weakened. By calculating the similarity between the deviation pattern vector and each pattern in the pattern library, when the matching degree exceeds the preset matching degree, it is determined that the current abnormal offset trajectory corresponds to an abnormal pattern of compensatory behavior. At the same time, in multi-joint collaborative scenarios, the occurrence frequency and evolution of compensatory patterns are comprehensively judged by combining the temporal trends of multiple training stages, thereby identifying newly emerging or gradually aggravated compensatory behaviors.

[0038] After identifying abnormal patterns of compensatory behavior, corresponding feature labels are generated for these abnormal patterns and stored along with relevant trajectory data. These feature labels are used to semantically identify the type and degree of compensatory behavior, such as "high elbow joint compensation intensity" or "increasing shoulder compensation trend," so that they can be used as direct input parameters in the subsequent personalized training protocol generation process to adjust training content, training rhythm, and control parameters. Through the above technical solution, a complete logical link is realized, starting from the distribution of temporal coordination deviations, through abnormal trajectory separation and cause analysis, to the identification of compensatory patterns and label generation. This enables the rehabilitation robot and cloud platform to optimize training strategies in a targeted manner based on the clear behavioral mechanism identification results, thereby effectively avoiding the solidification of compensatory behavior and improving the safety and effectiveness of rehabilitation training.

[0039] Step S3: Based on the compensatory behavior pattern and individualized patient data, generate a personalized training protocol; simulate multi-joint collaborative pathways and perform collaborative parameter corrections based on the personalized training protocol to determine the optimized motion guidance path; In step S3, a personalized training protocol scheme is generated, including: Based on the judgment results of the compensatory behavior pattern, key limiting factors of the patient's motor function are extracted; combined with the patient's individualized data, the patient's exercise habits and physical condition are analyzed; through staggered interval dynamic configuration technology, the functional movements are decomposed and the training rhythm is adjusted; based on the decomposed movement units, targeted training modules are designed; and the training modules and individualized data are integrated to generate the personalized training protocol scheme.

[0040] Specifically, based on the identification of compensatory behavior patterns and the acquisition of corresponding abnormal pattern feature labels, in order to effectively transform the above analysis results into an executable, adjustable rehabilitation training program that conforms to the patient's actual ability, a personalized training protocol program that is highly matched with the patient's individual state is generated. The compensatory behavior pattern is combined with the patient's individual data, and the personalized training protocol program is generated through the structured decomposition of functional movements and the dynamic configuration of training rhythm.

[0041] Specifically, based on the assessment results of the aforementioned compensatory behavior patterns, key limiting factors for the patient's motor function are extracted. This extraction process is not simply listing abnormal joints, but rather analyzing the root causes of compensation from the compensatory behavior patterns. Specifically, by identifying the joint parameters corresponding to abnormal deviation trajectories, it clarifies which joints or joint combinations exhibit abnormal rotation, hyperextension, or abnormal involvement in functional movements. For example, when abnormal rotation of the shoulder joint is detected in a functional movement, or hyperextension of the elbow joint occurs during the synergistic phase, these joint parameters are compared with preset motor function benchmarks to quantify the impact of coordination deviations on muscle load, joint pressure, and movement efficiency. The resulting key limiting factors not only indicate the specific locations of motor function limitations but also reflect the degree of limitation and its impact on overall movement completion, providing a clear direction for subsequent training interventions. After identifying the key limiting factors, a comprehensive analysis of the patient's individualized data, including their exercise habits and physical condition, is conducted. The individualized data mentioned above includes, but is not limited to, the patient's age, weight, medical history, daily activity level, and long-term motor preferences. Furthermore, by comparing this individualized data with standard physiological templates, the study assesses whether the patient exhibits a preference for using their dominant limb, insufficient muscle strength, decreased endurance, or limited recovery ability, and analyzes how these individual characteristics interact with identified compensatory behaviors. The standard physiological templates are reference templates constructed based on healthy individuals or recognized physiological principles to characterize normal human motor abilities and physiological characteristics, used for comparative analysis and evaluation of individual patient motor states. For example, for patients who long rely on their dominant limb to complete tasks, the study identifies their greater tendency to compensate during training on their non-dominant limb, and subsequently reduces the initial training intensity or extends the recovery interval in the training protocol. In this way, the results of the compensatory pattern analysis are integrated with individualized physiological and behavioral characteristics into the same decision-making framework, avoiding training programs that target only a single abnormality while ignoring the overall physical condition.

