Neural rehabilitation progress prediction method based on multimode data fusion

Abstract

The invention relates to the technical field of rehabilitation training, in particular to a neural rehabilitation progress prediction method based on multimode data fusion. According to the method, the intensity adaptation factor is analyzed through the respiratory rate increase situation, and the suitability between the training intensity and the physiological tolerance of the patient is analyzed; based on the myoelectricity value periodic change and the joint angle deviation degree, the muscle fatigue degree and the motion nonstandard degree are quantified, and the execution quality of each training item is obtained; and in combination with a training effect index obtained by the strength adaptation factor, analyzing change trend mechanical energy prediction. Through multi-dimensional data collaborative analysis, trend prediction is carried out on more comprehensive and objective quantitative evaluation of the neural rehabilitation process, the rehabilitation management accuracy is effectively improved, and reliable data support is provided.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation training technology, and specifically to a method for predicting the progress of neurorehabilitation based on multimodal data fusion. Background Technology

[0002] Brain injury patients require regular rehabilitation training to stimulate the plasticity of the nervous system, helping the brain and nerves re-establish connections and thus improve motor, sensory, and cognitive functions. For example, stroke patients can gradually regain limb motor abilities by repeatedly practicing specific movements. Neurorehabilitation training plays a crucial role in restoring brain and nervous system function; through specific training and treatment methods, it can help promote neural remodeling and restore damaged functions.

[0003] During rehabilitation, a patient's progress may be gradual and subtle. To accurately assess the actual rehabilitation effect during each session, it is common practice to combine physiological data analysis, such as respiratory rate analysis. However, neurorehabilitation is complex, and relying on a single data source cannot fully reflect the complexity and diversity of rehabilitation progress. It may overlook the patient's progress in muscle function, physical activity, and other aspects during training, thus reducing the accuracy of predictions. Summary of the Invention

[0004] To address the technical problem that existing technologies relying solely on a single data source cannot fully reflect the complexity and diversity of rehabilitation progress, and may overlook the rehabilitation progress of patients' muscles and physical activity during training, thus reducing the accuracy of prediction results, the present invention aims to provide a method for predicting neurorehabilitation progress based on multi-modal data fusion. The specific technical solution adopted is as follows: This invention provides a method for predicting the progression of neurorehabilitation based on multimodal data fusion, the method comprising: Acquire multi-dimensional time-series data for each training item in each historical rehabilitation training process of the patient; the multi-dimensional time-series data includes respiratory rate time-series data, electromyography value time-series data, and reference angle deviation time-series data at each joint; By analyzing the growth trend of respiratory rate time-series data, the patient's physiological adaptability to each training item in a single rehabilitation training session is obtained, and the intensity adaptation factor in each rehabilitation training session is obtained. Muscle fatigue was analyzed based on the periodic changes in electromyography (EMG) values ​​in each training exercise, and the degree of non-standard movement was analyzed based on the deviation in the time series data of the baseline angle deviation at each joint, thus determining the training quality indicators for each training exercise. By combining the intensity adaptation factor and training quality indicators in each rehabilitation training session, the training effect indicators for each rehabilitation training session are obtained. Predictions are made based on the changing trends of historical rehabilitation training effectiveness indicators.

[0005] Furthermore, the method for obtaining the intensity adaptation factor includes: For any rehabilitation training program, the time of increase is selected based on the increase of respiratory rate values ​​between adjacent time points in the respiratory rate time sequence data of that training program. The adaptability of the training program is determined based on the degree of growth trend and the degree of stability and continuity of growth at each growth point. Each training item is weighted according to its order in the rehabilitation training, and the intensity adaptation factor of the rehabilitation training is obtained by weighting and combining the adaptability of all training items.

[0006] Furthermore, the method for obtaining the growth moment includes: For any given moment, the rate of increase at that moment is obtained based on the numerical difference between the respiratory rate data at the next moment and the moment itself; when the rate of increase is positive, the corresponding moment is recorded as the time of increase.

