A signal pattern recognition method and system for assessing motor function in a disabled patient

Abstract

The present application relates to the technical field of motor function assessment, and provides a signal mode recognition method and system for motor function assessment of disabled patients, which comprises the following steps: collecting motion signals of the disabled patients; performing quality inspection processing on the motion signals to obtain signal quality indexes corresponding to the motion signals; identifying task context parameters of the current rehabilitation tasks of the disabled patients; performing motor function assessment based on the motion signals to obtain motor function assessment results; adjusting the influence weight of the signal quality indexes on the reliability score of the motor function assessment results according to the task context parameters; determining the reliability score of the motor function assessment results according to the adjusted influence weight and the signal quality indexes; and displaying the reliability score and the motor function assessment results in a graphical interface display mode. The present application has the effect of improving the accuracy of the motor function assessment results of the disabled patients.

Description

Technical Field

[0001] This invention relates to the technical field of motor function assessment, and specifically to a signal pattern recognition method and system for assessing the motor function of disabled patients. Background Technology

[0002] Intelligent assessment systems perform well in controlled environments when evaluating motor function in disabled patients. However, when the assessment scenario expands to more complex environments closer to daily life, the system faces significant challenges from multiple dynamic interferences. These interferences originate from the environment itself, the contact state between the sensor and the skin, and changes in the physical position of the sensor, leading to a decline in the quality of the acquired signals and consequently affecting the accuracy and reliability of the assessment results. Rehabilitation therapists urgently need a method that allows them to understand the true state of the assessment data in real time, thereby more accurately judging the patient's rehabilitation progress.

[0003] In non-standardized scenarios such as simulated home rehabilitation training areas with multiple dynamic interference sources (e.g., electromagnetic interference, changes in sensor-skin contact, sensor physical position shifts), existing pattern recognition methods begin to exhibit significant performance degradation and misjudgments when processing these damaged, incomplete, or degraded signals. The original design and training of pattern recognition methods were based on clear, stable, and complete signal characteristics, assuming relatively pure input data. However, the received signals are now riddled with noise, initial potential fluctuations, uncertainties, and intermittent loss. For example, when surface electromyography (EMG) signals are contaminated by strong power frequency interference, pattern recognition methods may incorrectly identify noise peaks as strong muscle contractions, thus overestimating the patient's muscle strength or activation level. Alternatively, when inertial measurement unit (IMU) data drifts due to magnetic field interference, pattern recognition methods may incorrectly determine the patient's joint range of motion, movement trajectory, or movement stability, leading to inaccurate assessment reports of "functional improvement" or "functional deterioration." More seriously, when some sensor signals are intermittently lost due to poor contact, pattern recognition methods may fail to obtain complete motion information, leading to assessment interruptions, invalid outputs, or incorrect inferences based on incomplete data. Rehabilitation therapists have found that the assessment reports generated by the system have become unstable and unreliable, sometimes even differing significantly from the patients' actual performance as observed by the naked eye. This severely impacts their ability to formulate and adjust rehabilitation plans. They cannot trust this data because they are unsure whether these "abnormalities" represent the patient's true motor performance or are merely illusions caused by sensor or environmental interference. This skepticism about the reliability of the assessment results significantly diminishes the value of intelligent assessment systems and may even lead rehabilitation therapists to make incorrect treatment decisions, thereby delaying the patient's rehabilitation process.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] This application discloses a signal pattern recognition method and system for assessing motor function in disabled patients. It aims to solve the problem that existing intelligent assessment systems suffer from signal quality degradation due to multiple dynamic interferences in complex and non-standardized rehabilitation environments, which affects the accuracy and reliability of assessment results. It also addresses the problem that pattern recognition methods suffer from significant performance degradation and misjudgment when processing various types of damaged, incomplete, or degraded signals, which leads to rehabilitation therapists being unable to trust the assessment data and affects the development of rehabilitation plans and the patient's rehabilitation process.

[0006] The technical solution of this application is as follows: In a first aspect, this application discloses a signal pattern recognition method for assessing the motor function of disabled patients, comprising the following steps: Motion signals from disabled patients are collected, including surface electromyography (EMG) signals and inertial measurement unit (IMU) data. The IMU data includes at least triaxial acceleration data, triaxial angular velocity data, and attitude angle data. The motion signal is subjected to quality inspection processing to obtain the corresponding signal quality index. Identify the task context parameters of the current rehabilitation task for disabled patients; Motor function assessment is performed based on motion signals to obtain motor function assessment results; Adjust the weighting of signal quality indices on the reliability score of motor function assessment results based on task context parameters. The reliability score of the motor function assessment results is determined based on the adjusted influence weights and signal quality indicators. The reliability score and motor function assessment results are displayed in a graphical interface.

[0007] This technical solution effectively addresses multiple dynamic interferences in complex rehabilitation environments. By introducing task context parameters to adjust the influence weight of signal quality indicators on the reliability score of assessment results, it improves the accuracy and credibility of motor function assessment results, solving the problems of assessment results being greatly affected by interference and rehabilitation therapists finding it difficult to trust assessment data in existing technologies.

[0008] Secondly, this application also discloses a signal pattern recognition system for assessing the motor function of disabled patients, comprising: The motion signal acquisition module is used to acquire motion signals from disabled patients; these motion signals include surface electromyography signals and inertial measurement unit data. The quality index acquisition module is used to perform quality inspection on motion signals in order to obtain the signal quality index corresponding to the motion signal. The task context recognition module is used to identify the task context parameters of the current rehabilitation task for disabled patients; The motor function assessment module is used to perform motor function assessment based on motion signals and obtain motor function assessment results. The influence weight adjustment module is used to adjust the influence weight of signal quality indicators on the reliability score of motion function assessment results based on task context parameters. The reliability score determination module is used to determine the reliability score of the motion function assessment results based on the adjusted influence weights and signal quality indicators. The evaluation results display module is used to display the reliability score and motor function evaluation results in a graphical interface.

[0009] Through this technical solution, this application can provide a complete system that integrates signal acquisition, quality assessment, context recognition, functional assessment, weight adjustment and result display, thereby improving the intelligence and reliability of motor function assessment for disabled patients and effectively solving the problem of insufficient assessment accuracy of existing systems in complex environments.

[0010] Beneficial Effects: This application discloses a signal pattern recognition method for assessing motor function in disabled patients. It collects surface electromyography (EMG) signals and inertial measurement unit (IMU) data from disabled patients and performs quality control processing on these motion signals to obtain signal quality indices. Simultaneously, it identifies the task context parameters of the disabled patient's current rehabilitation task. Based on these parameters, a motor function assessment is performed using the motion signals to obtain preliminary assessment results. To address the problem of signal quality degradation caused by multiple dynamic interferences in complex, non-standardized rehabilitation environments, thus affecting the accuracy and reliability of assessment results, this application dynamically adjusts the influence weight of the signal quality index on the reliability score of the motor function assessment results according to the task context parameters. Subsequently, based on the adjusted influence weights and signal quality indices, the reliability score of the motor function assessment results is determined, and the reliability score and motor function assessment results are displayed in a graphical interface.