[0042] After completing the above analysis, an interleaved interval dynamic configuration technique is introduced to structurally decompose functional movements and adjust the training rhythm. Specifically, the complete functional movement is first broken down into multiple sub-movement units with clear start and end points. For example, the action of reaching for an object is decomposed into continuous units such as shoulder elevation, elbow extension, and wrist rotation. Subsequently, based on the above temporal feature extraction results and coordination deviation distribution, the execution time and completion quality of each sub-movement unit are monitored, and the interval adjustment coefficient between the movement units is calculated accordingly. The adjustment coefficient is calculated from the difference between the actual completion time and the expected completion time, and is scaled in combination with a configuration factor. The configuration factor is dynamically selected within a preset range according to the patient's overall coordination deviation level to ensure that the rhythm adjustment can both alleviate joint burden and avoid excessively reducing training efficiency. By applying the calculated adjustment coefficient to the starting interval time between adjacent movement units, the rhythm of movement connection can be dynamically lengthened or shortened, avoiding the induction of new compensatory behaviors due to continuous movement stacking, and gradually guiding the patient to form a more coordinated and stable movement pattern.

[0043] After completing the movement decomposition and rhythm configuration, targeted training modules are designed based on the functional goals of each sub-movement unit. Each training module revolves around a specific movement unit and incorporates the aforementioned key limiting factors to set training content, repetition counts, and intensity progression strategies. For example, for the shoulder elevation unit, the training module focuses on repetitive arm elevation, gradually increasing the range of motion or resistance level according to the degree of shoulder joint abnormality. For the elbow extension unit, its synergistic performance in the overall movement can be improved by controlling the extension speed and duration. All of these training modules are designed directly with compensatory behavior patterns, ensuring that the training objectives are clearly directed at the underlying causes of compensation, rather than merely pursuing movement completion. Finally, these training modules are integrated with individualized patient data to generate a complete personalized training program. The training protocol not only includes specific combinations of training movements, but also further clarifies the daily training duration, frequency, and intensity distribution. The training time is adapted to the patient's age, fitness level, and lifestyle. For example, for patients with weak physical recovery abilities or chronic injuries, low-intensity, longer-interval training programs are prioritized. The training effect is regularly evaluated using the cloud platform's feedback mechanism, and module parameters are gradually adjusted. This technical solution continuously incorporates coordination deviation data and compensatory behavior tags into the protocol update process, enabling the personalized training protocol to iteratively optimize during the rehabilitation robot's execution. This achieves data-driven personalized rehabilitation intervention within the overall technical solution, improving training safety, compliance, and long-term recovery outcomes.

[0044] Further, in step S3, the optimized motion guidance path is determined, including: Based on the personalized training protocol, a coordinated motion model of the shoulder, elbow, and wrist joints is constructed; the parameters of the coordinated motion model are adjusted using motion path simulation optimization technology; the motion trajectory parameters of each joint are corrected to address deviations during the simulation process; cross-device data synchronization is achieved through a time-series data cloud storage mechanism; and an optimized motion guidance path is generated based on the corrected parameters.

