Exercise ability analysis method for rehabilitation data enhancement processing

Abstract

The invention relates to an athletic ability analysis method for rehabilitation data enhancement processing, which relates to the field of data processing, and comprises the following steps: performing primary retrieval by taking a multi-modal rehabilitation data sequence as a reference to obtain a primary similar user set; determining exercise demand information in combination with a life scene, and screening the primary similar user set to obtain a secondary similar user set; obtaining an exercise ability analysis report set based on secondary similar user set analysis, and training to obtain an exercise ability analyzer; and performing analysis according to the multi-modal rehabilitation data sequence by using a motion ability analyzer, and outputting a target motion ability analysis report. Through the method, the problems that a traditional analysis method cannot comprehensively reflect the real exercise ability of a patient, and analysis accuracy and practicability are insufficient can be solved; athletic ability analysis is carried out by utilizing artificial intelligence and big data technologies and combining life scenes and exercise demands of patients, so that scientificity, accuracy and practicability of exercise ability analysis of the patients can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to a method for analyzing motor ability through enhanced processing of rehabilitation data. Background Art

[0002] In the field of modern rehabilitation medicine, the assessment of patients' motor ability is a key link in formulating rehabilitation plans. Especially for patients with cerebrovascular diseases (such as stroke and cerebral infarction), the recovery of motor ability is directly related to the patient's quality of life.

[0003] At present, most motor ability assessment methods rely on regular doctor examinations and standardized scoring systems. This method cannot fully reflect the dynamic changes of patients during the rehabilitation process, nor can it capture fluctuations in motor ability in a timely manner. In addition, each patient's rehabilitation process and life needs are different. Traditional methods are difficult to dynamically adjust the assessment criteria according to the patient's actual situation, and fail to effectively combine the patient's actual life needs (such as whether they can go to the toilet or walk independently), resulting in the lack of practical significance of the implementation of the assessment and plan. Summary of the invention

[0004] The present invention aims to solve the technical problems that traditional motor ability analysis methods cannot fully reflect the patient's true motor ability and have insufficient analysis accuracy and practicality, and provides a motor ability analysis method with enhanced processing of rehabilitation data to solve the problem.

[0005] The technical solution of the present invention to solve the above technical problems is as follows: In a first aspect, the present invention provides a method for analyzing athletic ability by enhanced processing of rehabilitation data, comprising: collecting a multimodal rehabilitation data sequence of a target user, taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data to obtain a similar user set, wherein the search includes a data similarity comparison and a data fluctuation similarity comparison; determining athletic demand information in combination with a life scenario of the target user, screening the primary similar user set based on the athletic demand information to obtain a secondary similar user set; based on the secondary similar user set, analyzing and obtaining an athletic ability analysis report set, and training a athletic ability analyzer; utilizing the athletic ability analyzer to perform analysis according to the multimodal rehabilitation data sequence, outputting a target athletic ability analysis report, and formulating an athletic rehabilitation plan for the next rehabilitation stage according to the target athletic ability analysis report.

[0006] Preferably, the method for analyzing athletic ability by enhanced processing of rehabilitation data also includes: taking the end of the rehabilitation stage as a data collection node, continuously recording the multimodal rehabilitation data of the target user, and constructing a multimodal rehabilitation data sequence; obtaining the number of data collection nodes, and configuring a first similarity threshold, wherein the first similarity threshold decreases with the increase in the number of collection nodes; taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data in accordance with the order of data collection nodes, setting users whose data of all data collection nodes meet the first similarity threshold as initial similar users, and obtaining an initial similar user set; performing rehabilitation volatility analysis on the multimodal rehabilitation data sequence, and screening the initial similar user set according to the multivariate rehabilitation volatility coefficient, and obtaining the primary similar user set.

[0007] Preferably, the method for analyzing motor ability by enhanced processing of rehabilitation data further includes: the multimodal rehabilitation data at least includes physiological data, functional data, imaging data and rehabilitation treatment records.

[0008] Preferably, the athletic ability analysis method for enhanced processing of rehabilitation data also includes: respectively calculating the data mean and data standard deviation of the multimodal rehabilitation data sequence, setting the ratio of the data mean to the data standard deviation as the rehabilitation fluctuation coefficient, and obtaining the multivariate rehabilitation fluctuation coefficient; respectively calculating to obtain the historical multivariate rehabilitation fluctuation coefficient set of the initial similar user set; performing similarity screening on the historical multivariate rehabilitation fluctuation coefficient set according to the multivariate rehabilitation fluctuation coefficient, selecting the initial similar users whose multivariate rehabilitation fluctuation coefficients all meet the first similarity threshold as primary similar users, and obtaining the primary similar user set.