[0007] Furthermore, the method for acquiring the adaptability includes: Obtain the slope of the respiratory rate time series data at each growth moment; combine the proportion and slope distribution of all growth moments in this training project to obtain the continuous growth adaptation index of this training project. The slope is negatively correlated with the continuous growth adaptation index. Based on the maximum value of the growth rate at each growth moment in the training project, the maximum growth interval is divided; based on the distribution stability of all maximum growth intervals in the training project and the continuous growth index, the adaptability of the training project is obtained.

[0008] Furthermore, the method for obtaining the training quality metrics includes: For any training program in any rehabilitation training session, the muscle fatigue level of that training program can be obtained based on the periodic activation stability and signal deviation changes of the electromyography (EMG) data in the time series data. The non-standard moments are determined by the magnitude of the time series data of the reference angle deviation at different joints, and the distribution of non-standard moments and the degree of deviation of the corresponding different joints are analyzed to obtain the non-standard degree of the movement in this training item. By combining muscle fatigue and movement irregularity in this training program, training quality indicators were obtained. Both muscle fatigue and movement irregularity were negatively correlated with the training quality indicators.

[0009] Furthermore, the method for obtaining muscle fatigue includes: The exercise cycle was divided by adjacent peak values ​​of the electromyography (EMG) time series data in this training project; the average peak value in each exercise cycle was used as the activation index for each exercise cycle. The muscle fatigue level of the training program is obtained based on the degree of deviation of activation indicators at the beginning and end of the exercise cycle, as well as the degree of deviation of electromyography (EMG) data at the beginning and end.

[0010] Furthermore, the method for obtaining the degree of non-standardization of the action includes: Based on the magnitude of the reference angle deviation at each joint at each moment in the training project, deviating joints are screened out; based on the distribution of the number of deviating joints, non-standard moments are determined; and the time period composed of adjacent non-standard moments is taken as the non-standard time period. In each non-standard time period, the magnitude of the deviation of the reference angle from the joint at each non-standard time point and the distribution ratio of the deviation from the joint are analyzed to obtain the standardization difference degree of each non-standard time period. By combining the difference in standardization and the proportion of non-standardization periods in all non-standardization periods in the training program, the degree of non-standardization of the movements in the training program is obtained.

[0011] Furthermore, the method for obtaining the training performance metrics includes: For any given rehabilitation training session, reference training items are selected based on the magnitude of the training quality indicators; the effective quality score is obtained by combining the magnitude of the training quality indicators of the training items in the rehabilitation training session with the number of reference training items. By combining the effective quality score and intensity adaptation factor, the effective indicators of this rehabilitation training are obtained.

[0012] Furthermore, the prediction of the trend of training effect indicators based on historical rehabilitation training includes: By analyzing the trend change of effective training indicators of continuous rehabilitation training within a preset time period before the current moment, and combining the differences in effective training indicators between the first and last rehabilitation training sessions within the preset time period, the predictability of the training progress rate at the current moment is obtained.

[0013] Furthermore, the method for obtaining the reference angle deviation time series data includes: At any joint, the difference between the angle data collected by the inertial sensor and the standard angle data is used to determine the reference angle deviation data at each moment. The reference angle deviation data in the time sequence is used as the reference angle deviation time sequence data at that joint.

[0014] The present invention has the following beneficial effects: This invention comprehensively analyzes the patient's physiological state and physical activity capacity based on respiratory rate, electromyography (EMG) values, and joint reference angle deviations. By analyzing the increase in respiratory rate, it identifies intensity adaptation factors and analyzes the fit between training intensity and the patient's physiological tolerance. Based on the periodic changes in EMG values ​​and the degree of joint angle deviation, it quantifies muscle fatigue and movement irregularities, ensuring accurate assessment of the patient's motor ability during training and reflecting the execution quality of each training program. Furthermore, by combining intensity adaptation factors and training quality indicators, it obtains training effect indicators to characterize the actual value of each training session. This allows the trend to more accurately reflect the actual training effect of rehabilitation progress and predict subsequent rehabilitation progress. Through multi-dimensional data collaborative analysis, this invention provides a more comprehensive and objective quantitative assessment of the neurorehabilitation process and trend prediction, effectively improving the accuracy of rehabilitation management and providing reliable data support. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a method for predicting the progression of neurorehabilitation based on multimodal data fusion, provided in one embodiment of the present invention; Figure 2 A flowchart illustrating a method for obtaining an intensity adaptation factor according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating a method for obtaining training quality metrics according to an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for predicting the progression of neurorehabilitation based on multimodal data fusion proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] The following description, in conjunction with the accompanying drawings, details a specific scheme for a method for predicting the progression of neurorehabilitation based on multimodal data fusion provided by the present invention.

