Remote assessment and monitoring method and system for elderly rehabilitation therapy

Abstract

The remote evaluation and monitoring method and system based on the rehabilitation treatment of the elderly according to the application belong to the technical field of remote evaluation and monitoring of the rehabilitation treatment of the elderly, comprising S1, collecting kinematic data and physiological data of a patient; S2, establishing a physiological safety baseline model for the patient according to standard test data, and obtaining a preset standard motion template; S3, calculating a kinematic deviation index based on the kinematic data and the standard motion template, and calculating a physiological stress index based on the physiological data and the physiological safety baseline model; S4, using the physiological stress index to perform nonlinear correction on the kinematic deviation index to generate a risk-adjusted deviation; S5, generating an adaptive evaluation score based on the risk-adjusted deviation; S6, determining a system risk level according to the physiological stress index, and performing closed-loop intervention corresponding to the system risk level, wherein the adaptive evaluation score generated by the application is more comprehensive, objective and humanized.

Description

Technical Field

[0001] This invention relates to the field of remote assessment and monitoring of rehabilitation treatment for the elderly, specifically to a method and system for remote assessment and monitoring of rehabilitation treatment for the elderly. Background Technology

[0002] Currently, rehabilitation systems for the elderly are mainly constructed using a single-dimensional assessment method, focusing primarily on evaluating the patient's movement patterns during rehabilitation training. In certain special scenarios, such as remote rehabilitation, the lack of real-time monitoring of the patient's physiological state leads to incomplete assessment results and the inability to promptly identify potential risks.

[0003] In related technologies, with the development of home rehabilitation, wearable devices, and telemedicine, assessment systems based on single movement patterns present some problems or weaknesses. For example, when a patient is in poor physical condition, due to fatigue or pain, they may subconsciously change their movement patterns to avoid further injury, but existing systems may interpret this as improper movement and give a negative evaluation. This leads to assessment results that do not match the patient's actual rehabilitation performance, and may even exacerbate risks due to incorrect guidance. In addition, existing assessment systems also struggle to capture hidden risks that appear to be standard in movement patterns but whose underlying physiological load has exceeded limits, which urgently needs improvement.

[0004] The information disclosed in the background section above is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for remote assessment and monitoring of rehabilitation treatment for the elderly, so as to solve the problems mentioned in the background art.

[0006] The technical solution of the present invention includes the following steps:

[0007] S1. Collect the patient's kinematic and physiological data;

[0008] S2. Establish a physiological safety baseline model for the patient based on standard test data and obtain a preset standard exercise template;

[0009] S3. Calculate the kinematic deviation index based on kinematic data and standard exercise templates; calculate the physiological stress index based on physiological data and physiological safety baseline model.

[0010] S4. Use the physiological stress index to perform nonlinear correction on the kinematic deviation index to generate risk-adjusted deviation;

[0011] S5. Generate an adaptive assessment score based on risk-adjusted bias;

[0012] S6. Based on the physiological stress index, determine the system risk level and implement closed-loop intervention corresponding to the system risk level.

[0013] Preferably, S2 specifically includes:

[0014] S21. Guide the patient to complete the standardized baseline test and collect physiological data under resting and standard load conditions;

[0015] S22. Based on physiological data, determine the baseline resting heart rate, maximum safe heart rate, and safe threshold for electromyographic signal energy.

[0016] S23. Based on baseline resting heart rate, maximum safe heart rate and electromyographic signal energy safety threshold, construct a physiological safety baseline model.

[0017] Preferably, the S3 calculation of the physiological stress index specifically includes:

[0018] S31. Normalize the real-time heart rate in the physiological data with the baseline heart rate and maximum safe heart rate in the physiological safety baseline model to obtain the heart rate risk level.

[0019] S32. Normalize the electromyographic signal energy in the physiological data with the electromyographic signal energy safety threshold in the physiological safety baseline model to obtain the electromyographic risk level.

[0020] S33. The heart rate risk score and electromyographic risk score are weighted and summed to generate the physiological stress index.

[0021] Preferred options also include:

[0022] After calculating the kinematic deviation index and the physiological stress index, the deviation source attribute is determined based on the two-dimensional state space of the kinematic deviation index and the physiological stress index.

[0023] When the kinematic deviation index is high and the physiological stress index is low, it is judged as intentional movement deviation; when the kinematic deviation index is high and the physiological stress index is high, it is judged as protective movement deviation; when the kinematic deviation index is low and the physiological stress index is high, it is judged as latent physiological risk.

[0024] Preferably, S4 specifically includes:

[0025] S41. Construct a risk regulation factor with physiological stress index as the independent variable. When physiological stress index increases, the risk regulation factor decreases smoothly.

[0026] S42. Multiply the risk adjustment factor by the kinematic deviation index to generate the risk-adjusted bias.

[0027] Preferably, S6 specifically includes:

[0028] S61. Calculate the rate of change of the physiological stress index within a preset time window;

[0029] S62. If the instantaneous value of the physiological stress index exceeds the first threshold, the system risk level is determined to be Level 1 risk; if the instantaneous value of the physiological stress index exceeds the second threshold which is higher than the first threshold, and the rate of change exceeds the preset rate, the system risk level is determined to be Level 2 risk.

[0030] S63. Execute voice prompt intervention corresponding to Level 1 risk, or execute pause training and send alarm intervention corresponding to Level 2 risk.

[0031] Preferred options also include:

[0032] Record risk events that trigger voice prompts for intervention, pause training, or send alarms to create a risk event log.

[0033] Based on risk event logs, subsequent training plans are dynamically adjusted. If a specific action repeatedly triggers a level 1 risk, the training parameters for that action are reduced; if a level 2 risk is triggered, the action is temporarily removed from the training plan.

[0034] Based on the remote assessment and monitoring system for rehabilitation treatment of the elderly, including:

[0035] The data acquisition module is used to collect patients' kinematic and physiological data;

[0036] The baseline and template management module is used to establish a physiological safety baseline model for patients based on standard test data and to obtain preset standard exercise templates.

[0037] The risk quantification module is used to calculate the kinematic deviation index based on kinematic data and standard exercise templates; and to calculate the physiological stress index based on physiological data and a physiological safety baseline model.