[0011] Through the above technical solutions, this application can effectively address various dynamic interference sources (such as electromagnetic interference, changes in sensor-skin contact state, and sensor physical position shift) in non-standardized scenarios such as simulated home rehabilitation training areas, overcoming the limitations of existing pattern recognition methods in terms of performance degradation and misjudgment when processing various types of damaged, incomplete, or degraded signals. Specifically, this application introduces task context parameters, enabling the system to adaptively adjust the degree of influence of signal quality on the reliability of evaluation results according to the characteristics of the rehabilitation task and environmental changes, avoiding misjudgments caused by signal quality issues. For example, when surface electromyography signals are affected by power frequency interference, the system no longer simply identifies noise peaks as muscle contraction, but adjusts the weight of the signal quality index according to the task context, thereby more accurately assessing muscle strength. When inertial measurement unit data drifts due to magnetic field interference, the system can also avoid misjudging joint range of motion or movement trajectory through weight adjustment. Furthermore, even if some sensor signals are intermittently lost due to poor contact, this application can provide more reliable evaluation results through dynamic weight adjustment of signal quality indicators, rather than simply interrupting or outputting invalid results.

[0012] In summary, this application significantly improves the accuracy, stability, and reliability of motor function assessment results for disabled patients through a dynamic adjustment mechanism for the influence weight of signal quality indicators. This enables rehabilitation therapists to understand the true status of assessment data in real time, more accurately judge the patient's rehabilitation progress, and thus effectively formulate and adjust rehabilitation plans. It avoids the problem of delaying the patient's rehabilitation process due to unreliable assessment results, and greatly enhances the application value of the intelligent assessment system. Attached Figure Description

[0013] Figure 1 This is a flowchart of a signal pattern recognition method for assessing motor function in disabled patients, as described in one embodiment of the present invention. Figure 2 This is a flowchart of a signal pattern recognition method for assessing motor function in disabled patients, as described in another embodiment of the present invention. Figure 3 This is a system block diagram of a signal pattern recognition system for assessing motor function in disabled patients, according to another embodiment of the present invention. Explanation of reference numerals in the attached figures: 1. Signal pattern recognition system for assessing motor function in disabled patients; 11. Motor signal acquisition module; 12. Quality indicator acquisition module; 13. Task context recognition module; 14. Motor function assessment module; 15. Influence weight adjustment module; 16. Reliability score determination module; 17. Assessment result display module. Detailed Implementation

[0014] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0015] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0016] This application proposes a signal pattern recognition method for assessing motor function in disabled patients, combining... Figure 1 As shown, it includes: S1, collect motion signals from disabled patients; motion signals include surface electromyography signals and inertial measurement unit (IMU) data; IMU data includes at least triaxial acceleration data, triaxial angular velocity data, and attitude angle data; S2 performs quality inspection on the motion signal to obtain the signal quality index corresponding to the motion signal. S3 identifies the task context parameters of the current rehabilitation task for disabled patients; S4, perform motor function assessment based on motion signals to obtain motor function assessment results; S5, adjust the weight of signal quality index on the reliability score of motor function assessment results according to task context parameters; S6. Determine the reliability score of the motion function assessment results based on the adjusted influence weights and signal quality indicators. S7 displays reliability scores and motion function assessment results in a graphical interface.

[0017] To facilitate understanding of the technical solutions in this application, the key terms used in the text will be explained in a unified manner.

[0018] "Motion signals" refer to the physiological electrical signals and physical motion data used to characterize the motor state of disabled patients. Motion signals mainly include surface electromyography (EMG) signals and inertial measurement unit (IMU) data. Surface EMG signals are muscle activity potential information collected through electrodes placed on the skin surface, reflecting muscle contraction intensity and activation patterns. IMU data is data collected by IMU sensors, including at least triaxial acceleration data, triaxial angular velocity data, and posture angle data. Triaxial acceleration data characterizes the linear acceleration changes of the limb in three orthogonal directions, triaxial angular velocity data characterizes the rotational velocity changes of the limb around three orthogonal axes, and posture angle data characterizes the limb's posture in space. By jointly acquiring surface EMG signals and IMU data, the motor function status of disabled patients can be characterized from both muscle activation and limb movement perspectives.

[0019] "Signal quality metrics" refer to a set of indicators that quantify the quality of acquired motion signals, reflecting the availability and reliability of motion signals under current acquisition conditions. Signal quality metrics may include signal-to-noise ratio, baseline drift, power frequency interference intensity, signal integrity, clipping, abnormal spikes, sensor drift, and data loss, among others.

[0020] "Task context parameters" refer to background and status information related to the current rehabilitation task of a disabled patient. These parameters can include the type of rehabilitation task, task difficulty level, patient fatigue level, known disturbances in the environment, the patient's current posture, and other contextual information that can affect the reliability of motor function assessment. By identifying task context parameters, the system can adopt a reliability assessment method that matches the current task in different rehabilitation scenarios.

[0021] "Motor function assessment results" refer to quantitative or qualitative results obtained based on motion signal analysis, used to characterize a patient's motor ability. Motor function assessment results may include joint range of motion, muscle strength, motor coordination, gait parameters, movement trajectory smoothness, or other rehabilitation assessment indicators.

[0022] A "reliability score" is a quantified result of the credibility of motor function assessment results, used to characterize the acceptability of the current motor function assessment results under signal quality conditions. Reliability scores can be expressed as percentages, grades, or other visually tiered formats.

[0023] "Influence weight" refers to the relative strength of each signal quality indicator's impact on the reliability of motion function assessment results when determining reliability scores. By adjusting the influence weights of different signal quality indicators, the system can highlight signal quality issues more relevant to the current assessment objective in different task scenarios.

[0024] Based on the above terminology definitions, this application proposes a method for determining the reliability of motor function assessment in the rehabilitation process of disabled patients. This method does not merely output motor function assessment results, but simultaneously quantifies the reliability of the assessment results by considering the quality of motor signals and the current task context, thereby providing rehabilitation therapists with a reliable basis corresponding to the assessment results.

[0025] During implementation, the first step is to collect motion signals from the disabled patient. This collection can be accomplished collaboratively using surface electromyography (SEMG) sensors deployed on the surface of target muscle groups and inertial measurement unit (IMU) sensors fixed to key limb sites. SEMG sensors can be deployed in the quadriceps, tibialis anterior, biceps brachii, deltoid, or other target muscle areas relevant to the current rehabilitation task to collect SEMG signals. IMU sensors can be deployed in the thigh, calf, upper arm, forearm, trunk, or other key sites to collect triaxial acceleration, triaxial angular velocity, and posture angle data. The sensors can be connected to the data acquisition terminal via wired or wireless means to form a motion signal acquisition link adapted to the current rehabilitation task.

[0026] After motion signal acquisition, quality control processing is performed on the motion signals to obtain corresponding signal quality indicators. This quality control process identifies signal problems in the acquisition chain that may affect the reliability of the motion function assessment results. For surface electromyography (EMG) signals, it analyzes the signal-to-noise ratio, baseline drift, power frequency interference intensity, signal saturation, and clipping. For inertial measurement unit (IMU) data, it analyzes data integrity, temporal continuity, drift, abnormal spikes, and signal anomalies caused by changes in sensor fixation. Through quality control, quality issues affecting the reliability of the assessment in the raw motion signals are transformed into quantifiable signal quality indicators, providing a basis for subsequent reliability scoring calculations.

[0027] After obtaining signal quality indicators, the system identifies the task context parameters for the current rehabilitation task of the disabled patient. The identification of task context parameters can be based on a combination of human input, environmental perception results, and preliminary analysis of motion signals. For example, the therapist can pre-input whether the current task is gait training, sit-to-stand transition training, upper limb lifting training, or fine motor skills training; environmental sensors can provide information on electromagnetic interference levels, ambient temperature, ambient humidity, or sound and light interference; inertial measurement unit data can also be used to identify whether the patient is currently standing, sitting, or lying down. By identifying task context parameters, the system can clearly define the task background corresponding to the current assessment and provide a contextual basis for subsequent weight adjustments.