[0020] Please see Figure 1 The diagram illustrates a flowchart of a method for predicting the progression of neurorehabilitation based on multimodal data fusion, according to an embodiment of the present invention. The method includes the following steps: S1: Obtain multi-dimensional time-series data for each training item in each historical rehabilitation training process of the patient; the multi-dimensional time-series data includes respiratory rate time-series data, electromyography value time-series data, and reference angle deviation time-series data at each joint.

[0021] The rehabilitation training cycle for patients with brain injuries typically lasts 6-12 weeks, with 5 training sessions per week, each lasting 30-45 minutes. Each rehabilitation training session usually includes multiple training programs to achieve comprehensive functional recovery. Because the patient's training intensity, compliance, and neurological state will change over time during the rehabilitation process, it is necessary to dynamically track the patient's training progress.

[0022] In this embodiment of the invention, patients can collect respiratory rate data during training by wearing a smartwatch. A surface electromyography (iEMG) sensor is used to record the time sequence of the patient's electromyography signals during each training session. The electromyography signals reflect the total intensity of muscle activity and are commonly used in muscle fatigue, motor control, and biofeedback research to assess the state and activity of muscles.

[0023] To provide more accurate feedback on the standardization of rehabilitation movements, in this embodiment of the invention, inertial sensors are used to collect the angular timing data of each joint, including the elbow and knee joints, during each rehabilitation training session. Simultaneously, standard angular timing data for each joint under each training exercise is also collected. Specifically, the patient completes a standard demonstration under the guidance of a rehabilitation therapist, and the angular data of each joint in the demonstration are recorded as the standard angular timing data for that patient's current training session, serving as a comparison benchmark.

[0024] Then, the difference between the angle data and the standard angle data at each time point is normalized to obtain the reference angle deviation data at each time point. The time-series reference angle deviation data is used as the time-series reference angle deviation data at the corresponding joint. It should be noted that normalization is a well-known technique in the art. The choice of normalization can be standardization of decimal scaling, linear normalization, or standard normalization, etc. For example, z-fractional standardization is used. The specific normalization method is not limited here.

[0025] It is understandable that all monitored data undergo preprocessing, including data standardization and time-scale normalization, to facilitate unified data analysis and remove the influence of units. It should be noted that data preprocessing is a well-known technique in the field, and the sampling frequency can be adjusted by the implementer; therefore, it will not be elaborated upon or restricted here.

[0026] S2: By analyzing the growth trend of respiratory rate time series data, the patient's physiological adaptability to each training item in a single rehabilitation training session is obtained, thus obtaining the intensity adaptation factor in each rehabilitation training session.

[0027] During rehabilitation training, changes in respiratory rate can reflect a patient's physiological load, exercise intensity, and physical condition. For example, a patient's respiratory rate usually fluctuates with changes in training intensity. When training intensity increases, the respiratory rate will increase accordingly, as the body needs more oxygen to maintain higher exercise intensity. Therefore, the suitability of training intensity to the patient's physical condition can be judged by the changes in the patient's respiratory rate during each training session.

[0028] Since respiratory rate is affected by a variety of factors during training, such as coughing, emotional stress, and short rests, not all fluctuations are related to training intensity. Therefore, by analyzing the increase in respiratory rate caused by training intensity, invalid fluctuations that are not related to training intensity can be excluded.

[0029] Preferably, in this embodiment of the invention, the method for obtaining the intensity adaptation factor is described in [reference needed]. Figure 2 The diagram illustrates a flowchart of a method for obtaining an intensity adaptation factor according to an embodiment of the present invention, the method comprising the following steps: S201: For any rehabilitation training program, based on the increase in respiratory rate values ​​between adjacent time points in the respiratory rate time series data of the training program, filter the time points where the increase occurs.