[0038] The adaptive assessment module is used to perform nonlinear correction of the kinematic deviation index using the physiological stress index, generate risk-adjusted deviation, and generate an adaptive assessment score based on the risk-adjusted deviation.

[0039] The closed-loop intervention module is used to determine the system risk level based on the physiological stress index and to execute closed-loop interventions corresponding to the system risk level.

[0040] This invention provides an improved method and system for remote assessment and monitoring of rehabilitation treatment in the elderly, which, compared with the prior art, has the following improvements and advantages:

[0041] 1. This invention represents a breakthrough in assessment dimensions, transforming rehabilitation assessment from a singular evaluation of post-mortem morphology to a comprehensive insight into the process state. Existing technologies only assess the quality of movements based on kinematic data, failing to reflect the intrinsic physiological costs incurred by patients when performing these movements. This invention overcomes the limitations of universal assessment standards by simultaneously collecting patients' kinematic and physiological data and establishing an individualized physiological safety baseline model for each patient. It couples and analyzes quantified kinematic deviation indices with physiological stress indices, using the latter to nonlinearly correct the former. This design enables the assessment results to dynamically and reasonably reflect the patient's true physiological load, especially scientifically evaluating protective movement deviations caused by physiological discomfort. The resulting adaptive assessment scores are more comprehensive, objective, and humane.

[0042] 2. A proactive closed-loop safety mechanism has been constructed, transforming remote monitoring from a passive observer to an active guardian, greatly enhancing the safety of remote rehabilitation. Existing technologies lack the ability to effectively identify and intervene in potential physiological risks. This invention achieves precise quantification of the patient's comprehensive physiological risks by constructing a physiological stress index composed of a weighted fusion of real-time heart rate and electromyographic signal energy. Furthermore, the system not only analyzes the instantaneous value of this index but also its rate of change within a time window, thereby establishing a graded early warning system capable of distinguishing risk levels. For Level 1 risks, the system executes voice prompt intervention; for Level 2 risks where both the instantaneous value and rate of change exceed limits, it executes a forced suspension of training and sends an alarm. This mechanism ensures that any abnormal deterioration trend in the physiological state can be detected and dealt with in a timely manner, effectively avoiding the occurrence of serious safety incidents.

[0043] 3. This invention deepens the understanding of the rehabilitation process and achieves an intelligent upgrade from deviation identification to deviation attribution. Based on a two-dimensional state space constructed from kinematic deviation index and physiological stress index, the invention can identify the intrinsic motivations of movement deviations. The system can clearly distinguish between intentional movement deviations caused by inattention, protective movement deviations caused by physiological discomfort, and hidden physiological risks of standard movements but excessive physiological load. This deep-level discrimination ability makes subsequent assessment, correction, and intervention more targeted, avoiding the drawbacks of homogenizing all deviations and making the system's feedback and guidance more accurate and intelligent.

[0044] 4. This invention introduces a long-term adaptive correction capability based on risk logs, enabling rehabilitation programs to self-optimize and continuously iterate. It not only focuses on intervention for instantaneous risks but also creates a detailed risk event log by recording all risk events that trigger intervention. Based on the analysis of this log, the system can dynamically correct subsequent training programs. For example, it can automatically reduce the training parameters of actions that repeatedly trigger low-level risks, or temporarily remove actions that have caused high-level risks from the training plan and mark them for manual review. This mechanism allows rehabilitation programs to learn and adjust based on the patient's long-term performance and risk feedback, truly achieving personalized and long-term rehabilitation guidance under the premise of safety. Attached Figure Description

[0045] The present invention will be further explained below with reference to the accompanying drawings and embodiments:

[0046] Figure 1 This is a flowchart of the remote assessment and monitoring method for rehabilitation treatment of the elderly based on the present invention. Detailed Implementation

[0047] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0048] Example 1

[0049] Please see Figure 1 This invention provides a remote assessment and monitoring method for rehabilitation treatment of the elderly, comprising the following steps:

[0050] S1. Collect the patient's kinematic and physiological data;

[0051] S2. Establish a physiological safety baseline model for the patient based on standard test data and obtain a preset standard exercise template;

[0052] S3. Calculate the kinematic deviation index based on kinematic data and standard exercise templates; calculate the physiological stress index based on physiological data and physiological safety baseline model.

[0053] S4. Use the physiological stress index to perform nonlinear correction on the kinematic deviation index to generate risk-adjusted deviation;

[0054] S5. Generate an adaptive assessment score based on risk-adjusted bias;

[0055] S6. Based on the physiological stress index, determine the system risk level and implement closed-loop intervention corresponding to the system risk level;

[0056] This invention provides a remote assessment and monitoring method for rehabilitation treatment of the elderly, aiming to solve the problem that traditional remote rehabilitation assessment only focuses on movement patterns while ignoring the patient's internal physiological state, resulting in one-sided assessment results and the inability to detect potential risks in a timely manner; this method achieves comprehensive, dynamic and adaptive assessment and monitoring of the rehabilitation process by coupling and analyzing kinematic data and physiological data.

[0057] By constructing a multivariate evaluation model, the kinematic deviation index and physiological stress index are used as endogenous variables to dynamically adjust the evaluation of motor performance, thereby intelligently distinguishing between protective movement deviation caused by physiological discomfort and intentional deviation caused by inattention and other reasons.

[0058] The complete process of this method includes the following steps:

[0059] Step S1: Collect the patient's kinematic and physiological data;

[0060] In this step, kinematic data is used to accurately capture the external morphology of the patient performing rehabilitation movements. In this embodiment, one or more machine vision sensors are used to capture multi-dimensional kinematic data of the patient in real time at a frequency of 30 frames per second during the performance of rehabilitation movements, forming a kinematic sequence. This sequence mainly covers the three-dimensional coordinates, joint movement angles, and angular velocities of key joint points.

[0061] Through deep learning models, such as pose estimation models, video streams can be analyzed in real time to identify and track the three-dimensional coordinates of key points of the human skeleton, such as the shoulder, elbow, wrist, hip, knee, and ankle.