[0009] Preferably, the method for analyzing sports ability by enhanced processing of rehabilitation data also includes: determining sports demand information in combination with life scenario analysis of the target user, wherein the sports demand information includes a number of autonomously completed sports types; based on the several autonomously completed sports types, screening the first-level similar user set, setting the first-level similar users whose sports type intersection ratio is greater than a predetermined threshold as secondary similar users, and obtaining a secondary similar user set, wherein the sports type intersection ratio is the ratio of the number of sports type intersections to the total number of sports types of the first-level similar users, and the predetermined threshold is 80%.

[0010] Preferably, the method for analyzing athletic ability by enhanced processing of rehabilitation data also includes: based on big data, obtaining multiple sample multimodal rehabilitation data sequences of multiple secondary similar users in the secondary similar user set, and obtaining multiple sample athletic ability analysis reports; constructing K athletic ability analysis plug-ins based on a feedforward neural network, wherein K is an integer greater than or equal to 10; using the multiple sample multimodal rehabilitation data sequences and the multiple sample athletic ability analysis reports as training data, and dividing them into K parts, selecting them with replacement K times to construct a first training set, and iteratively selecting them K times to obtain K training sets; using the K training sets to respectively perform supervised training and verification on the K athletic ability analysis plug-ins until predetermined constraints are met, and obtaining K converged athletic ability analysis plug-ins, which are combined to obtain the athletic ability analyzer.

[0011] Preferably, the method for analyzing athletic ability by enhancing rehabilitation data further includes: counting the proportion of exercise types completed independently by each secondary similar user within a preset time range after the end of treatment, and constructing a sample athletic ability analysis report based on the autonomously completed exercise types; and analyzing in sequence to obtain multiple sample athletic ability analysis reports for multiple secondary similar users.

[0012] Preferably, the method for analyzing athletic ability by enhancing rehabilitation data processing also includes: performing mean calculation on the multiple sample multimodal rehabilitation data sequences to obtain a sample multimodal rehabilitation data mean sequence; performing deviation analysis on the multimodal rehabilitation data sequence and the sample multimodal rehabilitation data mean sequence to obtain a data deviation ratio; calculating the ratio of the data deviation ratio to the historical maximum data deviation ratio, multiplying it by K and rounding it to the integer to obtain the number of plug-in selections P; randomly selecting P convergent athletic ability analysis plug-ins from the K convergent athletic ability analysis plug-ins of the athletic ability analyzer, analyzing the multimodal rehabilitation data sequence, outputting P predicted athletic ability analysis reports, and obtaining the target athletic ability analysis report after mean calculation.

[0013] The beneficial effects of the present invention are as follows: by collecting the multimodal rehabilitation data sequence of the target user, taking the multimodal rehabilitation data sequence as a benchmark, a search is performed based on big data to obtain a similar user set, wherein the search includes data similarity comparison and data fluctuation similarity comparison; then the motion demand information is determined in combination with the life scene of the target user, and the motion demand information is used as a benchmark to screen the first similar user set to obtain a secondary similar user set; further based on the secondary similar user set, a set of motion ability analysis reports is analyzed and obtained, and a motion ability analyzer is trained; then the motion ability analyzer is used to analyze according to the multimodal rehabilitation data sequence, and a target motion ability analysis report is output; finally, a motion rehabilitation plan for the next rehabilitation stage is formulated according to the target motion ability analysis report. In other words, by using artificial intelligence and big data technology, and combining the patient's life scene and motion needs to perform motion ability analysis, the scientificity, accuracy and practicality of the patient's motion ability analysis can be significantly improved, so that a more accurate and effective motion rehabilitation plan can be formulated for the patient. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A schematic diagram of a flow chart of a method for analyzing motor ability by enhanced processing of rehabilitation data provided by the present invention; Figure 2 A schematic diagram of a process of obtaining a similar user set in a motor ability analysis method for enhanced rehabilitation data processing provided by the present invention. DETAILED DESCRIPTION

[0015] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0016] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0017] In the description of the present invention, the term "for example" is used to mean "used as an example, illustration or explanation". Any embodiment described as "for example" in the present invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is given to enable any technician in the field to implement and use the present invention. In the following description, details are listed for the purpose of explanation. It should be understood that a person of ordinary skill in the art can recognize that the present invention can be implemented without using these specific details. In other examples, well-known structures and processes will not be elaborated in detail to avoid obscuring the description of the present invention with unnecessary details. Therefore, the present invention is not intended to be limited to the embodiments shown, but is consistent with the widest scope consistent with the principles and features disclosed in the present invention.