[0028] After completing motion signal acquisition, quality control processing, and task context parameter identification, a motor function assessment is performed based on the motion signals to obtain the assessment results. The motor function assessment process can employ corresponding assessment models or analysis strategies depending on the specific rehabilitation task. When the current task is gait training, parameters such as stride length, cadence, swing phase time, stance phase time, gait cycle symmetry, and joint range of motion can be calculated based on inertial measurement unit (IMU) data. When the current task is upper limb lifting training, muscle activation sequence, activation intensity, and synergy can be analyzed based on surface electromyography (EMG) signals, and lifting angle, height, speed, and movement stability can be analyzed in conjunction with IMU data. When the current task is fine motor skills training, further attention can be paid to muscle fine control and local movement trajectory stability. Through the above analysis process, a motor function assessment result corresponding to the current rehabilitation task is generated.

[0029] After obtaining the motor function assessment results, the influence weights of each signal quality index on the reliability score of the motor function assessment results are adjusted according to the task context parameters. This process ensures that the calculation logic of the reliability score is consistent with the assessment focus of the current rehabilitation task. When the task context parameters indicate that the current task has high requirements for muscle activation precision and local movement control, the influence weights of indicators such as surface electromyography signal-to-noise ratio and baseline drift can be increased; when the task context parameters indicate that the current task is mainly based on large-amplitude limb movements and overall trajectory analysis, the influence weights of inertial measurement unit data integrity, drift degree, and posture continuity can be increased. Through dynamic adjustment of influence weights, the system can avoid using fixed reliability assessment rules to uniformly process different tasks, thereby improving the matching degree between the reliability score and the actual assessment scenario.

[0030] After adjusting the influence weights, the reliability score of the motion function assessment result is determined based on the adjusted influence weights and signal quality indicators. The reliability score determination process can be based on a weighted fusion calculation of multiple signal quality indicators and their corresponding influence weights, ensuring that the degree of influence of various signal quality issues on the final reliability score is consistent with the current task context. When a certain type of signal quality indicator has a high weight in the current task and the quality state corresponding to that indicator is poor, the reliability score decreases accordingly; conversely, when key signal quality indicators affecting the core assessment content of the current task remain at a good level, the reliability score increases accordingly. Thus, the system can output a quantitative result of the credibility corresponding to the current motion function assessment result.

[0031] After obtaining the reliability score, the reliability score and motor function assessment results are displayed in a graphical interface. The graphical interface can be implemented on a computer, tablet, rehabilitation assessment terminal, or other display device, and is used to simultaneously display the patient's motor function assessment results and their corresponding reliability score. Motor function assessment results can be displayed in the form of line graphs, bar charts, trajectory graphs, gait diagrams, or muscle activation diagrams; the reliability score can be presented using numerical displays, color-coded grading, level labels, or icon prompts. By displaying the reliability score and motor function assessment results synchronously, rehabilitation therapists can intuitively understand the reliability of the current results while reviewing them, and further analyze the main factors affecting reliability in conjunction with specific signal quality indicators, thereby improving the interpretability and usability of the rehabilitation assessment results.

[0032] The key technical point of this application lies in not separating the motor function assessment results from their reliability. Instead, it establishes a correspondence between motor signal quality issues and current task requirements through signal quality indicators, task context parameters, and an influence weight adjustment mechanism. This allows the reliability score to reflect the actual reliability of the motor function assessment results under different rehabilitation tasks. Therefore, it avoids misinterpretation, over-reliability, or incorrect judgment of assessment results due to fluctuations in sensor signal quality in complex rehabilitation scenarios, thereby improving the relevance, stability, and clinical auxiliary value of motor function assessment for disabled patients.

[0033] Optional, combined Figure 2 As shown, the steps for adjusting the influence weight of signal quality indicators on the reliability score of motor function assessment results based on task context parameters include: A1. Perform spectral analysis on the surface electromyography signal to obtain the instantaneous energy in the high-frequency band; A2, perform high-pass filtering on the inertial measurement unit data to obtain the root mean square value after filtering; A3. Based on the task context parameters, determine the macroscopic movement pattern and the allowable perturbation range corresponding to the current rehabilitation task; the macroscopic movement pattern is a templated description of the target movement trajectory and rhythm corresponding to the current rehabilitation task, including at least the main joint movement direction, movement amplitude envelope and movement rhythm parameters; the allowable perturbation range is the allowable small perturbation interval relative to the macroscopic movement pattern. A4, determine whether the instantaneous energy and root mean square value in the high-frequency band show perturbation characteristics within the same time window, and obtain the perturbation characteristic display judgment result; A5. When the perturbation feature shows a negative result, the interference type is determined to be no significant perturbation interference, and the weight of the signal quality index on the reliability score of the motion function assessment result is maintained at the weight preset in the previous cycle. A6. When the perturbation feature shows a judgment result of yes, based on the allowable perturbation range, the consistency between the perturbation feature and the macroscopic action pattern is compared to obtain the perturbation consistency comparison result. A7. When the perturbation feature shows a judgment result of yes, compare whether the abnormal features detected by different sensors corroborate each other to obtain the abnormal verification comparison result. A8. Determine the type of interference based on the perturbation consistency comparison results and the anomaly confirmation comparison results; A9, adjust the weighting of signal quality indicators on the reliability score of motion function assessment results according to the type of interference.

[0034] Specifically, in the spectral analysis of surface electromyography (EMG) signals, the acquisition of high-frequency instantaneous energy aims to capture rapid and transient signal changes that may occur during muscle activity. These changes may be related to factors such as muscle tremors, poor electrode contact, or external electromagnetic interference. Simultaneously, high-pass filtering is applied to the inertial measurement unit (IMU) data to remove low-frequency components such as gravity, thereby obtaining the filtered root mean square (RMS) value, which reflects the dynamic intensity and variability of limb movement. Macroscopic movement patterns can be understood as the standard or ideal execution method of rehabilitation tasks. Their templated description ensures the precise definition of the target movement trajectory, rhythm, and the envelope of the direction and amplitude of major joint movements. The permissible perturbation range defines the acceptable slight deviations from the standard pattern when the patient performs the task, which helps distinguish between normal physiological variations and abnormal interference.

[0035] The process involves determining whether the instantaneous energy and root mean square (RMS) values ​​in the high-frequency band exhibit perturbation characteristics within the same time window. This aims to initially identify any abnormal fluctuations through cross-validation of multimodal signals. If the result is negative, it is assumed that there is no significant perturbation interference, and the influence weight of the signal quality index can remain unchanged to ensure the continuity of the assessment. If the result is positive, the consistency between the perturbation characteristics and the macroscopic movement pattern is further compared based on the allowable perturbation range to distinguish between normal patient movement variations and abnormal movements inconsistent with the task objectives. Simultaneously, the abnormal characteristics detected by different sensors are compared to see if they corroborate each other. For example, if the surface electromyography (SEMG) signal shows high-frequency noise, and the inertial measurement unit (IMU) data also shows abnormal jitter, it may indicate sensor instability or external mechanical interference. By combining the perturbation consistency comparison results and the abnormality corroboration comparison results, the type of interference can be more accurately determined, such as physiological tremors, sensor displacement, poor electrode contact, or environmental noise. Finally, based on the determined type of interference, the influence weight of the signal quality index on the reliability score of the motor function assessment results is adjusted accordingly.