[0062] Physiological data aims to simultaneously monitor the patient's physiological load and stress response during the performance of actions; in this embodiment, a wearable physiological sensor device integrating a heart rate sensor and an electromyography sensor is used to simultaneously collect the patient's core physiological indicators, including real-time heart rate. and electromyographic signal energy All collected kinematic and physiological data are appended with precise hardware timestamps to ensure the accuracy of kinematic sequences. With physiological indicators Strict alignment in the time dimension forms a coupled multimodal dataset, providing a high-quality and highly correlated data foundation for subsequent coupled analysis;

[0063] Step S2: Establish a physiological safety baseline model for the patient based on standard test data, and obtain a preset standard exercise template;

[0064] The physiological safety baseline model is a reference model that includes individualized physiological safety parameters. Its purpose is to provide a dynamic, personally-specific definition of the safety zone for subsequent quantification of physiological stress levels; standard exercise templates. It is an idealized kinematic sequence of rehabilitation movements used as an evaluation standard. It can be a multidimensional time series, where each frame of data contains patient kinematic data. The same fields, such as joint 3D coordinates and joint angles, are intended to provide an accurate reference for calculating kinematic deviations. In this embodiment, the template can be formed by a rehabilitation therapist wearing the same motion capture device as the patient and personally completing a standard rehabilitation movement and recording it; or it can be directly retrieved from a standardized database containing a variety of common rehabilitation movements. The model can be constructed based on all the collected core physiological indicators, including but not limited to heart rate and electromyographic signal energy.

[0065] Step S3: Calculate the kinematic deviation index based on kinematic data and standard exercise template; calculate the physiological stress index based on physiological data and physiological safety baseline model.

[0066] Kinematic deviation index It is a scalar value used to quantify the morphological differences between a patient's actual movements and standard movements;

[0067] One common method for calculating this is the dynamic time warping algorithm; this algorithm can find the patient's kinematic sequence. With standard motion template The optimal matching path between them is determined, and the cumulative distance along the path is calculated to quantify the morphological differences between them. The calculation formula can be expressed as:

[0068]

[0069] in, Indicates the currently collected patient kinematic time series, subscript Take from the current value, This indicates a preset standard rehabilitation movement template sequence, with subscripts indicating the sequence. Taken from a reference value; in this embodiment, A dynamic time warping algorithm based on Euclidean distance is used to calculate the distance between sequences. To limit excessive distortion of the warped path, the Sakoe-Chiba global path constraint window width is set. For the sequence length ,Right now ;

[0070] Specifically, The function uses dynamic programming to solve for the minimum cumulative distance between two sequences; let... The length is , The length is Construct the cumulative distance matrix , of which elements express Center front points and Center front The optimal matching path distance for each point is calculated using the following formula:

[0071]

[0072] in, and This includes joint angles, angular velocities, and end effector spatial coordinates. Multidimensional feature vectors; in calculating Euclidean distance Previously, the Z-score method was used to normalize the data in each dimension in order to eliminate the influence of different physical dimensions;

[0073] This is the time step index for the standard action sequence, with a value range of [value range missing]. ; This is the time step index for the real-time action sequence, with a value range of [value range missing]. ; Represents the Euclidean distance between two eigenvectors; eigenvectors and Specifically, it includes joint angle and angular velocity data, which have been normalized using the Z-score method before calculation to eliminate dimensional differences;

[0074] Physiological stress index It is a comprehensive indicator used to quantify the degree to which a patient's current physiological state deviates from their individualized safety baseline; the specific calculation methods for both will be detailed in subsequent implementation methods.

[0075] In this embodiment, the DTW algorithm based on Euclidean distance is used, and a fixed window constraint is adopted, such as Sakoe-Chiba, to improve computational efficiency.

[0076] Step S4: Use the physiological stress index to perform nonlinear correction on the kinematic deviation index to generate risk-adjusted deviation;

[0077] The purpose of this step is to use the patient's physiological state as an endogenous variable in the evaluation function to dynamically adjust the assessment of movement morphology deviations. When the patient performs the movement with great effort under a high physiological load, the system should provide a more reasonable evaluation even if there are some deviations in movement morphology. This non-linear correction reflects an understanding and tolerance for protective movement deviations, and the corrected deviation is used to generate a risk-adjusted bias. ;

[0078] Step S5: Generate an adaptive assessment score based on the risk-adjusted bias;

[0079] Adaptive evaluation score This is the final score presented to the user or therapist that comprehensively reflects the rehabilitation performance; the purpose of this step is to mitigate risk-adjusted biases that incorporate physiological cost information. Mapping to an intuitive, bounded scoring range allows the assessment results to not only reflect the degree of action completion but also implicitly reflect the physiological cost of completing the action.

[0080] It is important to note that, to ensure the robustness of the model, when the physiological stress index... Exceeding the preset upper limit, for example, when At this time, the system will no longer calculate the adaptive assessment score, but will directly determine it as high risk and trigger the corresponding safety intervention to prevent the assessment score from being unreasonably lenient due to extremely high physiological stress;

[0081] Step S6: Determine the system risk level based on the physiological stress index, and implement closed-loop intervention corresponding to the system risk level;

[0082] The purpose of this step is to build a proactive security protection mechanism, enabling the system to evolve from a passive assessment tool into an intelligent guardian with proactive protection capabilities; the determination of the system's risk level is based on the physiological stress index. The instantaneous value and short-term trend of change; closed-loop intervention is based on different risk levels, and the system automatically executes a series of graded intervention measures, from voice prompts to forced suspension of training and alarms;

[0083] The method described in this invention constructs a multi-dimensional assessment and monitoring framework by deeply coupling and analyzing kinematic and physiological data. It overcomes the one-sidedness of existing technologies that rely solely on movement patterns for assessment, and can comprehensively and objectively reflect the true state of elderly patients in rehabilitation training. In particular, it can identify and quantify protective movement deviations caused by physiological discomfort. The generated adaptive assessment scores are more scientific and humane. At the same time, the closed-loop intervention mechanism based on physiological stress greatly improves the safety of remote rehabilitation, providing a complete technical solution for achieving truly personalized, safe, and efficient remote rehabilitation guidance.

[0084] Although this method effectively combines kinesiological and physiological data, future research could further incorporate more dimensions of information, such as patients' subjective pain ratings and emotional states, into the assessment model to achieve more comprehensive rehabilitation monitoring and guidance.