[0045] Specifically, to further transform the aforementioned training protocol into specific motion commands that can be directly executed by rehabilitation robots and dynamically adapted to the patient's functional state, an optimized motion guidance path is determined through multi-joint collaborative modeling, path simulation optimization, and parameter correction. Specifically, a collaborative motion model of the shoulder, elbow, and wrist joints is first constructed based on the aforementioned personalized training protocol. This collaborative motion model is not simply a superposition of the independent trajectories of each joint, but rather, based on the temporal characteristics, coordination deviation distribution, and joint weight parameters already defined in the personalized training protocol, it abstracts the multi-joint motion relationships into a unified three-dimensional mathematical model. During the modeling process, the relative positions of the shoulder, elbow, and wrist joints in functional movements are extracted from the personalized training protocol. By considering the initiation sequence, time delay, and spatial motion characteristics, and combining the historical motion patterns reflected in the standardized motion trajectory dataset, the relative position vectors and velocity vectors between each joint are calculated to establish a collaborative motion representation describing the linkage relationship of multiple joints. Among them, the motion vector of the shoulder joint is set as the reference axis of the overall collaborative model, while the motion vectors of the elbow and wrist joints are weighted and adjusted according to their performance in the coordination deviation distribution, so that the model can truly reflect the patient's current functional state and potential compensatory characteristics. For patients with changes in nerve conduction ability, a nerve conduction velocity adaptation parameter is introduced during the modeling process, making the collaborative model more sensitive to initiation delay in the time dimension, thereby improving the targeting of subsequent path optimization.

[0046] After constructing the coordinated movement model, key parameters in the model are iteratively adjusted using motion path simulation optimization technology. This ensures that the simulated multi-joint coordinated path closely approximates the natural movement baseline and matches the patient's current ability level. Specifically, the coordinated movement model and standardized movement trajectory dataset are input into the path simulation optimization module, and the model parameters are searched and optimized using a path planning method based on iterative algorithms. In one implementation, the system employs a particle swarm optimization algorithm to jointly optimize parameters such as joint angles, joint velocities, and initiation delays. Each particle represents a possible combination of parameters. A fitness function is constructed by calculating the degree of matching between the simulated path and the standardized trajectory, and the position and velocity of the particles are continuously updated during multiple iterations until the fitness function converges. Through this simulation optimization process, the deviation between the multi-joint coordinated path and the natural movement pattern can be gradually reduced while ensuring the safety of the movement. This ensures that the generated path provides the challenge required for rehabilitation training without inducing new compensatory behaviors.

[0047] After the path simulation optimization is completed, the motion trajectory parameters of each joint are finely corrected to address the local deviations that still exist in the simulation. This correction process uses the natural motion reference range as a reference, analyzes the trajectory positions in the simulated path that exceed the reference range point by point, and calculates the corresponding deviation vector. Furthermore, it incorporates a neural conduction velocity adaptation mechanism to dynamically determine compensation factors for different joints, and superimposes the compensation vectors onto the original trajectory parameters. This achieves targeted correction of abnormal offsets, effectively suppressing the over-involvement or abnormal offset that may occur in non-target joints in the simulated path, while preserving the dominant motion characteristics of the target joints. This ensures that the corrected collaborative path maintains spatial continuity and temporal consistency. In terms of consistency, all aspects meet the requirements of rehabilitation training. For patients requiring long-term, gradual recovery, the magnitude of the compensation factor is appropriately reduced to ensure that the corrected path supports a gradual training pace. After parameter correction is completed, the corrected coordinated movement parameters and path data are uploaded to the cloud platform through a time-series data cloud storage mechanism, enabling data synchronization between the rehabilitation robot, mobile terminal, and remote monitoring system. Through the above cloud synchronization mechanism, different devices can share the latest movement guidance path when executing the same training protocol, avoiding deviations in training effects due to data inconsistencies between devices. It also provides doctors or rehabilitation management systems with the technical conditions for remote viewing and intervention.