[0030] An analysis of individual training sessions within a single rehabilitation training program is conducted to initially screen for increases in respiratory rate. In this embodiment of the invention, for any given moment, the degree of increase at that moment is obtained based on the numerical difference between the respiratory rate data at the next moment and the moment itself. When the degree of increase is positive, the corresponding moment is recorded as the moment of increase. The moment of respiratory increase reflects the moment when the physiological state is significantly affected by the training intensity.

[0031] S202: Determine the adaptability of the training program based on the degree of growth trend and the degree of stable and continuous growth at each growth point.

[0032] During rehabilitation training, the ideal respiratory rate should increase slowly as the training progresses. If a rapid and unstable increase occurs, it is highly likely that the patient is unable to adapt to the abnormal fluctuations in physiological state caused by the training, which will lead to an unsatisfactory training effect.

[0033] Therefore, in this embodiment of the invention, the slope of the respiratory rate time-series data at each growth moment is obtained to measure the rate of respiratory rate increase. The smaller the slope, the more stable the growth, and the better the patient can withstand the current training load. Thus, by combining the proportion and slope distribution of all growth moments in the training program, a continuous growth adaptation index for the training program is obtained, and the slope is negatively correlated with the continuous growth adaptation index.

[0034] In one specific embodiment of the invention, the average slope of all growth moments is negatively correlated to obtain the growth smoothness; the smaller the overall slope, the smoother and more stable the patient's physiological state. Then, the proportion of the total number of all growth moments to the total number of moments in this training program is used as the persistence; the greater the persistence, the longer the high-intensity training can be sustained. The product of growth smoothness and persistence is calculated to obtain a sustained growth adaptation index; the longer and smoother the high-intensity training is sustained, the stronger the effective training capacity.

[0035] It should be noted that negative correlation mapping is a technique well known to those skilled in the art, and can be implemented in an inverse proportional form or a negative exponential form, etc. For example, negative exponential functions with the natural constant as the base are used for negative correlation, which will not be restricted or elaborated here.

[0036] Considering the fluctuation phase distribution caused by high-intensity training, based on the temporal maximum of the growth rate at each growth moment in the training project, a maximum growth interval is defined. That is, the maximum growth rate is obtained temporally, and the duration interval between any two adjacent maximum values ​​is taken as the maximum growth interval. It should be noted that the method for obtaining the maximum value is a technique well-known to those skilled in the art, and methods such as the derivative method can be used; no restrictions are placed here.

[0037] Finally, the adaptability of the training program is obtained based on the distribution stability of all maximum growth intervals and the sustained growth index. The more consistent and stable the duration between maximum growth intervals, the more stable the respiratory rise rhythm, and the higher the patient's adaptability. In one specific embodiment of the invention, the product of the negatively correlated standard deviation of all maximum growth intervals and the sustained growth index is used as the adaptability of the training program. A higher adaptability indicates that the patient can maintain a stable physiological state even under high intensity.

[0038] S203: Weight each training item according to its order in the rehabilitation training, and combine the adaptive capacity of all training items to obtain the intensity adaptation factor in the rehabilitation training.

[0039] Patients undergoing neurorehabilitation typically have weak physiological reserves, such as poor respiratory muscle endurance and low cardiovascular compensatory capacity. When the training intensity exceeds the patient's physiological load, it can not only lead to a decline in effectiveness but may also trigger risks such as respiratory failure and muscle damage. Therefore, it is necessary to analyze the continuous adaptation of multiple training programs and comprehensively assess the match between the patient and the training intensity for each rehabilitation training session.

[0040] In this embodiment of the invention, the training items in rehabilitation training are sorted in chronological order to obtain the serial number of each training item. The product of the serial number of each training item and the corresponding adaptive ability is used as the weighted fitness of each training item. The sum of all weighted fitness values ​​is used as the intensity adaptation factor of rehabilitation training.

[0041] The greater the patient's adaptability during the training program and the later the program is in the overall rehabilitation training, the better the patient's physiological adaptability is in later and more intense training programs. This indicates that the patient's adaptability gradually improves during the training process, and the better the training effect.