[0085] Example 2

[0086] S2 specifically includes:

[0087] S21. Guide the patient to complete the standardized baseline test and collect physiological data under resting and standard load conditions;

[0088] S22. Based on physiological data, determine the baseline resting heart rate, maximum safe heart rate, and safe threshold for electromyographic signal energy.

[0089] S23. Construct a physiological safety baseline model based on baseline resting heart rate, maximum safe heart rate and electromyographic signal energy safety threshold.

[0090] This embodiment is a specific implementation of step S2 in embodiment 1. The core is to construct an individualized physiological safety baseline model for each patient. This individualized modeling is the cornerstone of the accuracy of all subsequent physiological risk assessments.

[0091] This step may include:

[0092] Step S21: Guide the patient to complete the standardized baseline test and collect physiological data under resting and standard load conditions;

[0093] Standardized baseline testing is a testing procedure involving specific combinations of movements conducted before the formal start of rehabilitation training. Its purpose is to safely and controllably acquire key physiological parameters of patients under different physiological states. In this embodiment, the testing procedure includes: sitting quietly for 3 minutes to collect physiological data in a resting state; and completing a set of standard load movements at low intensity and slow speed to collect physiological data under standard load.

[0094] Step S22: Based on physiological data, determine the baseline resting heart rate, maximum safe heart rate, and safe threshold for electromyographic signal energy.

[0095] The purpose of this step is to extract the three core parameters that constitute the physiological safety baseline model from the collected raw physiological data.

[0096] Baseline resting heart rate Determination: It is the patient's heart rate in a completely relaxed state; it is obtained by processing the 3-minute resting heart rate data collected in step S21, removing outliers, and calculating its average value.

[0097] Maximum safe heart rate Determination: It is the theoretically upper limit of the heart rate that a patient should not exceed during exercise; it adopts a recognized exercise heart rate formula, such as the 208-0.7 age formula for the elderly, and makes fine adjustments based on the patient's health condition and the professional judgment of the rehabilitation therapist.

[0098] Electromyographic signal energy safety threshold Determination: This represents the reasonable upper limit of the electromyographic signal energy of the target muscle group when the patient performs the movement normally. It is derived from analyzing the electromyographic signals collected during the standard load exercise in step S21, calculating their energy values, and taking the 95th percentile of their statistical distribution as the safe threshold. Analyze the electromyographic signals collected during the standard load exercise in step S21, calculate their energy values, and take the 95th percentile of their statistical distribution as the safety threshold. ;

[0099] Step S23: Construct a physiological safety baseline model based on baseline resting heart rate, maximum safe heart rate, and electromyographic signal energy safety threshold;

[0100] The purpose of this step is to apply the three core parameters calculated above... , , The data is integrated into a structured data model, namely the physiological safety baseline model; this model is stored and linked to the patient's personal file, serving as a personalized reference benchmark for all subsequent real-time physiological data assessments.

[0101] Compared to existing technologies that use universal physiological parameter standards, this embodiment greatly improves the accuracy and sensitivity of physiological stress assessment by establishing an individualized physiological safety baseline model for each patient. It ensures that risk assessment is based on the patient's own physiological condition, rather than a vague population average, thereby enabling earlier and more accurate detection of individualized physiological risks and providing a solid foundation for truly personalized safety monitoring and intervention.

[0102] The S3 calculation of the physiological stress index specifically includes:

[0103] S31. Normalize the real-time heart rate in the physiological data with the baseline heart rate and maximum safe heart rate in the physiological safety baseline model to obtain the heart rate risk level.

[0104] S32. Normalize the electromyographic signal energy in the physiological data with the electromyographic signal energy safety threshold in the physiological safety baseline model to obtain the electromyographic risk level.

[0105] S33. Weighted summation of heart rate risk and electromyographic risk to generate physiological stress index;

[0106] This embodiment is a specific implementation of the calculation of the physiological stress index in step S3 of embodiment 1; the purpose is to fuse multi-source physiological data into a single physiological stress index that can intuitively reflect the overall physiological risk. ;

[0107] To quantify the comprehensive physiological stress risk, this embodiment introduces a physiological stress index. The calculation method is as follows:

[0108]

[0109] in, Physiological stress index, dimensionless, calculated by this formula; its physical meaning is the comprehensive degree to which the patient's current physiological state deviates from its safe baseline;

[0110] Real-time heart rate, beats / minute, collected in real time via wearable physiological sensors;

[0111] Baseline resting heart rate, beats / minute, derived from the physiological safety baseline model established in the previous steps;

[0112] Maximum safe heart rate, beats / minute, derived from the physiological safety baseline model established in the previous steps;

[0113] Electromyographic signal energy, a unit of energy, obtained by analyzing recent... The parameter is obtained by applying a fast Fourier transform to a raw electromyographic signal sample within a 2-second interval and calculating its power spectral density integral within a preset tremor characteristic frequency band; this parameter is used to quantify the muscle contraction intensity, abnormal tremor, or degree of fatigue.

[0114] : Safety threshold for electromyographic signal energy, and The dimensions are the same, and their units are the same as those of the same quantity. Maintaining consistency stems from the physiological safety baseline model established in the preceding steps;

[0115] Weighting coefficients are dimensionless and their values ​​are derived from statistical analysis of the correlation between different physiological indicators and adverse events in historical clinical data. These values ​​are optimized using models such as logistic regression to maximize the index. The accuracy of predictions for known risk events; for example, it can be set to... ,and ;

[0116] in, To estimate the power spectral density using the Welch method, a Hamming window was selected for calculation, with a window length of 256 and an overlap rate of 50%; the integration frequency range was... The selected frequency range is 4-12Hz, which corresponds to the characteristic frequency range of pathological tremors in the target population. For example, integration can be performed within the 4-12Hz range.

[0117]

[0118] Regarding weighting coefficients and The settings are determined quantitatively based on the focus of rehabilitation treatment as follows:

[0119] If the rehabilitation program is marked as cardiopulmonary function priority, then set... ;

[0120] If the rehabilitation program is marked as limb motor control priority, then set... ;

[0121] If it is a balanced type, then set ;

[0122] This embodiment uses a linear weighted summation model, which has good interpretability and computational efficiency in most clinical scenarios. Although in some complex cases, the nonlinear interaction between heart rate and electromyographic signals may require a more complex mathematical model to characterize it.