[0018] Examples, such as Figure 1 As shown, the embodiment of the present invention provides a method for analyzing sports ability through enhanced processing of rehabilitation data, which specifically includes the following steps: S10: Collecting a multimodal rehabilitation data sequence of a target user, taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data, and obtaining a similar user set, wherein the search includes a data similarity comparison and a data fluctuation similarity comparison.

[0019] Further, if Figure 2 As shown, step S10 of the present invention further includes: S11: Taking the end of the rehabilitation stage as the data collection node, continuously record the multimodal rehabilitation data of the target user and construct a multimodal rehabilitation data sequence.

[0020] Furthermore, step S11 of the present invention further includes: S111: The multimodal rehabilitation data at least includes physiological data, functional data, imaging data and rehabilitation treatment records.

[0021] Specifically, during the patient's rehabilitation process, each rehabilitation stage usually has different treatment goals and evaluation criteria. The end of the rehabilitation stage refers to the evaluation node at the end of each rehabilitation cycle (which can be set according to the actual scenario, such as setting a rehabilitation cycle every 5 days), which is usually the patient's final state of treatment and recovery at this stage. This node is crucial for recording the patient's rehabilitation effect at a specific stage, because it can reflect the patient's overall rehabilitation status and provide guidance for the next step of treatment; at the end of each rehabilitation stage, the data collection node can comprehensively evaluate the patient's motor ability, physical function and overall rehabilitation progress at that stage.

[0022] The end of the rehabilitation stage is used as the data collection node to continuously record the multimodal rehabilitation data of the target user, that is, during the rehabilitation treatment process, the patient's rehabilitation process is comprehensively collected and recorded from multiple dimensions and different angles. These data can reflect the patient's rehabilitation progress and status in real time. Among them, the multimodal rehabilitation data at least includes physiological data, functional data, imaging data and rehabilitation treatment records. Physiological data reflects the basic indicators of the patient's physical health status, such as heart rate, blood oxygen saturation, blood pressure, body temperature, weight, etc.; functional data mainly involves the patient's motor function recovery during the rehabilitation process, such as joint range of motion, muscle strength, balance ability, walking distance and speed, etc.; imaging data can provide morphological basis for motor ability assessment, and common imaging data include CT scans, MRI images, etc.; rehabilitation treatment records include all treatment content and progress received by the patient, including but not limited to physical therapy, exercise therapy, drug therapy, etc. By integrating the above-mentioned multiple data sources (physiological data, functional data, imaging data, rehabilitation treatment records), a multimodal rehabilitation data sequence is constructed. This data sequence can record the patient's rehabilitation process in detail by tracking the changes in various data of the patient during the rehabilitation stage for a long time. Through comprehensive analysis of multimodal data, the accuracy and practicality of rehabilitation assessment can be significantly improved.

[0023] S12: Obtain the number of data acquisition nodes and configure a first similarity threshold, wherein the first similarity threshold decreases as the number of acquisition nodes increases; S13: Based on the multimodal rehabilitation data sequence and in accordance with the order of data acquisition nodes, perform a search based on big data, set users whose data of all data acquisition nodes meet the first similarity threshold as initial similar users, and obtain an initial similar user set.

[0024] Specifically, first, determine the number of data collection nodes. Data collection nodes refer to events that record data at different time points during rehabilitation treatment based on the patient's specific situation. The number of these nodes directly affects the accuracy and coverage of similarity calculations. Next, a first similarity threshold is configured according to the number of data acquisition nodes. The first similarity threshold is a standard for measuring the consistency of similarity between two sets of multimodal data sequences. The first similarity threshold decreases with the increase in the number of acquisition nodes, that is, with the increase in the number of acquisition nodes, the comparable points and information in the data sequence increase. The initial similarity threshold should be gradually reduced to allow more data sets to be compared, thereby expanding the range of similar user sets. When there are fewer data acquisition nodes, a higher first similarity threshold is set to ensure that the data similarity is high and a more accurate similar user set is screened out; as the number of acquisition nodes increases, the threshold is gradually reduced, allowing a wider range of user data sets to be selected into the similar user set. For example, the initial similarity threshold is set to 0.85, and the similarity threshold is reduced by 0.015 for each additional data acquisition node. That is, if the number of data acquisition nodes is 10, the first similarity threshold is 0.85-0.015*10, which is equal to 0.7.