[0036] In some preferred embodiments, a specific example is given below. Suppose a disabled patient is undergoing elbow flexion-extension rehabilitation training. First, the system collects surface electromyography (EMG) signals from the patient's elbow muscles and data from an inertial measurement unit (IMU) worn on the arm. During training, the system continuously performs spectral analysis on the EMG signals to obtain high-frequency instantaneous energy and performs high-pass filtering on the IMU data to obtain the filtered root mean square (RMS) value. Simultaneously, based on the task context parameter of "elbow flexion-extension," the system pre-determines the macroscopic movement pattern of the action, such as the specific angle range, speed, and rhythm of the elbow joint from full extension to full flexion, and sets an allowable perturbation range, for example, allowing a slight deviation of ±5 degrees in the joint angle.

[0037] If the system detects perturbation characteristics in both the high-frequency instantaneous energy of the surface electromyography (EMG) signal and the root mean square (RMS) value of the inertial measurement unit (IMU) data within the same time window, further analysis will be performed. For example, if the perturbation characteristics are highly consistent with the macroscopic movement pattern, and the abnormal characteristics detected by different sensors (such as multiple EMG sensors or between EMG and IMU sensors) corroborate each other, it may be determined to be physiological tremor occurring when the patient is exerting effort to complete the movement. In this case, the system will appropriately reduce the weight of high-frequency noise indicators in the EMG signal on the reliability score based on the interference type of "physiological tremor," because this tremor is a manifestation of the patient's effort, rather than a pure signal quality issue. Conversely, if the perturbation characteristics are inconsistent with the macroscopic movement pattern, and the abnormal characteristics detected by different sensors are contradictory (e.g., the EMG signal shows high-frequency noise, but the IMU data shows the arm is stationary), it may be determined to be poor electrode contact or external electromagnetic interference. In this case, the system will significantly increase the weight of relevant signal quality indicators (such as signal-to-noise ratio and baseline drift) on the reliability score based on the interference type of "poor electrode contact" or "external electromagnetic interference," to warn that the evaluation results may be severely affected. In this way, the system can dynamically adjust the weights according to the specific type of interference, thereby providing a more accurate reliability score for motion function assessment results.

[0038] Optionally, adjusting the weighting of signal quality metrics on the reliability score of motion function assessment results based on the type of interference includes: Read the current signal quality indicators and interference types; Based on the type of interference, the corresponding impact mode and impact intensity are obtained from the preset interference impact feature library; When only one type of interference is detected, the instantaneous impact weight of each signal quality index on the reliability score at the current moment is calculated based on the impact mode and impact intensity. When multiple interference types are detected to exist simultaneously, the comprehensive impact weight is calculated based on the impact intensity of each interference type and the sensitivity of each signal quality index to each interference type. The instantaneous impact weight or the comprehensive impact weight is used as the impact weight and output.

[0039] Specifically, when performing this step, it is first necessary to read the various signal quality indicators monitored by the system at the current moment, such as the signal-to-noise ratio, baseline drift, and motion artifact level of the surface electromyography signal, as well as the noise level and drift amount of the inertial measurement unit data. At the same time, it is also necessary to obtain the types of interference identified through the above steps, such as motion artifacts, electrode detachment, and power line interference.

[0040] The interference impact feature library can be understood as a pre-established database or model that stores the specific impact patterns and intensities of different interference types (such as motion artifacts, electrode detachment, power line interference, etc.) on various signal quality indicators (such as signal-to-noise ratio, baseline drift, noise level, etc.). The impact pattern describes how the interference alters the characteristics of the signal quality indicators (for example, motion artifacts may lead to an increase in the high-frequency components of surface electromyography signals), while the impact intensity quantifies the degree of this alteration.

[0041] In practical applications, when the system detects only one type of interference, such as motion artifacts, it calculates the instantaneous impact weights of each signal quality index on the reliability score at the current moment, based on the impact pattern and intensity of the motion artifact obtained from the interference impact feature library. These instantaneous impact weights reflect the degree to which a specific signal quality index contributes to the final reliability score under a single interference scenario.

[0042] Furthermore, when the system detects multiple interference types simultaneously, such as motion artifacts and electrode detachment, a comprehensive impact weight needs to be calculated based on the intensity of each interference type and the sensitivity of each signal quality index to each interference type. The sensitivity of each signal quality index to each interference type refers to the degree to which different signal quality indices are affected by different interference types. For example, the signal-to-noise ratio of surface electromyography (EMG) signals may be more sensitive to power line interference, while baseline drift may be more sensitive to electrode detachment. By comprehensively considering these factors, a more comprehensive and accurate comprehensive impact weight can be calculated.

[0043] Ultimately, both the calculated instantaneous impact weight and the comprehensive impact weight will be used as the final impact weight and output for subsequent determination of the reliability score of the motor function assessment results.

[0044] Optionally, when multiple interference types are detected simultaneously, the steps for calculating the comprehensive impact weight based on the influence intensity of each interference type and the sensitivity of each signal quality index to each interference type include: Acquire environmental parameters as well as the movement status and limb posture of disabled patients; Based on environmental parameters and the movement status and limb posture of disabled patients, multiple interference types were identified and filtered to obtain multiple filtered interference types. Read the influence patterns and influence intensities corresponding to the multiple selected interference types, and use them as the independent influence components of each selected interference type; Based on the preset interference interaction rules, determine whether there is a synergistic effect between multiple interference types; Based on the existing synergistic effects, the independent influencing components are modified to obtain the actual influence characteristics under the condition of multiple interferences coexisting; Based on the actual impact characteristics and the sensitivity of each signal quality index to each type of interference, the comprehensive impact weight is calculated and output. When it is determined that there is no synergistic effect, the independent influence components of each interference type are directly determined as the actual influence characteristics under the condition of multiple interference coexistence. Based on the actual influence characteristics and the sensitivity of each signal quality index to each interference type, the comprehensive influence weight is calculated and output.

[0045] Specifically, environmental parameters can be understood as external conditions that affect signal acquisition and transmission, such as ambient background noise levels, electromagnetic interference intensity, temperature, and humidity. The movement status and limb posture of disabled patients refer to their specific physical activities and joint position information during rehabilitation tasks, such as whether they are in a resting state, performing active or passive movements, and the specific angles and relative positions of the joints. These parameters can be acquired by integrating environmental sensors, posture sensors, or visual recognition systems.

[0046] The identification and screening of multiple interference types involves intelligently analyzing and prioritizing various interference sources in the current environment based on acquired environmental parameters and the disabled patient's movement status and limb posture. For example, motion artifacts may be more prominent when the patient is engaged in high-intensity exercise, while electromagnetic interference may become a major problem when near large medical equipment. This identification and screening allows focusing on the interference types that have the greatest impact on signal quality in the current context, avoiding unnecessary computational burdens and improving the targeting of processing.

[0047] In practical applications, independent influencing components refer to the individual impact patterns and intensities of each selected interference type on signal quality indicators, without considering the interactions between interferences. These components can be obtained from a pre-defined interference impact feature library and preliminarily adjusted according to the current context.

[0048] Furthermore, the pre-defined interference interaction rules are a knowledge base or model used to describe the possible interaction patterns between different interference types. For example, some interferences may have a synergistic effect, meaning that when they coexist, their negative impact on signal quality is greater than the sum of their individual effects; while other interferences may have an antagonistic effect, meaning that they partially cancel each other out in terms of negative impact. These rules can be established based on experimental data, expert experience, or machine learning models. The determination of synergistic effect involves analyzing, based on these pre-defined rules, whether there are mutually reinforcing or weakening effects among the selected interference types.