[0123] The design of this formula is based on a comprehensive judgment logic of exercise risk in elderly patients, namely, excessive deviation of heart rate and abnormal muscle activity are two core risk signals; the use of a weighted linear model is an efficient mathematical simulation of this multi-factor decision-making process; both components in the formula have been normalized to solve the problem of inconsistent dimensions of different physiological signals, so that they can be added meaningfully.

[0124] In this embodiment, the physiological stress index is calculated. The steps are as follows:

[0125] Step S31: Normalize the real-time heart rate in the physiological data with the baseline heart rate and maximum safe heart rate in the physiological safety baseline model to obtain the heart rate risk level;

[0126] Heart rate risk is the first term in the formula. It will show the current heart rate. The position within the safe heart rate zone is linearly mapped to the vicinity of [0, 1], intuitively reflecting the relative level of heart rate load;

[0127] Step S32: Normalize the electromyographic signal energy in the physiological data with the electromyographic signal energy safety threshold in the physiological safety baseline model to obtain the electromyographic risk level.

[0128] Electromyographic risk is the second term in the formula. It directly calculates real-time electromyographic energy. Relative to the safety threshold The proportion of risk is such that when muscles experience abnormal fatigue or tremors, the risk value will be significantly greater than 1, thus effectively identifying the risk at the muscle level.

[0129] Step S33: Weight the heart rate risk score and electromyography risk score to generate a physiological stress index;

[0130] This step integrates the two dimensions of risk; by assigning different weights, the model's sensitivity to different physiological risks can be flexibly adjusted according to different recovery stages or patient types; the final generated physiological stress index As a unified and quantitative risk indicator, its output directly serves the subsequent adaptive assessment module and closed-loop intervention module;

[0131] This embodiment integrates two physiological indicators with different properties and dimensions into a single, continuous, and standardized physiological stress index by normalization and weighted summation. This not only solves the problem of the difficulty in comprehensively evaluating multi-source heterogeneous physiological data, but also endows the model with high clinical adaptability and flexibility through adjustable weight coefficients, making risk quantification more accurate and comprehensive.

[0132] Also includes:

[0133] After calculating the kinematic deviation index and the physiological stress index, the deviation source attribute is determined based on the two-dimensional state space of the kinematic deviation index and the physiological stress index.

[0134] When the kinematic deviation index is high and the physiological stress index is low, it is identified as intentional movement deviation; when both the kinematic deviation index and the physiological stress index are high, it is identified as protective movement deviation; when both the kinematic deviation index and the physiological stress index are low, it is identified as latent physiological risk.

[0135] This embodiment adds a step of deviation source attribute identification based on the method of Embodiment 1; this step calculates the kinematic deviation index. With physiological stress index The purpose of subsequent implementation is to qualitatively distinguish the underlying causes of action deviations, so as to provide a deeper basis for decision-making in subsequent evaluation, correction and human intervention.

[0136] The underlying logic is that the system is based on the kinematic deviation index. With physiological stress index A two-dimensional state space is constructed, and the source attribute of the deviation is determined based on the combined state of the two indices using a preset discrimination rule; the discrimination rule is as follows:

[0137] When the kinematic deviation index is high and the physiological stress index is low, it is judged as intentional movement deviation;

[0138] Intentional movement deviation refers to the distortion of movement caused by factors such as lack of concentration, misunderstanding of the key points of the movement, or subjective laxity when the patient is in a good physiological state.

[0139] The threshold for judging high and low values ​​is derived from the boundary line determined after performing K-means cluster analysis on a large amount of historical rehabilitation sample data; for example, The threshold can be taken from all samples. 75th percentile of the value The threshold can be taken as the first-level risk threshold. 50%;

[0140] In this state, the system can infer that the deviation is not caused by physiological discomfort, and subsequent intervention strategies can focus on strengthening movement guidance;

[0141] When both the kinematic deviation index and the physiological stress index are high, it is judged as a protective movement deviation;

[0142] Protective movement deviation refers to the movement deformation that occurs when a patient subconsciously changes their movement pattern to avoid further damage due to physiological discomfort such as pain or fatigue.

[0143] In this state, the system recognizes that action deviation and physiological stress occur simultaneously, which provides a direct logical basis for the subsequent adaptive evaluation module to handle such deviations tolerantly.

[0144] When the kinematic deviation index is low and the physiological stress GLISH index is high, it is judged as a latent physiological risk.

[0145] Hidden physiological risks refer to situations where, although a patient appears to be able to perform actions correctly, their internal physiological systems are already under tremendous strain.

[0146] In this state, the system can detect this high-risk signal of discrepancy between appearance and reality and trigger timely intervention to prevent accidents caused by potential physiological problems.

[0147] This embodiment achieves a cognitive upgrade from identifying the nature of the deviation to understanding the reasons for it by introducing deviation source attribute discrimination. This enables the system to gain insight into the patient's internal state. This qualitative discrimination provides more refined and intelligent decision input for subsequent adaptive assessment and graded intervention. For example, it can strictly deduct points for intentional deviations while making lenient adjustments for protective deviations, greatly improving the rationality of the assessment and the pertinence of the intervention.

[0148] S4 specifically includes:

[0149] S41. Construct a risk regulation factor with physiological stress index as the independent variable. When physiological stress index increases, the risk regulation factor decreases smoothly.

[0150] S42. Multiply the risk adjustment factor by the kinematic deviation index to generate the risk-adjusted bias;

[0151] This embodiment is a specific implementation of step S4 in embodiment 1; the core purpose is to integrate the physiological state determined in the preceding steps into the evaluation of kinematic performance, thereby achieving the adaptability of the evaluation model.

[0152] To utilize the physiological stress GLISH index for nonlinear correction of the kinematic deviation index, this embodiment introduces risk-adjusted post-bias. The concept and calculation method are as follows:

[0153]

[0154] in, Risk-adjusted bias, and The units are the same, and the result is calculated using this formula;

[0155] : Kinematic deviation index, a dimensionless distance value, calculated from the preceding steps;

[0156] Physiological stress index, dimensionless, calculated from the preceding steps;

[0157] : This is a risk moderating factor used to control the degree of influence of physiological stress on the performance difference score; The value of is determined by grid search optimization based on standard rehabilitation dataset, and the objective function of optimization is to maximize the Pearson correlation coefficient between the automatic system score and the rehabilitation physician expert score;

[0158] In this embodiment, The preferred value range is The specific values ​​are obtained from a pre-defined lookup table based on the patient's disease type, such as stroke or post-fracture surgery; the source is determined through grid search optimization on a gold-standard dataset labeled by rehabilitation experts. The specific process of grid search is as follows: constructing a dataset containing... The gold standard dataset contains 100 samples, each containing the raw bias calculated by the system. Physiological stress index and expert ratings ;set up The search scope is Step size is The optimization is performed with the objective function of minimizing the mean square error between the predicted score and the expert score.