[0042] S3: Analyze muscle fatigue based on the periodic changes in electromyography (EMG) values ​​in each training exercise, and analyze the degree of movement non-standardization based on the deviation in the time series data of the reference angle deviation at each joint, thereby determining the training quality indicators for each training exercise.

[0043] Patients undergoing neurorehabilitation need to strictly control the intensity of each training session to ensure safety. However, some safety guarantees may be accompanied by insufficient stimulation. For example, high adaptability only indicates that the intensity is within the physiological tolerance range, but it cannot determine whether the rehabilitation training has reached the level of stimulation required for functional recovery. Therefore, in order to more accurately assess the actual training effect of patients in each rehabilitation training session, it is necessary to further analyze the patients' muscle vitality status.

[0044] For patients with impaired motor function, rehabilitation training often requires highly regular repetitive movements to activate dormant neural pathways, strengthen muscle memory of movements, and help the brain and muscles re-establish signal connections. Without regularity, stable functional compensation cannot be formed.

[0045] Therefore, considering the patient's movement process, the training quality of behavioral movements is analyzed. Preferably, in this embodiment of the invention, the method for obtaining training quality indicators... Figure 3 The diagram illustrates a method for obtaining training quality metrics according to an embodiment of the present invention, which includes the following steps: S301: For any training program in a rehabilitation training session, the muscle fatigue level of that training program is obtained based on the periodic activation stability and signal deviation changes of the electromyography (EMG) data in the EMG time series data.

[0046] By analyzing changes in electromyography (EMG) values, periodic movement periods are determined. The maintenance and changes in muscle activation intensity within these periods reflect the fatigue state of the neuromuscular system. In this embodiment of the invention, the movement cycle is divided using adjacent peak values ​​in the EMG time-series data of the training program. After obtaining the peak values ​​in the EMG time-series data, the time interval between each two adjacent peak values ​​is taken as each movement cycle.

[0047] In each exercise cycle, the average peak value is used as the activation index, reflecting the distribution of muscle activation during the cycle. Furthermore, based on the deviation of the activation index between the first and last exercise cycles of the training program, as well as the deviation of the electromyography (EMG) values ​​at the beginning and end, the muscle fatigue level of the training program is obtained. The degree of fatigue is assessed by observing the changes throughout a single training session.

[0048] In one specific embodiment of the present invention, the difference in activation indicators between the first and last exercise cycles in the training program is used as the fatigue accumulation degree. The greater the fatigue accumulation degree, the weaker the muscle's ability to maintain stable contraction strength during continuous movement, the more obvious the peripheral muscle fatigue accumulation, and the lower the muscle's physiological reserves. Here, the difference is the absolute value of the numerical difference.

[0049] Further calculations were made of the difference in electromyographic (EMG) values ​​between the first and last moments of the training program, which was used as the signal attenuation. The product of the signal attenuation and the fatigue accumulation was used as the muscle fatigue level. The greater the deviation in the EMG values ​​during the training program, the higher the intensity of the training loss and the more significant the fatigue. Combined with the fatigue status of the cyclic movements, the overall muscle fatigue level was obtained.

[0050] S302: Determine the non-standard moments based on the magnitude of the time series data of the reference angle deviation at different joints, and analyze the distribution of non-standard moments and the degree of deviation of the corresponding different joints to obtain the non-standard degree of the movement in this training item.

[0051] One of the core goals of neurorehabilitation is to activate the signaling connections between the brain and muscles through training, helping the brain regain control over the muscles. In patients with nerve damage, motor control is often imprecise, potentially leading to irregular or asymmetrical joint angles. These irregular movement patterns can cause joint deformities, pain, or further functional impairments, hindering the recovery of motor function. Therefore, it is necessary to assess the standardization of movements based on changes in the angles of each joint during training.

[0052] In this embodiment of the invention, based on the magnitude of the reference angle deviation value at each joint at each moment of the training project, deviating joints are screened out. In a specific embodiment of the invention, the preset deviation threshold is 0.2. When the reference angle deviation data at the joint at a moment is greater than or equal to the preset deviation threshold, it indicates that the angle deviates from the reference to a high degree. At this moment, the joint has a high deviation from the reference and is marked as a deviating joint.