[0049] When a synergistic effect is identified, the correction of independent influence components refers to adjusting the independent influence components of each interference type based on the identified type and intensity of the synergistic effect. For example, if two interferences have a synergistic enhancement effect, their independent influence components will be appropriately amplified to reflect the actual impact under their combined effect. This allows for the acquisition of the actual impact characteristics under conditions of multiple interferences coexisting, which more accurately reflects the comprehensive impact on signal quality under the combined effect of various interferences.

[0050] Finally, based on the actual impact characteristics and the sensitivity of each signal quality index to each interference type, the comprehensive impact weight is calculated and output. When it is determined that there is no synergistic effect, the independent impact components of each interference type are directly identified as the actual impact characteristics under the condition of multiple interference coexistence, and the comprehensive impact weight is calculated based on this.

[0051] Optionally, when multiple interference types are detected simultaneously, the steps for calculating the comprehensive impact weight based on the influence intensity of each interference type and the sensitivity of each signal quality index to each interference type include: When multiple interference types are detected to exist simultaneously, the spatial distribution of each interference type in the rehabilitation area is obtained. Obtain the patient location of the disabled patient; Based on the patient's location, determine the corresponding position of each sensor in the spatial distribution; Based on the corresponding location, calculate the local interference intensity of each sensor; where local interference intensity refers to the intensity value at the corresponding location in the spatial distribution, and adjust it according to the influence intensity of the interference type. Based on the local interference intensity and the sensitivity of each signal quality index to each type of interference, the comprehensive influence weight is calculated and output.

[0052] Specifically, when multiple interference types are detected simultaneously, the first step is to obtain the spatial distribution of each interference type within the rehabilitation area. Spatial distribution can be understood as the intensity or probability distribution map of various interference types at different locations within the rehabilitation area. This can be obtained, for example, through pre-deployed sensor networks, environmental modeling techniques, or historical data analysis. Further, the patient's location needs to be obtained. Patient location refers to the real-time spatial coordinates of the disabled patient within the rehabilitation area, which can be monitored in real-time using indoor positioning systems (e.g., based on ultra-wideband (UWB), visual recognition, or inertial navigation technologies).

[0053] Based on this, the corresponding position of each sensor in the spatial distribution can be determined according to the patient's location. Specifically, since disabled patients typically wear multiple sensors, and the relative positions of these sensors on the patient's limbs are pre-set or calibrated, once the patient's overall position is determined, the specific coordinates of each sensor in the spatial distribution of the rehabilitation area can be calculated. Therefore, based on the corresponding position, the local interference intensity experienced by each sensor is calculated. Local interference intensity refers to the local intensity value obtained by taking the intensity value at the corresponding position in the spatial distribution and adjusting it in conjunction with the influence intensity of the interference type. Specifically, this local intensity reflects the actual degree of interference experienced by a specific sensor at a specific location and under a specific interference type, comprehensively considering the intensity of the interference source, distance attenuation, and the influence of the environmental medium. Finally, based on the local interference intensity and the sensitivity of each signal quality index to each interference type, the comprehensive influence weight is calculated and output.

[0054] Optionally, the steps for calculating and outputting the comprehensive influence weights based on the local interference intensity and the sensitivity of each signal quality index to each interference type include: Obtain contact state parameters and fixation stability parameters; wherein, the contact state parameters include at least one or more of the following: electrode-skin contact impedance, contact pressure, or contact area parameters; the fixation stability parameters include at least one or more of the following: strap tightness, sensor slippage, fixation point displacement, or fixation structure vibration amplitude. The current sensitivity of each sensor to local disturbances is evaluated based on the contact state parameters and fixed stability parameters. Based on the current sensitivity, the sensitivity of each signal quality index to each type of interference is corrected in real time to obtain the corrected sensitivity used for the calculation of the comprehensive influence weight. Calculate and output the overall impact weight based on the local disturbance intensity and correction sensitivity.

[0055] Specifically, contact state parameters refer to the physical or electrical characteristics of the interface between the sensor and the patient's body, which directly affect the signal acquisition quality. For example, the contact impedance between the electrode and the skin is a key parameter for surface electromyography (EMG) signal acquisition; high impedance usually indicates poor contact and susceptibility to external electromagnetic interference. Contact pressure and contact area parameters affect the degree of physical coupling between the sensor and the skin, thus affecting signal stability and signal-to-noise ratio. Fixation stability parameters refer to the degree of secure fixation of the sensor to the patient's limb, which affects the relative displacement and vibration of the sensor during movement. For example, the tightness of the strap directly relates to whether the sensor will slip; the amount of sensor slippage, the displacement of the fixation point, and the vibration amplitude of the fixation structure quantify the mechanical instability that may occur during sensor movement. These instabilities can introduce motion artifacts or cause signal distortion.

[0056] Optionally, the step of evaluating the current sensitivity of each sensor to local disturbances based on contact state parameters and fixed stability parameters includes: Read the contact state parameters and the fixed stability parameters; A time-series analysis is performed on the contact state parameters and fixed stability parameters to identify the change patterns within a preset short time window. The change patterns are the temporal changes of the contact state parameters and fixed stability parameters within the preset short time window, including at least one or more of the following: rapid shaking, periodic loosening, continuous offset, and combinations thereof. Based on the change pattern, obtain sensitivity correction parameters from the preset sensitivity calibration rule library; The preset sensor response model is adjusted in real time based on the sensitivity correction parameters. Based on the adjusted sensor response model, the current sensitivity of each sensor to local disturbances is evaluated and output.

[0057] Specifically, reading contact state parameters and fixation stability parameters refers to the system acquiring in real time contact state parameters such as contact impedance, contact pressure, and contact area between the sensor and the patient's skin, as well as fixation stability parameters such as strap tightness, sensor slippage, fixation point displacement, and vibration amplitude of the fixation structure. These parameters are key indicators reflecting the sensor's working status and the degree of interference.

[0058] The temporal analysis of contact state parameters and fixed stability parameters to identify change patterns within a preset short period can be understood as processing the data of these parameters within a continuous time window, such as using methods like moving average, Fourier transform, or wavelet analysis, to detect their trends and characteristics over time. Change patterns refer to the specific dynamic behaviors exhibited by these parameters within a short time window. For example, rapid jitter may indicate a momentary impact on the sensor or rapid movement of the patient's limb; periodic loosening may indicate gradual loosening of the strap or periodic pressure changes during repetitive movements by the patient; and persistent offset may mean a slow but continuous displacement of the sensor position. The purpose of identifying these patterns is to capture the dynamic changes in the sensor state, providing a basis for subsequent sensitivity correction.

[0059] In practical applications, retrieving sensitivity correction parameters from a pre-defined sensitivity calibration rule base based on the change pattern means that the system searches for the corresponding correction parameters in the pre-established rule base according to the identified specific change pattern (such as rapid jitter, periodic loosening, or continuous shift). This rule base stores empirical or model-based data on how sensor sensitivity should be adjusted under different change patterns. For example, when a "rapid jitter" pattern is detected, it may be necessary to increase the sensitivity correction parameters for high-frequency interference; when a "periodic loosening" pattern is detected, it may be necessary to adjust the sensitivity correction parameters for low-frequency or specific-frequency interference.

[0060] Furthermore, adjusting the preset sensor response model in real time based on the sensitivity correction parameters refers to dynamically modifying the internal parameters or structure of the sensor response model using the acquired correction parameters. A sensor response model is a mathematical model describing how a sensor responds to external stimuli and disturbances, aiming to accurately simulate the sensor's output characteristics under different operating conditions. By adjusting this model in real time, it can better reflect the sensor's true performance under current contact conditions and fixed stability.