[0048] Finally, based on the simulated, optimized, and parameter-corrected collaborative motion model, an optimized motion guidance path corresponding to the current training stage is generated. This motion guidance path explicitly specifies the target angle, speed, and initiation sequence of each joint during the training process in the form of a time series, and embeds safety thresholds and rhythm constraints, enabling the rehabilitation robot to guide the patient to complete functional movements that conform to the laws of neural recovery in real time during execution. The above technical solution, by combining the above motion guidance path with a real-time feedback mechanism, can continuously evaluate the patient's execution status during training and update the path parameters through a cloud platform when necessary. Thus, the overall technical solution forms a rehabilitation training execution system with personalized training protocols as the core, collaborative path optimization as the means, and closed-loop control as the goal, achieving continuous optimization of the functional recovery path.

[0049] Step S4: Extract feedback indicators based on the optimized motion guidance path, determine the training effect, update the control parameters of the rehabilitation robot, and generate a continuously optimized functional recovery path; specifically including: Based on the optimized exercise guidance path, muscle function feedback indicators are extracted; the changing trends of the feedback indicators are analyzed through a real-time feedback loop update mechanism; if the feedback indicators meet the preset neural recovery threshold assessment criteria, the degree of improvement in the training effect is determined; based on the determination of the training effect, the parameters of the rehabilitation robot control module are dynamically adjusted; a closed-loop regulation is formed through the feedback data stored in the cloud to generate a continuously optimized functional recovery path.

[0050] Specifically, to achieve objective evaluation of rehabilitation training effects and dynamically adjust the control parameters of the rehabilitation robot accordingly, thereby forming a continuously optimized functional recovery path, a feedback indicator extraction, trend analysis, and closed-loop control mechanism based on motion guidance paths is further introduced within the technical framework of integrating rehabilitation robots and cloud platforms. Specifically, during training execution, the patient's multi-joint movement performance during actual training is collected and analyzed in real time based on the optimized motion guidance path, extracting feedback indicators that reflect muscle function status. The motion guidance path itself is generated by the aforementioned collaborative motion model, path simulation optimization, and parameter correction process, clearly defining the collaborative trajectory of the shoulder, elbow, and wrist joints in the time dimension. Therefore, the system can... The collaborative trajectory data of each joint during the training process is directly extracted from this path. The collaborative trajectory data is consistent in data structure with the standardized motion trajectory data previously collected by multi-angle sensors and processed by interpolation, smoothing and normalization. This ensures that the data source and historical analysis results are traceable during the extraction of feedback indicators. The collaborative trajectory data is processed by a preset feature extraction algorithm to calculate indicators such as peak velocity and acceleration in the trajectory of each joint. Peak velocity is used to characterize the instantaneous force exertion ability of the corresponding muscle group, and acceleration is used to reflect the response speed and coordination of the joint in the collaborative movement. This transforms the abstract training process into quantifiable muscle function feedback indicators, laying a data foundation for subsequent effect judgment.

[0051] After obtaining the aforementioned muscle function feedback indicators, a real-time feedback loop update mechanism is used to continuously analyze the changing trends of these indicators during training. Specifically, continuous feedback indicator data points are periodically selected at preset time intervals and organized into an indicator sequence in chronological order. Subsequently, a moving average filter is applied to the indicator sequence to eliminate the interference of instantaneous fluctuations on trend judgment. Based on this, the slope of the indicator changes over time is calculated to determine whether the muscle function state is improving, stagnating, or declining. This real-time feedback loop update mechanism allows the system to continuously introduce new indicator data during training and integrate it with existing trend information, thereby forming a cumulative change curve reflecting short-term changes and long-term evolution. When the trend analysis results show that the feedback indicators are continuously rising, they are marked as positive changes. Conversely, if there are obvious fluctuations or a downward trend, the corresponding adjustment logic is triggered to prevent compensatory behavior from intensifying or the training load from becoming inappropriate.