[0053] For each moment, based on the distribution of the number of deviating joints, an irregular moment is determined. When a large number of joints at a moment show a high degree of deviation from the baseline, the movement is highly irregular. In a specific embodiment of the present invention, the preset joint number threshold is set to half of the total number of detected joints. When the number of deviating joints at a moment is greater than the preset joint number, the corresponding moment is marked as an irregular moment.

[0054] It should be noted that the threshold setting value can be adjusted by the implementer according to the specific implementation scenario, and there are no restrictions here.

[0055] The time intervals formed by adjacent non-standard moments are defined as non-standard time intervals, reflecting the periods when non-standard movements occur. Multiple non-standard time intervals may exist in a training program. Therefore, within each non-standard time interval, the magnitude of the deviation from the reference joint angle corresponding to each non-standard moment, as well as the distribution percentage of deviations, are analyzed to obtain the standardization difference degree for each non-standard time interval. After a non-standard movement occurs, the higher the percentage of deviations from the joint and the greater the corresponding deviation from the reference, the greater the difference between the movement and the standard movement, and the lower the training quality.

[0056] In one specific embodiment of the present invention, for any non-standard moment within a non-standard time period, the ratio of the number of deviating joints at that non-standard moment to the total number of detected joints is taken as the non-standard percentage. The mean of the reference angle deviation data of all deviating joints at that non-standard moment is taken as the non-standard deviation degree. The product of the non-standard percentage and the non-standard deviation degree is calculated to obtain the local non-standard degree for each non-standard moment, reflecting the degree of difference from the standard movement at that moment. Finally, the sum of the local non-standard degrees of all non-standard moments within the non-standard time period is taken as the standard difference degree for that non-standard time period.

[0057] Finally, by combining the degree of difference in standardization and the proportion of non-standard periods in the training program, the degree of non-standardization of the training program is obtained. The higher the proportion of non-standard periods, the more non-standard movements occur. The greater the degree of difference in standardization, the higher the degree of non-standardization. Therefore, the greater the degree of non-standardization, the lower the training quality should be.

[0058] In one specific embodiment of the present invention, the ratio of the total duration of all non-standard periods in the training project to the total duration of the training project is taken as the non-standard duration percentage, and the product of the normalized sum of the standardization differences of all non-standard periods in the training project and the non-standard duration percentage is taken as the non-standardization degree of the training project.

[0059] S303: Combine the muscle fatigue and movement irregularity of this training program to obtain the training quality index. Both muscle fatigue and movement irregularity are negatively correlated with the training quality index.

[0060] Training quality is comprehensively assessed by combining muscle fatigue levels with movement standardization. The greater the muscle fatigue and the more improper the movement during training, the less likely the expected training effect of the program will be achieved. Therefore, in this embodiment of the invention, the product of muscle fatigue and movement standardization is calculated, and the product is negatively correlated and normalized to serve as the training quality indicator for that program.

[0061] Specifically, when there are no non-standard moments in the training program, only the value of muscle fatigue after negative correlation mapping and normalization is used as the training quality indicator, without considering the impact of non-standard movements.

[0062] S4: Combine the intensity adaptation factor and training quality indicators in each rehabilitation training session to obtain the training effect indicators for each rehabilitation training session; make predictions based on the changing trends of the training effect indicators of historical rehabilitation training sessions.

[0063] The effectiveness of each rehabilitation training session is comprehensively quantified by analyzing the overall quality and intensity of the training. Higher overall training quality indicators and a larger intensity adaptation factor in a patient's past rehabilitation training indicate a smoother training session, a greater likelihood of achieving the expected results, and thus a better overall rehabilitation outcome for the patient.

[0064] In this embodiment of the invention, for any rehabilitation training session, reference training items are selected based on the magnitude of the training quality index. The distribution ratio of high-quality training items is analyzed, and the more efficient the training items, the more significant the training effect. In one specific embodiment of the invention, a quality threshold of 0.8 is set, and training items in the rehabilitation training whose training index is greater than the preset quality threshold are used as reference training items.