[0061] Therefore, based on the adjusted sensor response model, assessing and outputting the current sensitivity of each sensor to local interference means using the updated model, after real-time adjustments, to calculate or predict the sensor's sensitivity under the current local interference environment. This assessment is dynamic and real-time, and can more accurately reflect the sensor's resistance or response capability to various interferences in actual use.

[0062] Optionally, the step of performing time-series analysis on contact state parameters and fixed stability parameters to identify change patterns within a preset short period includes: The contact state parameters and fixed stability parameters are decomposed into signals over a continuous time window to separate components in different frequency ranges. Based on the components in different frequency ranges, calculate the energy distribution and instantaneous phase change in each frequency range; By comparing the fluctuation characteristics of energy distribution and instantaneous phase change within a continuous time window, the change pattern is identified and output.

[0063] The continuous-time-window signal decomposition of contact state parameters and fixation stability parameters refers to processing the time-series data of contact state parameters (such as electrode-skin contact impedance, contact pressure, or contact area parameters) and fixation stability parameters (such as strap tightness, sensor slippage, fixation point displacement, or vibration amplitude of the fixation structure) collected by the sensor within a preset continuous-time window. Specifically, signal processing techniques such as Short-Time Fourier Transform (STFT), Wavelet Transform, or Empirical Mode Decomposition (EMD) can be used to decompose the original complex time-series signal into multiple components in different frequency ranges. For example, low-frequency components may reflect slow, trend-like changes, while high-frequency components may correspond to rapid, instantaneous disturbances or noise.

[0064] Furthermore, based on the components within different frequency ranges, the energy distribution and instantaneous phase change of each frequency range are calculated. Energy distribution refers to the intensity or power of the signal energy within the frequency band corresponding to each frequency component. This helps quantify the activity level of a specific frequency component in the signal. Instantaneous phase change describes how the phase of the signal changes over time at a specific frequency component, which is crucial for capturing the instantaneous synchronization or delay characteristics of the signal. For example, the energy distribution can be obtained by calculating the sum of squares of the instantaneous amplitude of each frequency component, and instantaneous phase information can be extracted using methods such as the Hilbert transform.

[0065] Therefore, by comparing the fluctuation characteristics of energy distribution and instantaneous phase changes within a continuous time window, change patterns can be identified and output. Fluctuation characteristics refer to the dynamic behavior of energy distribution and instantaneous phase changes within a continuous time window, including their amplitude, frequency, duration, and trend. For example, if the energy of high-frequency components fluctuates violently and irregularly within a short period, it may indicate rapid jitter; if the energy or phase of a certain frequency component shows periodic increases or decreases, it may indicate periodic loosening; and if the energy or phase of low-frequency components continuously shifts in a certain direction, it may indicate persistent shift. Using a pre-defined pattern recognition algorithm (e.g., threshold-based judgment, machine learning classifiers, or pattern matching), specific contact states or fixed stability change patterns can be identified and output based on these fluctuation characteristics.

[0066] Optionally, the step of identifying and outputting the change pattern by comparing the fluctuation characteristics of energy distribution and instantaneous phase change within a continuous time window includes: Obtain the spectral characteristics of ambient background noise; Based on the spectral characteristics of ambient background noise, noise suppression processing is performed on energy distribution and instantaneous phase changes; Statistical features were extracted from the energy distribution and instantaneous phase changes after noise suppression; the statistical features included mean, variance, kurtosis, and skewness. Obtain the time trend of statistical features; Based on the time-varying trend of statistical characteristics, distinguish between real changes and noise fluctuations; Based on the differentiation results, the change patterns are identified and output; the change patterns include rapid shaking, periodic loosening, continuous shift, and combinations of rapid shaking, periodic loosening, and continuous shift.

[0067] Specifically, before identifying the change pattern, it is first necessary to obtain the spectral characteristics of the ambient background noise. The spectral characteristics of ambient background noise can be understood as the energy distribution of the ambient noise signal at different frequencies when there is no patient movement or significant sensor changes. The purpose is to provide baseline information for subsequent noise suppression processing. For example, this can be obtained by pre-collecting idle signals within the rehabilitation area for a period of time, or by collecting signals while the patient is stationary, and then using spectral analysis methods such as Fourier transform.

[0068] Furthermore, based on the spectral characteristics of the environmental background noise, noise suppression processing is applied to the energy distribution and instantaneous phase changes. Noise suppression processing refers to using digital signal processing techniques, such as adaptive filtering, wavelet denoising, or spectral subtraction, to remove or weaken noise components from the energy distribution and instantaneous phase change signals decomposed from the contact state parameters and fixed stability parameters. The aim is to improve the signal-to-noise ratio, making subsequent extraction of signal features more accurate.

[0069] After noise suppression, statistical features are extracted from the noise-suppressed energy distribution and instantaneous phase changes. These statistical features include mean, variance, kurtosis, and skewness. The mean reflects the average level of the signal; variance reflects the degree of signal fluctuation; kurtosis describes the sharpness of the signal distribution and can be used to identify abnormal peaks or flat regions; skewness describes the symmetry of the signal distribution and can reveal the tendency of signal changes. These statistical features can quantify the intrinsic characteristics of the signal from different dimensions, providing a quantitative basis for distinguishing between real changes and noise fluctuations.

[0070] Subsequently, the temporal trends of the statistical characteristics are obtained. Temporal trends refer to the regular changes in the aforementioned statistical characteristics such as mean, variance, kurtosis, and skewness over a continuous time window. For example, these trends can be captured using methods such as sliding window averaging, regression analysis, or differencing. The aim is to identify continuous, regular signal changes, rather than instantaneous or random noise fluctuations.

[0071] Based on the time-varying trends of statistical features, a distinction can be made between real changes and noise fluctuations. Specifically, methods such as setting thresholds, pattern matching, or machine learning classifiers can be used to identify statistical features with specific time-varying trends as real changes, while fluctuations that do not conform to these trends and have strong randomness can be identified as noise. For example, a sustained mean shift may indicate a sustained shift, while fluctuations with high variance and no obvious trend are more likely to be noise.

[0072] Finally, based on the differentiation results, the change patterns are identified and output. Change patterns include rapid jitter, periodic loosening, persistent shift, and combinations of these three patterns. For example, if statistical characteristics show high-frequency, short-term, and large-amplitude fluctuations, it may be identified as rapid jitter; if they show regular, recurring fluctuations, it may be identified as periodic loosening; and if they show slow but persistent changes in the mean or variance, it may be identified as persistent shift.

[0073] This application also discloses a signal pattern recognition system for assessing motor function in disabled patients, combined with... Figure 3 As shown, the signal pattern recognition system 1 for assessing motor function in disabled patients includes: The motion signal acquisition module 11 is used to acquire motion signals from disabled patients; the motion signals include surface electromyography signals and inertial measurement unit data. The quality index acquisition module 12 is used to perform quality inspection processing on motion signals in order to obtain the signal quality index corresponding to the motion signal. Task context recognition module 13 is used to identify the task context parameters of the current rehabilitation task of the disabled patient; The motor function assessment module 14 is used to perform motor function assessment based on motion signals and obtain motor function assessment results. The influence weight adjustment module 15 is used to adjust the influence weight of the signal quality index on the reliability score of the motion function assessment result according to the task context parameters. The reliability score determination module 16 is used to determine the reliability score of the motion function assessment results based on the adjusted influence weights and signal quality indicators. The evaluation results display module 17 is used to display the reliability score and motor function evaluation results in a graphical interface.