[0159]

[0160] in, The value of is determined by a grid search method, with the search range set as . The search step size is The goal is to make the calculated score consistent with the expert score. Minimize the mean square error;

[0161] The goal is to ensure that the final generated assessment scores are most consistent with the subjective assessment results of the experts; for example, The value can range from [0.5, 2.0].

[0162] in, It is a non-negative regulatory factor used to control the strength of physiological stress correction of motor deviations. The value of was determined by optimizing the historical clinical dataset using a grid search method, and the preferred value range was . ;when When the time is right, it indicates that the system has a higher tolerance for motor deviations under high physiological stress, that is, it significantly reduces the corrected deviation value. ;

[0163] The core technical consideration in designing this formula is the need for a regulatory mechanism, when the physiological stress index... When the value is zero or very low, this mechanism should not or minimally affect the original kinematic bias. ; and when As the value increases, the mechanism should smoothly and non-linearly decrease the impact on... The penalty; this embodiment uses an inverse proportional function form. The mathematical properties are manifested as follows: when When the adjustment factor is 1, ;along with As it increases, the regulating factor decreases smoothly and approaches 0;

[0164] Mathematically, this factor is a typical nonlinear decay function, whose graph shows a decreasing trend, and the slope increases with... The increase and smooth decrease of the value precisely simulates the clinical logic of being more tolerant of biased evaluation under high physiological stress.

[0165] In this embodiment, risk-adjusted bias is generated. The steps are as follows:

[0166] Step S41: Construct a risk regulation factor with physiological stress index as the independent variable;

[0167] The risk adjustment factor is in the formula Partial; this is based on the physiological stress index. The function is the only independent variable; its role is to dynamically generate a multiplier between [0, 1] based on real-time physiological risk; when the physiological stress index increases, the risk regulation factor decreases smoothly.

[0168] Step S42: Multiply the risk adjustment factor by the kinematic deviation index to generate the risk-adjusted bias;

[0169] This step is the core operation for correction; it involves adjusting the original kinematic deviation index. Multiply by the risk adjustment factor calculated in the previous step; for example, if the patient's movement deviation... Under normal physiological conditions, , If the patient is under extremely high physiological stress, for example... The risk adjustment factor is At this point, the risk-adjusted deviation Although the deviation values ​​of the movement patterns are the same, the corrected deviation values ​​are significantly reduced.

[0170] This embodiment successfully internalizes physiological costs into the evaluation criteria by constructing a nonlinear correction model driven by the physiological stress index. This method enables the evaluation results to intelligently distinguish between what is not done well and what cannot be done, which is especially significant for identifying and reasonably evaluating deviations from protective movements. It makes the final evaluation score not only measure the movement pattern but also reflect the physiological state, greatly improving the scientific and humanistic level of remote evaluation.

[0171] We choose the inverse proportional function form. Because it can smoothly convert the original deviation The punishment is reduced, and when physiological stress occurs... When the level is low, the effect of punishment is not obvious, which is consistent with the clinical assessment logic of normal conditions;

[0172] To further clarify the implementation of step S5 in Example 1, this example focuses on the implementation of step S5 based on risk-adjusted deviation. Generate adaptive evaluation scores The process is described in detail; the purpose is to map an unbounded bias measure to a bounded, easily understood rating interval.

[0173] To achieve this objective, this embodiment employs an exponential decay model, calculated as follows:

[0174]

[0175] in, The adaptive assessment score, with a score unit such as 0-100, is calculated using this formula.

[0176] The preset full score, and The units are the same and are preset according to the application scenario, for example, 100;

[0177] The base of the natural logarithm;

[0178] : This represents the rating sensitivity coefficient. The value varies Dynamic adjustment, among which This represents the cumulative number of effective training sessions completed by the patient under this rehabilitation program; it also applies when the rehabilitation program is changed or reassessed. Reset to ;

[0179] Risk-adjusted bias, and They have the same dimensions and are calculated from the preceding steps;

[0180] in, The maximum score for evaluating this action, for example, is set as follows: ; This is the scoring sensitivity coefficient, used to adjust the rate at which scores decay as deviation increases; The value is dynamically adjusted according to the rehabilitation stage:

[0181] Dynamic adjustment specifically follows the following piecewise function rules, where The total number of effective rehabilitation training sessions completed by the patient to date:

[0182] .

[0183] In the early stages of rehabilitation, i.e., the low-demand phase, smaller requirements are set. Values, for example To slow down the decline in scores; in the later stages of rehabilitation, i.e., the high-precision stage, a larger [score] is set. Values, for example To improve sensitivity to subtle deviations;

[0184] The logic behind this formula is that it corrects for the physiological information already included in the previous calculation. As the core independent variable;

[0185] Exponential decay model Ensured the score The value is always in Between; when deviation As it approaches infinity, the fraction The score approaches 0, and when the deviation is 0, the score is full. This makes the scoring results not only intuitive, but also always within a meaningful range;

[0186] The exponential function ensures that when When the score is full marks ;along with Increase, fraction The descent is smooth and non-linear, which aligns with the conventional intuition of error penalty.

[0187] Through this step, the system will convert a technical, potentially difficult-to-understand, deviation value... This was converted into standardized assessment scores that are easy for users and therapists to interpret. ; due to its input After physiological regulation, the final score It can comprehensively and fairly reflect the patient's overall rehabilitation performance under specific physiological costs.

[0188] Example 3

[0189] S6 specifically includes:

[0190] S61. Calculate the rate of change of the physiological stress index within a preset time window;

[0191] S62. If the instantaneous value of the physiological stress index exceeds the first threshold, the system risk level is determined to be Level 1 risk; if the instantaneous value of the physiological stress index exceeds the second threshold which is higher than the first threshold, and the rate of change exceeds the preset rate, the system risk level is determined to be Level 2 risk.