[0025] Next, taking the multimodal rehabilitation data sequence as a benchmark, a search is performed based on big data in accordance with the order of data collection nodes, that is, for each data collection node, the similarity between users is calculated. The similarity calculation can use a variety of methods, such as cosine similarity, Euclidean distance, etc., or extract features through deep learning algorithms (such as autoencoders or convolutional neural networks) for comparison; a search is performed based on big data to calculate the similarity of each user at each data collection node, and to determine whether the multimodal data sequence of each user meets the set first similarity threshold. Only when the similarity of the user at all data collection nodes is greater than the threshold condition, is the user included in the initial similar user set; then the user who meets the similarity requirements of all data collection nodes is set as the initial similar user, and the initial similar user set is obtained. The initial similar user set contains multiple candidate users, and the performance of these users on the multimodal data is highly similar to that of the target user. Through big data retrieval, a comprehensive comparison of multimodal data sequences is ensured, an accurate similar user set is obtained, and the reliability and practicality of the analysis results are further improved.

[0026] S14: performing rehabilitation volatility analysis on the multimodal rehabilitation data sequence, and screening the initial similar user set according to the multivariate rehabilitation volatility coefficient to obtain the primary similar user set.

[0027] Further, step S14 of the present invention also includes: S141: Calculate the data mean and data standard deviation of the multimodal rehabilitation data sequence respectively, set the ratio of the data mean to the data standard deviation as the rehabilitation fluctuation coefficient, and obtain the multivariate rehabilitation fluctuation coefficient; S142: Calculate and obtain the historical multivariate rehabilitation fluctuation coefficient set of the initial similar user set respectively; S143: According to the multivariate rehabilitation fluctuation coefficient, perform similarity screening on the historical multivariate rehabilitation fluctuation coefficient set, select the initial similar users whose multivariate rehabilitation fluctuation coefficients all meet the first similarity threshold as the first similar users, and obtain the first similar user set.

[0028] Specifically, for the multimodal rehabilitation data sequence of the target user, the data mean and data standard deviation of each data sequence (such as physiological data, functional data, imaging data, etc.) are calculated respectively, and then the ratio of the data mean to the data standard deviation of each data sequence is set as the rehabilitation fluctuation coefficient. The rehabilitation fluctuation coefficient is an important indicator to measure the volatility of rehabilitation data. The higher the fluctuation coefficient, the stronger the volatility of the sequence data. For example, if the fluctuation coefficient of a data sequence is high, it means that the sequence shows a large volatility during the acquisition process, which means that there is a large fluctuation or instability in the patient's rehabilitation process; if the fluctuation coefficient is low, it means that the data changes relatively smoothly and the rehabilitation process may be relatively stable. The multivariate rehabilitation fluctuation coefficient is calculated in sequence, where the multivariate rehabilitation fluctuation coefficient includes multiple rehabilitation fluctuation coefficients of multiple rehabilitation data sequences, that is, the multimodal rehabilitation data sequence of the target user contains multiple different data types, such as physiological data, functional data, imaging data, etc., and each data type has its corresponding rehabilitation fluctuation coefficient.

[0029] Next, for each user in the initial similar user set, the historical multivariate rehabilitation fluctuation coefficient is calculated to obtain a historical multivariate rehabilitation fluctuation coefficient set. Then, according to the multivariate rehabilitation fluctuation coefficient, the historical multivariate rehabilitation fluctuation coefficient set is similarly screened, for example, the Euclidean distance or cosine similarity between the target user and the initial similar user is calculated to quantify the difference; further, the initial similar users whose multivariate rehabilitation fluctuation coefficients are all greater than the first similarity threshold are selected as primary similar users to obtain a primary similar user set. By analyzing the rehabilitation volatility to screen the initial similar user set, the screening range can be further narrowed to select users who are more similar to the target user in terms of volatility.

[0030] S20: determining exercise demand information in combination with the target user's life scenario, and screening the primary similar user set based on the exercise demand information to obtain a secondary similar user set.