[0074] This system aims to address the challenges of traditional assessment systems facing multiple dynamic interferences in complex, non-standardized environments. By integrating modules for motion signal acquisition, quality indicator acquisition, task context identification, motor function assessment, influence weight adjustment, reliability score determination, and assessment result display, this system can work collaboratively to accurately assess the motor function of disabled patients. Specifically, the system first acquires the patient's motion signals through the motion signal acquisition module, followed by quality indicator acquisition module quality control to quantify signal quality. The task context identification module identifies the background information of the current rehabilitation task. After the initial motor function assessment results are generated, the influence weight adjustment module dynamically adjusts the influence weight of the signal quality indicators on the reliability score based on task context parameters, enabling the reliability score determination module to generate a more realistic reliability score. Finally, the assessment result display module presents the assessment results and reliability score intuitively to the rehabilitation therapist. This modular design and dynamic adjustment mechanism allow the system to adaptively cope with various interferences, significantly improving the accuracy and reliability of the assessment results.

[0075] The methods for acquiring motion signals have already been described in the above embodiments and will not be repeated here. It is important to emphasize that the motion signal acquisition module can be configured to include multiple sensor units, such as surface electromyography (EMG) sensors and inertial measurement unit (IMU) sensors. These sensor units can be designed as independent hardware devices, connected to the main processing unit via wired or wireless means (e.g., Bluetooth, Wi-Fi). As one implementation, this module can be a hardware circuit integrated into a wearable device, responsible for converting raw analog or digital signals into a data stream that the system can process. For example, the sensor can simply transmit the raw data to the processing unit via a serial interface without any preprocessing.

[0076] The methods for quality control processing of motion signals to obtain signal quality indicators have been described in the above embodiments and will not be repeated here. It is important to emphasize that the quality indicator acquisition module can be implemented as a software component running on the processing unit to receive the raw motion signal output by the motion signal acquisition module. This module can contain a series of preset algorithms, such as detecting signal saturation or clipping through simple threshold judgment, or evaluating baseline drift by calculating the signal's mean and standard deviation. In some implementations, this module may only provide basic signal statistics without performing complex spectral analysis or pattern recognition.

[0077] The methods for identifying task context parameters have already been described in the above embodiments and will not be repeated here. It is important to emphasize that the task context recognition module can be configured to receive data from a user interface or auxiliary sensors. For example, this module can provide a simple input interface, allowing therapists to manually select the current rehabilitation task type and difficulty. Furthermore, this module can also integrate a simple environmental sensor interface to acquire basic environmental parameters such as ambient light or temperature. In some implementations, this module can perform context judgment based solely on preset rules, without involving complex machine learning models.

[0078] The above embodiments have already described the methods for performing motor function assessment based on motion signals, and will not be repeated here. It is important to emphasize that the motor function assessment module can be implemented as a software module containing an algorithm library for processing motion signals. For example, this module can use simple threshold comparisons or linear regression models to calculate joint range of motion or muscle strength. For gait analysis, this module can perform basic step counting and cadence calculation based on inertial measurement unit data. In some implementations, this module may only provide preliminary, uncalibrated assessment results.

[0079] The above embodiments have already described the method of adjusting the influence weights according to task context parameters, and will not be repeated here. It is important to emphasize that the influence weight adjustment module can be implemented as a software component, whose core function is to modify preset weight values ​​based on task context parameters. For example, this module can contain a simple lookup table that directly reads the corresponding weight set based on different task context parameters (such as "walking training" or "fine motor skills"). In some implementations, this module can only provide static weight adjustment, without considering real-time changes in signal quality indicators or interference types.

[0080] The above embodiments have already described the method for determining the reliability score based on the adjusted influence weights and signal quality indicators, and will not be repeated here. It is important to emphasize that the reliability score determination module can be implemented as a calculation unit to receive the adjusted influence weights and each signal quality indicator. This module can perform a simple weighted average calculation, multiplying each signal quality indicator by its corresponding influence weight, and then summing all products to obtain the final reliability score. For example, this module can simply add all weighted indicator values ​​and normalize them to a preset range.

[0081] The methods for displaying reliability scores and motor function assessment results using a graphical interface have already been described in the above embodiments, and will not be repeated here. It is important to emphasize that the assessment result display module can be implemented as a user interface component, responsible for presenting the assessment results and reliability scores to the user in an intuitive way. For example, this module can generate a simple text report listing various assessment indicators and reliability scores. In some implementations, this module may only provide basic numerical displays, without including complex charts or dynamic visualizations.

[0082] The signal pattern recognition system for assessing motor function in disabled patients disclosed in this application aims to solve the problem of multiple dynamic interferences faced by existing assessment systems in complex and non-standardized environments. Traditional systems experience significant performance degradation when processing damaged, incomplete, or degraded signals, leading to unstable and unreliable assessment results, which seriously affects rehabilitation therapists' formulation and adjustment of rehabilitation plans.

[0083] This system dynamically adjusts the weighting of signal quality indicators by introducing a task context recognition module and an influence weight adjustment module. Specifically, the system can intelligently weigh the importance of different signal quality issues based on the characteristics of the current rehabilitation task, thereby providing more realistic reliability scores and motor function assessment results. For example, in fine motor training, the system will adjust the weights to focus more on the signal-to-noise ratio of surface electromyography (EMG) signals; while in high-amplitude movement training, it may focus more on the integrity of inertial measurement unit (IMU) data. This modular design and dynamic adjustment mechanism allows the system to adaptively cope with various interferences, significantly improving the accuracy and reliability of the assessment results. Compared with existing technologies, this system can understand the real-time status of assessment data, avoiding misjudgments caused by signal interference, thereby improving the scientific rigor and effectiveness of rehabilitation program development and adjustment, and enhancing rehabilitation therapists' trust in the intelligent assessment data.

[0084] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A signal pattern recognition method for assessing motor function in disabled patients, characterized in that, include: The motion signals of disabled patients are collected; the motion signals include surface electromyography signals and inertial measurement unit data; The inertial measurement unit data includes at least three-axis acceleration data, three-axis angular velocity data, and attitude angle data; The motion signal is subjected to quality inspection processing to obtain the signal quality index corresponding to the motion signal; Identify the task context parameters of the current rehabilitation task for disabled patients; Based on the motion signals, a motor function assessment is performed to obtain the motor function assessment results; Based on the task context parameters, adjust the weight of the signal quality index on the reliability score of the motor function assessment result; Based on the adjusted influence weights and the signal quality index, the reliability score of the motion function assessment result is determined; The reliability score and the motion function evaluation results are displayed in a graphical interface.

2. The signal pattern recognition method for assessing motor function in disabled patients according to claim 1, characterized in that, The step of adjusting the influence weight of the signal quality index on the reliability score of the motor function assessment result according to the task context parameters includes: Spectral analysis was performed on the surface electromyography signal to obtain the instantaneous energy in the high-frequency band; The inertial measurement unit data is subjected to high-pass filtering to obtain the filtered root mean square value; Based on the task context parameters, determine the macroscopic movement pattern and the allowable micro-perturbation range corresponding to the current rehabilitation task; the macroscopic movement pattern is a templated description of the target movement trajectory and rhythm corresponding to the current rehabilitation task, including at least the main joint movement direction, movement amplitude envelope and movement rhythm parameters; the allowable micro-perturbation range is the allowable small-amplitude perturbation interval relative to the macroscopic movement pattern. Determine whether the instantaneous energy of the high-frequency band and the root mean square value exhibit perturbation characteristics within the same time window, and obtain the perturbation characteristic display judgment result; When the perturbation feature shows a negative result, the interference type is determined to be no significant perturbation interference, and the influence weight of the signal quality index on the reliability score of the motion function evaluation result is maintained at the weight preset in the previous period. When the perturbation feature shows a yes result, the consistency between the perturbation feature and the macroscopic action pattern is compared based on the allowable perturbation range to obtain a perturbation consistency comparison result. When the perturbation feature shows a judgment result of yes, compare whether the abnormal features detected by different sensors corroborate each other to obtain the abnormal verification comparison result; The type of interference is determined based on the perturbation consistency comparison results and the anomaly confirmation comparison results; Based on the type of interference, adjust the weighting of the signal quality index on the reliability score of the motion function assessment result.