[0192] S63. Execute voice prompt intervention corresponding to Level 1 risk, or execute pause training and send alarm intervention corresponding to Level 2 risk;

[0193] Also includes:

[0194] Record risk events that trigger voice prompts for intervention, pause training, or send alarms to create a risk event log.

[0195] Based on risk event logs, the subsequent training plan is dynamically adjusted. If a specific action repeatedly triggers a level 1 risk, the training parameters for that action are reduced; if a level 2 risk is triggered, the action is temporarily removed from the training plan.

[0196] This embodiment is a further specification of step S6 in embodiment 1, and incorporates the above-mentioned long-term adaptive correction logic, together forming a complete closed-loop intervention and correction system from instantaneous risk intervention to long-term solution optimization;

[0197] Classification and intervention of instantaneous risks:

[0198] The purpose of this section is to establish a graded, automated risk response mechanism based on the severity and worsening trend of physiological stress index, so as to ensure the safety of patients during remote training.

[0199] Step S61: Calculate the rate of change of the physiological stress index within a preset time window;

[0200] Rate of change of physiological stress index Physiological stress index It measures short-term trends; its function is to identify whether a physiological state is rapidly deteriorating; it is derived from data from N sampling points over the past period, for example, N=10, corresponding to 5 seconds of data. The value sequence is linearly fitted using the first-order least squares method, and its slope is calculated. Let the past... The time series of each sampling point is The corresponding physiological stress index is The rate of change of the physiological stress index The slope of the fitted line is calculated as follows:

[0201]

[0202] in, Indicates the current sampling time, in seconds; For example, the fixed sampling time interval of the sensor. ; The number of sampling points within the sliding time window, for example, taking ,correspond The formula reflects the time window in the most recent period; The average rate of change of physiological stress index within each sampling point;

[0203] Step S62: Determine the system risk level;

[0204] This step is based on the physiological stress index. instantaneous value and rate of change The system risk level is divided into two levels; the first threshold Second threshold The settings, for example , and preset rate The settings are based on the American College of Sports Medicine's guidelines for exercise physiology and safety in older adults, and are further adjusted to individual patient baseline models; these adjustments are based on the patient's resting heart rate. Compared to the standard resting heart rate of peers The deviation is corrected; the corrected first threshold The calculation formula is:

[0205]

[0206] in, The recommended value is, for example, 0.5. The sensitivity adjustment coefficient is determined based on linear regression analysis of historical clinical data. In this embodiment, it is taken as... ; The patient's measured resting heart rate; Standard resting heart rate values ​​derived from age-matched population health norms;

[0207] If the instantaneous value of the physiological stress index exceeds the first threshold, If so, the system risk level will be determined as Level 1 risk;

[0208] If the instantaneous value of the physiological stress index exceeds a second threshold that is higher than the first threshold, and the rate of change exceeds a preset rate, and If so, the system risk level will be determined as Level 2 risk;

[0209] Step S63: Implement closed-loop interventions corresponding to the risk level;

[0210] For Level 1 risks, the system will automatically issue a voice prompt to intervene, such as: Your heart rate is too high, please slow down and take a deep breath;

[0211] For Level 2 risks, the system immediately implements intervention by suspending training and sending an alarm, that is, forcibly suspending the current rehabilitation training and automatically sending an alarm message to the preset guardian or remote therapist.

[0212] Dynamically correct long-term training schemes based on risk logs;

[0213] The purpose of this section is to continuously optimize future training programs using historical risk data, thereby achieving long-term, learning-based adaptive rehabilitation guidance.

[0214] The system records risk events that trigger interventions, creating a risk event log. This log includes the timestamp of each intervention trigger, the associated rehabilitation actions, the relevant risk indicators, and the risk level at that time. Based on this risk event log, the system dynamically adjusts subsequent training plans.

[0215] If a specific action repeatedly triggers a Level 1 risk during recent training, the system will automatically reduce the training parameters for that action when generating the next training plan, for example, by reducing the number of repetitions by 20%-40%.

[0216] If a Level 2 risk has been triggered, the system will temporarily remove the specific action that caused the Level 2 risk from the training plan and prompt the therapist to conduct a manual assessment.

[0217] This invention constructs a dual-loop intelligent safety system; the first loop enables real-time monitoring and immediate intervention of instantaneous physiological risks; the second loop learns from historical risk events to achieve dynamic optimization and self-correction of long-term training programs; this combination of immediate response and long-term adaptation enables the system to fundamentally improve the personalization level and long-term safety of remote rehabilitation.

[0218] Example 4

[0219] Based on the remote assessment and monitoring system for rehabilitation treatment of the elderly, including:

[0220] The data acquisition module is used to collect patients' kinematic and physiological data;

[0221] The baseline and template management module is used to establish a physiological safety baseline model for patients based on standard test data and to obtain preset standard exercise templates.

[0222] The risk quantification module is used to calculate the kinematic deviation index based on kinematic data and standard exercise templates; and to calculate the physiological stress index based on physiological data and a physiological safety baseline model.

[0223] The adaptive assessment module is used to perform nonlinear correction of the kinematic deviation index using the physiological stress index, generate risk-adjusted deviation, and generate an adaptive assessment score based on the risk-adjusted deviation.

[0224] The closed-loop intervention module is used to determine the system risk level based on the physiological stress index and to execute closed-loop interventions corresponding to the system risk level.

[0225] Kinematic data is captured in real time by one or more machine vision sensors, and physiological data is collected synchronously by a wearable physiological sensor device that integrates a heart rate sensor and an electromyography sensor.

[0226] This invention also provides a remote assessment and monitoring system for rehabilitation treatment of the elderly, comprising the following functional modules:

[0227] The data acquisition module aims to simultaneously acquire the patient's kinematic and physiological data. In this embodiment, the module consists of a machine vision sensor, a wearable physiological sensor, and corresponding data synchronization and transmission software, and is responsible for executing step S1 in the aforementioned method.

[0228] The baseline and template management module aims to establish a personalized physiological safety baseline model for patients and manage standard motion templates for comparison; in this embodiment, this is a software module responsible for executing step S2 in the aforementioned method.