[0052] Based on trend analysis, the aforementioned feedback indicators are compared with preset neural recovery threshold assessment criteria to determine the degree of improvement in training effectiveness. These criteria are set based on clinical statistical data and consider not only the absolute change in a single indicator but also a persistence condition to distinguish between short-term fluctuations and true functional recovery. Specifically, when the trend of change in the feedback indicator exceeds the preset threshold and continues to meet at least a certain number of training cycles, the patient's functional state is determined to have entered a new recovery stage. The corresponding improvement score is quantified by calculating the relative change ratio between the current indicator and the initial indicator. Subsequently, the training effect is graded based on the improvement score, and the judgment results are appropriately corrected by combining the patient's individual data, such as age and initial coordination deviation level, to avoid misjudgment due to individual differences. This ensures that the training effect assessment results are both quantitatively based and consistent with the patient's actual recovery rhythm.

[0053] After completing the phased assessment of training effectiveness, the parameters of the rehabilitation robot control module are dynamically adjusted based on the assessment results to achieve adaptive optimization of training intensity and difficulty. Specifically, corresponding adjustment factors are extracted from the improvement scores and mapped to key parameters in the rehabilitation robot control module, such as movement speed, joint resistance, or range of motion. Parameter adjustment follows a gradual principle: the training challenge is moderately increased when functional improvement is significant, and the load is reduced when recovery is slow or fluctuating, ensuring that the training process remains within a safe and effective range. To verify the rationality of the adjusted parameter combination, the adjusted movement path is rapidly simulated using a neural conduction velocity adaptation mechanism to confirm that it still meets the coordination and safety requirements in both time and space dimensions. Subsequently, the verified parameters are sent to the rehabilitation robot control module in real time, allowing the next round of training to be executed directly based on the updated parameters. After the above parameter adjustment is completed, through... The feedback data stored in the cloud forms a closed-loop control mechanism to support the continuous optimization of the functional recovery path. Specifically, the feedback indicators, trend analysis results, phased judgment conclusions, and corresponding control parameter adjustment information generated in the latest round of training are uniformly uploaded to the cloud platform and integrated with historical training data. Based on this, the cloud platform compares and analyzes the data from different training stages. If a consistent trend in functional improvement is found, the current training path is strengthened; if fluctuations or abnormal trends are found, a basis is provided for subsequent path adjustments. Through the cloud-based closed-loop control mechanism, the optimized movement guidance path is no longer a static result, but a dynamic path that continuously evolves with changes in the patient's functional state. This enables data-driven long-term rehabilitation management within the overall technical solution, ensuring that the rehabilitation robot adaptively adjusts itself around the patient's actual recovery process throughout the entire training cycle, ultimately generating a continuously optimized functional recovery path.

[0054] This invention also provides a rehabilitation training system integrating a rehabilitation robot and a cloud platform, used to implement the above-mentioned methods, such as... Figure 3 As shown, the system includes: The joint trajectory acquisition unit is used to acquire joint motion trajectory data of the patient during functional movements through sensors, generate raw three-dimensional trajectory information, and preprocess the raw three-dimensional trajectory information to obtain a standardized motion trajectory dataset. The timing deviation analysis unit is used to extract joint timing features based on the standardized motion trajectory dataset, analyze coordination deviations based on the joint timing features, and determine the timing coordination deviation distribution. The compensation pattern recognition unit is used to separate abnormal offset trajectories and determine compensation behavior patterns when the timing coordination deviation distribution exceeds a preset range. The training path generation unit is used to generate a personalized training protocol scheme based on the compensatory behavior pattern and individualized patient data; simulate multi-joint collaborative paths and perform collaborative parameter corrections based on the personalized training protocol scheme to determine the optimized motion guidance path; The feedback control and optimization unit is used to extract feedback indicators based on the optimized motion guidance path, judge the training effect, update the control parameters of the rehabilitation robot, and generate a continuously optimized functional recovery path.

[0055] The present invention also provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the above-described method.