[0031] Further, step S20 of the present invention further includes: S21: Determine the exercise demand information in combination with the life scenario analysis of the target user, wherein the exercise demand information includes a number of self-completed exercise types; S22: Based on the several self-completed exercise types, screen the first similar user set, set the first similar users whose exercise type intersection ratio is greater than a predetermined threshold as secondary similar users, and obtain the secondary similar user set, wherein the exercise type intersection ratio is the ratio of the number of exercise type intersections to the total number of exercise types of the first similar users, and the predetermined threshold is 80%.

[0032] Specifically, first, the exercise demand information is determined in combination with the target user's life scenario analysis, where the exercise demand information includes several types of autonomous exercises. By observing and recording the patient's daily life activities, the patient's actual exercise needs are understood, and these needs are converted into quantifiable exercise types, such as walking indoors, going to the toilet independently, climbing stairs, dressing independently, cooking independently, standing up and walking, etc. These autonomous exercise types provide basic data for subsequent motor ability analysis, which can help to more accurately evaluate the patient's rehabilitation status and formulate personalized rehabilitation treatment plans.

[0033] Next, based on the several self-completed exercise types, the first similar user set is screened. First, the exercise type intersection needs to be calculated, that is, how many exercise types are common between the exercise needs of the target user and the exercise needs of the first similar user (that is, exercise types that can be completed autonomously), and then the ratio of the number of exercise type intersections to the total number of exercise types of the first similar user is set as the exercise type intersection ratio, and the intersection ratios of multiple exercise types of multiple first similar users are obtained. Then, a predetermined threshold (80%) is obtained, and the first similar users whose exercise type intersection ratio is greater than the predetermined threshold are set as secondary similar users, that is, if the intersection ratio is greater than 80%, the user is considered to be a user with similar exercise needs to the target user, and a secondary similar user set is obtained, wherein the exercise needs of the secondary similar users are consistent with those of the target user to a high degree, which can provide more accurate data support for subsequent exercise ability analysis, rehabilitation program formulation, etc., and provide a more reliable reference basis for personalized rehabilitation programs.

[0034] S30: Based on the secondary similar user set, analyze and obtain a set of sports ability analysis reports, and train a sports ability analyzer.

[0035] Further, step S30 of the present invention further includes: S31: Based on the big data, a plurality of sample multimodal rehabilitation data sequences of a plurality of secondary similar users in the secondary similar user set are obtained, and a plurality of sample motion ability analysis reports are obtained.

[0036] Further, step S31 of the present invention further includes: S311: Count the proportion of exercise types completed independently by each secondary similar user within a preset time range after the end of treatment, and construct a sample exercise ability analysis report based on the self-completed exercise types; S312: Analyze in sequence to obtain multiple sample exercise ability analysis reports for multiple secondary similar users.

[0037] Specifically, based on big data, for each secondary similar user, multimodal data is extracted from the rehabilitation process to obtain multiple sample multimodal rehabilitation data sequences of multiple secondary similar users in the secondary similar user set. Then, for each secondary similar user, within a preset time range after the end of rehabilitation treatment (such as within 3 months after the end of treatment), the proportion of exercise types that they can complete independently is counted. For example, the types of exercise that a user can complete independently after treatment include indoor walking, climbing stairs, going to the toilet independently, etc., and the completion ratio of these types of exercise within the preset time range is recorded; and a sample exercise ability analysis report is constructed in combination with the self-completed exercise types, where the sample exercise ability analysis report is shown in Table 1: Table 1: Sample athletic performance analysis report

[0038] Then, the multiple secondary similar users are analyzed in sequence to obtain multiple sample sports ability analysis reports of the multiple secondary similar users.

[0039] S32: Construct K motor ability analysis plug-ins based on a feedforward neural network, where K is an integer greater than or equal to 10; S33: Use the multiple sample multimodal rehabilitation data sequences and multiple sample motor ability analysis reports as training data, and divide them into K equal parts, select them with replacement K times, construct a first training set, and iterate and select them K times to obtain K training sets; S34: Use the K training sets to perform supervised training and verification on the K motor ability analysis plug-ins respectively until the predetermined constraints are met, and obtain K convergent motor ability analysis plug-ins, which are combined to obtain the motor ability analyzer.

[0040] Specifically, a feedforward neural network is one of the most basic artificial neural network models, which is widely used in machine learning and deep learning tasks. First, K motor ability analysis plug-ins are constructed based on the feedforward neural network, where K is an integer greater than or equal to 10. The value of K can be set according to actual needs. For example, K is set to 20. The choice of K affects the capacity and computational complexity of the model. The larger the K value, the more features and patterns the model can learn, which may improve the accuracy of motor ability analysis. However, at the same time, a large K value will also increase computing resources and time consumption. Each plug-in represents an independent feedforward neural network, which is used to analyze rehabilitation data and predict patients' motor ability, including an input layer, multiple hidden layers and an output layer. The input data of the input layer is a multimodal rehabilitation data sequence, and the output data is a motor ability analysis report.