[0229] The risk quantification module aims to calculate the kinematic deviation index in real time. and physiological stress index In this embodiment, this is the core algorithm module, which receives real-time data streams from the data acquisition module and calls the dynamic time warping algorithm to calculate the kinematic deviation index. Meanwhile, the formula detailed above calculates the physiological stress index. This module is responsible for executing step S3 in the aforementioned method.

[0230] The adaptive assessment module aims to generate a final assessment score that comprehensively reflects the physiological costs; in this embodiment, the software module receives the output from the risk quantification module. and The risk-adjusted deviation is calculated based on the formula detailed above. and further based on Generate adaptive evaluation scores This module is responsible for executing steps S4 and S5 in the aforementioned method.

[0231] The closed-loop intervention module aims to implement proactive safety interventions and long-term program adjustments based on risk levels. In this embodiment, the module continuously monitors the physiological stress index output by the risk quantification module. Based on the logic defined above, the system risk level is determined and corresponding intervention actions are performed according to its rate of change; this module is responsible for executing step S6 in the aforementioned method.

[0232] These modules are tightly coupled through internal data interfaces and control logic, working together to form a complete system from data acquisition, risk quantification, adaptive assessment to closed-loop intervention;

[0233] This system, through its modular design, solidifies complex assessment and monitoring methods into a set of operational hardware and software entities; it provides an end-to-end solution that can automate and intelligently manage the entire process of remote rehabilitation for the elderly; compared to assessment software that relies on manual supervision or has limited functionality, this system integrates proactive risk protection and long-term adaptive learning capabilities, resulting in more scientific assessment results and higher security, truly enabling the elderly to enjoy safe, effective, and personalized rehabilitation guidance at home.

[0234] It should be noted that the above 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for remote assessment monitoring based on rehabilitation therapy for the elderly, characterized in that, The method comprises the following steps: S1, collecting kinematic data and physiological data of a patient; S2, establishing a physiological safety baseline model for the patient according to standard test data, and obtaining a preset standard movement template; S3, calculating a kinematic deviation index based on the kinematic data and the standard movement template, and calculating a physiological stress index based on the physiological data and the physiological safety baseline model; S4, using the physiological stress index to nonlinearly correct the kinematic deviation index to generate a risk-adjusted deviation; S5, generating an adaptive evaluation score based on the risk-adjusted deviation; S6, determining a system risk level according to the physiological stress index, and performing a closed-loop intervention corresponding to the system risk level; Further comprising: After the kinematic deviation index and the physiological stress index are calculated, the source attribute of the deviation is determined based on a two-dimensional state space of the kinematic deviation index and the physiological stress index; When the kinematic deviation index is a high value and the physiological stress index is a low value, it is determined to be an intentional action deviation; when the kinematic deviation index is a high value and the physiological stress index is a high value, it is determined to be a protective action deviation; when the kinematic deviation index is a low value and the physiological stress index is a high value, it is determined to be an implicit physiological risk.

2. The remote assessment monitoring method based on rehabilitation therapy for the elderly according to claim 1, characterized in that, S2 specifically comprises: S21, guiding the patient to complete a standardized baseline test to collect physiological data under rest and standard load; S22, determining a baseline resting heart rate, a maximum safe heart rate, and an electromyographic signal energy safety threshold based on the physiological data; S23, constructing a physiological safety baseline model based on the baseline resting heart rate, the maximum safe heart rate, and the electromyographic signal energy safety threshold.

3. The remote assessment monitoring method based on rehabilitation therapy for the elderly according to claim 1, characterized in that, S3 calculating the physiological stress index specifically comprises: S31, normalizing the real-time heart rate in the physiological data with the baseline heart rate and the maximum safe heart rate in the physiological safety baseline model to obtain a heart rate risk degree; S32, normalizing the electromyographic signal energy in the physiological data with the electromyographic signal energy safety threshold in the physiological safety baseline model to obtain an electromyographic risk degree; S33, performing weighted summation on the heart rate risk degree and the electromyographic risk degree to generate the physiological stress index.

4. The remote assessment monitoring method based on rehabilitation treatment for the elderly according to claim 1, characterized in that, S4 specifically comprises: S41, constructing a risk adjustment factor with the physiological stress index as the independent variable, and the risk adjustment factor smoothly decreases as the physiological stress index increases; S42, multiplying the risk adjustment factor by the kinematic deviation index to generate a risk-adjusted deviation.

5. The remote assessment monitoring method based on rehabilitation therapy for the elderly according to claim 1, characterized in that, S6 specifically comprises: S61, calculating the change rate of the physiological stress index within a preset time window; S62, if the instantaneous value of the physiological stress index exceeds a first threshold, the system risk level is determined to be a first-level risk; if the instantaneous value of the physiological stress index exceeds a second threshold higher than the first threshold and the change rate exceeds a preset rate, the system risk level is determined to be a second-level risk; S63, performing a voice prompt intervention corresponding to the first-level risk, or performing a pause training and alarm sending intervention corresponding to the second-level risk.

6. The method for remote assessment monitoring based on rehabilitation therapy for the elderly according to claim 5, characterized in that, Further comprising: Recording the risk events triggering the voice prompt intervention or the pause training and alarm sending intervention to form a risk event log; Based on the risk event log, dynamically correcting a subsequent training scheme, and if a certain specific action repeatedly triggers the first-level risk, reducing the training parameters of the action; If a secondary risk is triggered, the exercise is temporarily removed from the training program.

7. A remote assessment monitoring system for rehabilitation therapy of the elderly, characterized by, The remote assessment and monitoring method for rehabilitation treatment of the elderly according to any one of claims 1-6, comprising: a data acquisition module for acquiring kinematic data and physiological data of the patient; a baseline and template management module for establishing a physiological safety baseline model for the patient according to standard test data and obtaining a preset standard movement template; a risk quantification module for calculating a kinematic deviation index based on the kinematic data and the standard movement template, and calculating a physiological stress index based on the physiological data and the physiological safety baseline model; an adaptive assessment module for nonlinearly correcting the kinematic deviation index using the physiological stress index to generate a risk-adjusted deviation, and generating an adaptive assessment score based on the risk-adjusted deviation; a closed-loop intervention module for determining a system risk level according to the physiological stress index and performing a closed-loop intervention corresponding to the system risk level.

Images