[0056] In summary, this invention utilizes multi-angle sensors to continuously collect and preprocess patient joint motion trajectories, constructing a high-quality, standardized motion trajectory dataset. This accurately characterizes the collaborative relationships of multiple joints over time, avoiding the limitations of traditional methods that rely solely on single-joint or instantaneous indicators. By analyzing joint temporal features and comparing them with natural motion benchmarks, coordination deviations in different training stages are identified, and abnormal deviation trajectories are separated. The compensatory behavior pattern is determined by combining neural conduction velocity adaptation mechanisms, ensuring that anomaly identification is based not only on trajectory morphology but also on physiological mechanisms, thereby improving the accuracy of compensation judgment and avoiding misinterpreting normal individual differences as abnormalities. Personalized training protocols are generated based on compensatory behavior patterns and individualized patient data. Through multi-joint collaborative modeling and path simulation optimization, the training strategy is transformed into an executable dataset. The exercise guidance pathway effectively connects "problem identification" to "intervention design," ensuring that the training pathway matches the patient's current ability level in terms of time, space, and rhythm, reducing training risks and improving training efficiency. Finally, by collecting and analyzing the execution feedback of the optimized exercise guidance pathway in real time, the degree of improvement in training effect is judged, and the control parameters of the rehabilitation robot are dynamically adjusted accordingly. At the same time, the feedback data is uploaded to the cloud platform to form a closed-loop control mechanism. Through the cooperation of the above technical solutions, the training pathway can continuously evolve with the patient's recovery state, achieving long-term and stable functional recovery optimization. This constructs an intelligent rehabilitation training method with data-driven as the core, rehabilitation robot execution as the carrier, and continuous optimization by the cloud platform as the support, effectively improving the safety, individual adaptability, and long-term recovery effect of rehabilitation training.

[0057] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0058] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or 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 the present invention. 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.

[0059] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A rehabilitation training method integrating a rehabilitation robot and a cloud platform, characterized in that, The method includes: Step S1: Collect joint motion trajectory data of the patient during functional movements using sensors to generate raw three-dimensional trajectory information; preprocess the raw three-dimensional trajectory information to obtain a standardized motion trajectory dataset; Step S2: Extract joint temporal features based on the standardized motion trajectory dataset; analyze coordination deviation and determine temporal coordination deviation distribution based on the joint temporal features; if the temporal coordination deviation distribution exceeds a preset range, separate abnormal offset trajectories and determine compensatory behavior patterns. Step S3: Based on the compensatory behavior pattern and individualized patient data, generate a personalized training protocol; simulate multi-joint collaborative pathways and perform collaborative parameter corrections based on the personalized training protocol to determine the optimized motion guidance path; Step S4: Extract feedback indicators based on the optimized motion guidance path, judge the training effect, update the control parameters of the rehabilitation robot, and generate a continuously optimized functional recovery path.

2. The method as described in claim 1, characterized in that, In step S1, the original three-dimensional trajectory information is generated, including: Multi-angle sensors are used to collect data on the patient's shoulder, elbow, and wrist joints during specific functional movements, recording the motion parameters of each joint; trajectory data containing three-dimensional spatial position information is generated based on the motion parameters; the trajectory data is initially integrated to form the original three-dimensional trajectory information; The original three-dimensional trajectory information is checked for integrity through a data verification mechanism to ensure that the data covers the range of motion of all target joints.

3. The method as described in claim 2, characterized in that, In step S1, a standardized motion trajectory dataset is obtained, including: For discontinuities in the original 3D trajectory information, interpolation techniques are used to complete the data; a smoothing algorithm is used to optimize the curve of the original 3D trajectory information to eliminate jitter interference during the acquisition process; the time axis and spatial axis of the original 3D trajectory information are normalized according to preset standardization rules; and the weight parameters of the original 3D trajectory information are adjusted according to the motion characteristics of different joints to generate a standardized motion trajectory dataset.