3. The signal pattern recognition method for assessing motor function in disabled patients according to claim 2, characterized in that, The step of adjusting the influence weight of the signal quality index on the reliability score of the motion function assessment result according to the type of interference includes: Read the signal quality indicators and interference types at the current moment; Based on the type of interference, the corresponding influence mode and influence intensity are obtained from a preset interference influence feature library; When only one type of interference is detected, the instantaneous impact weight of each signal quality index on the reliability score at the current moment is calculated based on the impact mode and impact intensity. When multiple interference types are detected to exist simultaneously, the comprehensive impact weight is calculated based on the impact intensity of each interference type and the sensitivity of each signal quality index to each interference type. The instantaneous influence weight or the comprehensive influence weight is used as the influence weight and output.

4. The signal pattern recognition method for assessing motor function in disabled patients according to claim 3, characterized in that, When multiple interference types are detected simultaneously, the step of calculating the comprehensive influence weight based on the influence intensity of each interference type and the sensitivity of each signal quality index to each interference type includes: Acquire environmental parameters as well as the movement status and limb posture of disabled patients; Based on the environmental parameters and the movement status and limb posture of the disabled patient, multiple interference types are identified and filtered to obtain multiple filtered interference types. Read the influence patterns and influence intensities corresponding to the multiple selected interference types, and use them as the independent influence components of each selected interference type; Based on the preset interference interaction rules, determine whether there is a synergistic effect between multiple interference types; Based on the existing synergistic effects, the independent influencing components are modified to obtain the actual influence characteristics under the condition of multiple interferences coexisting; Based on the actual impact characteristics and the sensitivity of each signal quality index to each type of interference, the comprehensive impact weight is calculated and output. When it is determined that there is no synergistic effect, the independent influence components of each interference type are directly determined as the actual influence characteristics corresponding to the coexistence of multiple interferences. Based on the actual influence characteristics and the sensitivity of each signal quality index to each interference type, the comprehensive influence weight is calculated and output.

5. The signal pattern recognition method for assessing motor function in disabled patients according to claim 3, characterized in that, When multiple interference types are detected simultaneously, the step of calculating the comprehensive influence weight based on the influence intensity of each interference type and the sensitivity of each signal quality index to each interference type includes: When multiple interference types are detected to exist simultaneously, the spatial distribution of each interference type in the rehabilitation area is obtained. Obtain the patient location of the disabled patient; Based on the patient's location, determine the corresponding position of each sensor in the spatial distribution; Based on the corresponding location, the local interference intensity of each sensor is calculated; wherein, the local interference intensity refers to the intensity value at the corresponding location in the spatial distribution, and is adjusted in combination with the influence intensity of the interference type to obtain the local intensity quantity; Based on the local interference intensity and the sensitivity of each signal quality index to each type of interference, the comprehensive influence weight is calculated and output.

6. The signal pattern recognition method for assessing motor function in disabled patients according to claim 5, characterized in that, The step of calculating and outputting the comprehensive influence weight based on the local interference intensity and the sensitivity of each signal quality index to each interference type includes: The contact state parameters and fixation stability parameters are obtained; wherein, the contact state parameters include at least one or more of the following: electrode-skin contact impedance, contact pressure, or contact area parameters; the fixation stability parameters include at least one or more of the following: strap tightness, sensor slippage, fixation point displacement, or fixation structure vibration amplitude. Based on the contact state parameters and the fixed stability parameters, evaluate the current sensitivity of each sensor to local disturbances; Based on the current sensitivity, the sensitivity of each signal quality index to each type of interference is corrected in real time to obtain the corrected sensitivity used for the comprehensive influence weight calculation. Based on the local disturbance intensity and the correction sensitivity, the comprehensive influence weight is calculated and output.

7. The signal pattern recognition method for assessing motor function in disabled patients according to claim 6, characterized in that, The step of evaluating the current sensitivity of each sensor to local disturbances based on the contact state parameters and the fixed stability parameters includes: Read the contact state parameters and the fixed stability parameters; A time-series analysis is performed on the contact state parameters and the fixed stability parameters to identify change patterns within a preset short time window; the change pattern is the temporal change pattern of the contact state parameters and the fixed stability parameters within the preset short time window, including at least one or more of rapid jitter, periodic loosening, continuous offset, and combinations thereof; Based on the change pattern, sensitivity correction parameters are obtained from a preset sensitivity calibration rule library; The preset sensor response model is adjusted in real time based on the sensitivity correction parameters. Based on the adjusted sensor response model, the current sensitivity of each sensor to local disturbances is evaluated and output.

8. The signal pattern recognition method for assessing motor function in disabled patients according to claim 7, characterized in that, The step of performing time-series analysis on the contact state parameters and the fixed stability parameters to identify change patterns within a preset short period includes: The contact state parameters and the fixed stability parameters are decomposed into signals over a continuous time window to separate components in different frequency ranges. Based on the components in different frequency ranges, calculate the energy distribution and instantaneous phase change in each frequency range; By comparing the fluctuation characteristics of the energy distribution and the instantaneous phase change within a continuous time window, the change pattern is identified and output.

9. The signal pattern recognition method for assessing motor function in disabled patients according to claim 8, characterized in that, The step of identifying and outputting the change pattern by comparing the fluctuation characteristics of the energy distribution and the instantaneous phase change within a continuous time window includes: Obtain the spectral characteristics of ambient background noise; Based on the spectral characteristics of the ambient background noise, noise suppression processing is applied to the energy distribution and the instantaneous phase change; Statistical features are extracted from the energy distribution and instantaneous phase changes after noise suppression; the statistical features include mean, variance, kurtosis, and skewness. Obtain the time trend of the statistical features; Based on the time-varying trend of the statistical characteristics, distinguish between real changes and noise fluctuations; Based on the differentiation results, the change patterns are identified and output; the change patterns include rapid shaking, periodic loosening, continuous shift, and combinations of rapid shaking, periodic loosening, and continuous shift.

10. A signal pattern recognition system for assessing motor function in disabled patients, characterized in that, include: A motion signal acquisition module is used to acquire motion signals from disabled patients; the motion signals include surface electromyography signals and inertial measurement unit data. The quality index acquisition module is used to perform quality inspection processing on the motion signal to obtain the signal quality index corresponding to the motion signal. The task context recognition module is used to identify the task context parameters of the current rehabilitation task for disabled patients; A motor function assessment module is used to perform a motor function assessment based on the motor signal and obtain a motor function assessment result. The influence weight adjustment module is used to adjust the influence weight of the signal quality index on the reliability score of the motion function evaluation result according to the task context parameters. The reliability score determination module is used to determine the reliability score of the motion function evaluation result based on the adjusted influence weights and the signal quality index. The evaluation result display module is used to display the reliability score and the motion function evaluation result in a graphical interface.

Images