[0065] Furthermore, by combining the training quality indicators of the training items in the rehabilitation training with the number of reference training items, an effective quality score is obtained. The higher the overall training quality in the rehabilitation training and the greater the proportion of highly efficient training items, the better the overall training effect. In a specific embodiment of the present invention, the proportion of reference training items in the rehabilitation training among all training items is used as the high efficiency proportion score. The product of the high efficiency proportion score and the mean of the training quality indicators of all training items is calculated to obtain the effective quality score of the rehabilitation training.

[0066] A higher effective quality score and a higher intensity adaptation factor indicate a better training effect from rehabilitation training. Therefore, by combining the effective quality score and the intensity adaptation factor, this effective index of rehabilitation training is obtained. In this embodiment of the invention, the product of the effective quality score and the intensity adaptation factor is normalized to obtain this effective index of rehabilitation training.

[0067] The trend of historical rehabilitation training effects can reflect the rate of recovery progress. In this embodiment of the invention, the trend of effective training indicators of continuous rehabilitation training within a preset time period before the current moment is analyzed. The two weeks before the current moment are used as the preset time period, which means analyzing the historical rehabilitation training of the most recent two weeks. If the duration before the current moment does not meet the two-week requirement, the actual length is used as the preset time period.

[0068] It is understandable that the progress of training effects requires a certain amount of training to be demonstrated. Therefore, when the number of rehabilitation training sessions exceeds the minimum progress number, progress analysis and prediction should be performed. The minimum progress number can be selected as 3. The implementer can adjust the minimum progress number and the preset time period length at their own discretion, and there are no restrictions here.

[0069] In one specific embodiment of the present invention, the effective training indicators of rehabilitation training within a preset time period are linearly fitted over time. The slope of the fitted line is used as the trend degree, reflecting the trend direction of the overall training effect. Combining the differences in effective training indicators between the first and last rehabilitation training sessions within the preset time period, the predictive value of the training progress rate at the current moment is obtained. The difference between the effective training indicator of the most recent rehabilitation training session and the effective training indicator of the first rehabilitation training session within the preset time period is used as the overall change degree of the recent rehabilitation training effect. Furthermore, the product of the trend degree and the overall change degree is used as the predictive value of the training progress rate, reflecting the predicted magnitude of the recovery rate. Based on the predicted value of the patient's neurological recovery speed at the current moment, the physician is assisted in adjusting the treatment or rehabilitation training plan to improve rehabilitation effects and optimize training resources.

[0070] In summary, this invention comprehensively analyzes the patient's physiological state and physical activity capacity based on respiratory rate, electromyography (EMG) values, and joint reference angle deviations. By analyzing the increase in respiratory rate, it identifies intensity adaptation factors and assesses the fit between training intensity and the patient's physiological tolerance. Based on the periodic changes in EMG values ​​and the degree of joint angle deviation, it quantifies muscle fatigue and movement irregularities, ensuring accurate assessment of the patient's motor ability during training and reflecting the execution quality of each training program. Furthermore, by combining intensity adaptation factors and training quality indicators, it obtains training effect indicators that characterize the actual value of each training session. This allows the trends to more accurately reflect the actual training effect of rehabilitation progress and predict subsequent rehabilitation progress. This invention, through multi-dimensional data collaborative analysis, provides a more comprehensive and objective quantitative assessment of the neurorehabilitation process and trend prediction, effectively improving the accuracy of rehabilitation management and providing reliable data support.

[0071] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0072] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for predicting the progression of neurorehabilitation based on multimodal data fusion, characterized in that, The method includes: Acquire multi-dimensional time-series data for each training item in each historical rehabilitation training process of the patient; the multi-dimensional time-series data includes respiratory rate time-series data, electromyography value time-series data, and reference angle deviation time-series data at each joint; By analyzing the growth trend of respiratory rate time-series data, the patient's physiological adaptability to each training item in a single rehabilitation training session is obtained, and the intensity adaptation factor in each rehabilitation training session is obtained. Muscle fatigue was analyzed based on the periodic changes in electromyography (EMG) values ​​in each training exercise, and the degree of non-standard movement was analyzed based on the deviation in the time series data of the baseline angle deviation at each joint, thus determining the training quality indicators for each training exercise. By combining the intensity adaptation factor and training quality indicators in each rehabilitation training session, the training effect indicators for each rehabilitation training session are obtained. Predictions are made based on the changing trends of historical rehabilitation training effectiveness indicators.

2. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 1, characterized in that, The method for obtaining the intensity adaptation factor includes: For any rehabilitation training program, the time of increase is selected based on the increase of respiratory rate values ​​between adjacent time points in the respiratory rate time sequence data of that training program. The adaptability of the training program is determined based on the degree of growth trend and the degree of stability and continuity of growth at each growth point. Each training item is weighted according to its order in the rehabilitation training, and the intensity adaptation factor of the rehabilitation training is obtained by weighting and combining the adaptability of all training items.

3. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 2, characterized in that, The method for obtaining the growth time includes: For any given moment, the rate of increase at that moment is obtained based on the numerical difference between the respiratory rate data at the next moment and the moment itself; when the rate of increase is positive, the corresponding moment is recorded as the time of increase.

4. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 3, characterized in that, The methods for acquiring the adaptability include: Obtain the slope of the respiratory rate time series data at each growth moment; combine the proportion and slope distribution of all growth moments in this training project to obtain the continuous growth adaptation index of this training project. The slope is negatively correlated with the continuous growth adaptation index. Based on the maximum value of the growth rate at each growth moment in the training project, the maximum growth interval is divided; based on the distribution stability of all maximum growth intervals in the training project and the continuous growth index, the adaptability of the training project is obtained.

5. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 1, characterized in that, The methods for obtaining the training quality metrics include: For any training program in any rehabilitation training session, the muscle fatigue level of that training program can be obtained based on the periodic activation stability and signal deviation changes of the electromyography (EMG) data in the time series data. The non-standard moments are determined by the magnitude of the time series data of the reference angle deviation at different joints, and the distribution of non-standard moments and the degree of deviation of the corresponding different joints are analyzed to obtain the non-standard degree of the movement in this training item. By combining muscle fatigue and movement irregularity in this training program, training quality indicators were obtained. Both muscle fatigue and movement irregularity were negatively correlated with the training quality indicators.

6. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 5, characterized in that, The method for obtaining muscle fatigue includes: The exercise cycle was divided by adjacent peak values ​​of the electromyography (EMG) time series data in this training project; the average peak value in each exercise cycle was used as the activation index for each exercise cycle. The muscle fatigue level of the training program is obtained based on the degree of deviation of activation indicators at the beginning and end of the exercise cycle, as well as the degree of deviation of electromyography (EMG) data at the beginning and end.

7. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 5, characterized in that, The methods for obtaining the degree of non-standardity of the action include: Based on the magnitude of the reference angle deviation at each joint at each moment in the training project, deviating joints are screened out; based on the distribution of the number of deviating joints, non-standard moments are determined; and the time period composed of adjacent non-standard moments is taken as the non-standard time period. In each non-standard time period, the magnitude of the deviation of the reference angle from the joint at each non-standard time point and the distribution ratio of the deviation from the joint are analyzed to obtain the standardization difference degree of each non-standard time period. By combining the difference in standardization and the proportion of non-standardization periods in all non-standardization periods in the training program, the degree of non-standardization of the movements in the training program is obtained.

8. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 1, characterized in that, The methods for obtaining the training performance metrics include: For any given rehabilitation training session, reference training items are selected based on the magnitude of the training quality indicators; the effective quality score is obtained by combining the magnitude of the training quality indicators of the training items in the rehabilitation training session with the number of reference training items. By combining the effective quality score and intensity adaptation factor, the effective indicators of this rehabilitation training are obtained.

9. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 1, characterized in that, The prediction of training effect indicators based on historical rehabilitation training trends includes: By analyzing the trend change of effective training indicators of continuous rehabilitation training within a preset time period before the current moment, and combining the differences in effective training indicators between the first and last rehabilitation training sessions within the preset time period, the predictability of the training progress rate at the current moment is obtained.

10. The method for predicting the progression of neurorehabilitation based on multimodal data fusion according to claim 1, characterized in that, The method for obtaining the reference angle deviation time series data includes: At any joint, the difference between the angle data collected by the inertial sensor and the standard angle data is used to determine the reference angle deviation data at each moment. The reference angle deviation data in the time sequence is used as the reference angle deviation time sequence data at that joint.

Images