4. The method as described in claim 1, characterized in that, In step S2, determining the timing coordination deviation distribution includes: Based on the standardized motion trajectory dataset, joint motion data for different training stages are divided; for each training stage, feature parameters of each joint in the time dimension are extracted; by using the delay parameter adjustment technique, the temporal differences between joints are calculated; the coordination of joint motion is analyzed based on the temporal differences to identify potential deviation patterns; and the deviation patterns are integrated to generate the temporal coordination deviation distribution.

5. The method as described in claim 4, characterized in that, In step S2, if the timing coordination deviation distribution exceeds a preset range, the abnormal offset trajectory is separated and the compensatory behavior pattern is determined, including: By comparing the temporal coordination deviation distribution with the preset natural motion reference range, the joints with deviations exceeding the limit are identified; abnormal offset trajectory data of non-target joints are separated from the original three-dimensional trajectory information; and the causes of the abnormal offset trajectory data are analyzed in conjunction with the neural conduction velocity adaptation mechanism. Based on the analysis results, determine whether there is an abnormal pattern of compensatory behavior; generate feature labels corresponding to the abnormal pattern.

6. The method as described in claim 1, characterized in that, In step S3, a personalized training protocol scheme is generated, including: Based on the judgment results of the compensatory behavior pattern, key limiting factors of the patient's motor function are extracted; combined with the patient's individualized data, the patient's exercise habits and physical condition are analyzed; through staggered interval dynamic configuration technology, the functional movements are decomposed and the training rhythm is adjusted; based on the decomposed movement units, targeted training modules are designed; and the training modules and individualized data are integrated to generate the personalized training protocol scheme.

7. The method as described in claim 6, characterized in that, In step S3, the optimized motion guidance path is determined, including: Based on the personalized training protocol, a coordinated motion model of the shoulder, elbow, and wrist joints is constructed; the parameters of the coordinated motion model are adjusted using motion path simulation optimization technology; the motion trajectory parameters of each joint are corrected to address deviations during the simulation process; cross-device data synchronization is achieved through a time-series data cloud storage mechanism; and an optimized motion guidance path is generated based on the corrected parameters.

8. The method as described in claim 1, characterized in that, In step S4, a continuously optimized function recovery path is formed, including: Based on the optimized exercise guidance path, muscle function feedback indicators are extracted; the changing trends of the feedback indicators are analyzed through a real-time feedback loop update mechanism; if the feedback indicators meet the preset neural recovery threshold assessment criteria, the degree of improvement in the training effect is determined; based on the determination of the training effect, the parameters of the rehabilitation robot control module are dynamically adjusted; a closed-loop regulation is formed through the feedback data stored in the cloud to generate a continuously optimized functional recovery path.

9. A rehabilitation training system integrating a rehabilitation robot and a cloud platform, used to implement the method as described in any one of claims 1-8, characterized in that, The system includes: The joint trajectory acquisition unit is used to acquire joint motion trajectory data of the patient during functional movements through sensors, generate raw three-dimensional trajectory information, and preprocess the raw three-dimensional trajectory information to obtain a standardized motion trajectory dataset. The timing deviation analysis unit is used to extract joint timing features based on the standardized motion trajectory dataset, analyze coordination deviations based on the joint timing features, and determine the timing coordination deviation distribution. The compensation pattern recognition unit is used to separate abnormal offset trajectories and determine compensation behavior patterns when the timing coordination deviation distribution exceeds a preset range. The training path generation unit is used to generate a personalized training protocol scheme based on the compensatory behavior pattern and individualized patient data; simulate multi-joint collaborative paths and perform collaborative parameter corrections based on the personalized training protocol scheme to determine the optimized motion guidance path; The feedback control and optimization unit is used to extract feedback indicators based on the optimized motion guidance path, judge the training effect, update the control parameters of the rehabilitation robot, and generate a continuously optimized functional recovery path.

10. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement the method as described in any one of claims 1-8.

Images