[0041] Next, the multiple sample multimodal rehabilitation data sequences and multiple sample motor ability analysis reports are used as training data and divided into K equal parts to obtain K data sets; then, K data sets are selected with replacement K times to construct the first training set, and the same method is used to iteratively select K times to obtain K training sets. The resampling method with replacement can effectively improve the generalization ability of the model and reduce the overfitting problem. Each iteration can ensure that each data sample participates in the training, and the accuracy and robustness of the model can also be tested through the validation set. Finally, the results of K training sets can be combined into a more powerful and stable motor ability analyzer.

[0042] Then, taking the sample multimodal rehabilitation data sequence as input and the sample motor ability analysis report as supervision, the K training sets are used to perform supervised training and verification on the K motor ability analysis plug-ins respectively. During the training process, the multimodal rehabilitation data sequence of each training sample is input into the corresponding plug-in model, and the input data is forward propagated through a feedforward neural network to calculate the output result, which represents the prediction of motor ability; then, the predicted result is compared with the actual motor ability analysis report, and the loss value is calculated. The loss function is usually the mean square error, which is used to measure the gap between the model output and the true value; then, the gradient of the loss function with respect to the model parameters is calculated through the back-propagation algorithm, and the weights are updated using the gradient descent method. The goal of each update is to minimize the loss function, thereby optimizing the parameters of the model; multiple iterations are performed until the loss function tends to be stable or reaches the preset stop condition; wherein, after each training, the model is evaluated using the validation set, which is data that is not involved in the training and is used to test the generalization ability of the model. The training sets of different motor ability analysis plug-ins are exchanged and used as validation sets for verification. When the training process is completed, based on the loss and evaluation indicators of the validation set, it can be judged whether each plug-in has converged, for example, the verification loss drops to a predetermined threshold; K converged motor ability analysis plug-ins are obtained, and a motor ability analyzer is obtained based on the combination of K converged motor ability analysis plug-ins. Through the above training and verification process, the motor ability analysis capabilities of each plug-in are gradually optimized using the multimodal rehabilitation data sequences and motor ability analysis reports of multiple samples, and finally combined into a system that can provide patients with personalized and scientific motor ability analysis, which can significantly improve the intelligence and efficiency of motor ability analysis.

[0043] S40: Utilizing the motor ability analyzer to analyze the multimodal rehabilitation data sequence, outputting a target motor ability analysis report, and formulating a motor rehabilitation program for the next rehabilitation stage according to the target motor ability analysis report.

[0044] Further, step S40 of the present invention further includes: S41: Calculate the mean of the multiple sample multimodal rehabilitation data sequences to obtain a sample multimodal rehabilitation data mean sequence; S42: Perform deviation analysis on the multimodal rehabilitation data sequence and the sample multimodal rehabilitation data mean sequence to obtain a data deviation ratio; S43: Calculate the ratio of the data deviation ratio to the historical maximum data deviation ratio, multiply it by K and round it up to obtain the number of plug-in selections P; S44: Randomly select P convergent motion ability analysis plug-ins from the K convergent motion ability analysis plug-ins of the motion ability analyzer, analyze the multimodal rehabilitation data sequence, output P predicted motion ability analysis reports, and obtain the target motion ability analysis report after mean calculation.

[0045] Specifically, first, the multimodal rehabilitation data sequences of multiple samples are processed, and the mean of each data sequence is calculated respectively. Assuming that the multimodal rehabilitation data sequence of each sample includes different types of data (such as physiological data, functional data, imaging data, etc.), the mean is calculated for each data type, and the means of all data types are merged in sequence order to obtain the sample multimodal rehabilitation data mean sequence. Next, the multimodal rehabilitation data sequence and the sample multimodal rehabilitation data mean sequence are subjected to deviation analysis. The goal of the deviation analysis is to quantify the deviation between each data point and the mean, and calculate the data deviation ratio. For example, for each sample data point, the difference between it and the sample multimodal rehabilitation data mean sequence is calculated, and the deviations of all data points are normalized. After the mean is calculated, the data deviation ratio is obtained. The data deviation ratio reflects the overall deviation between the multimodal data sequence and the mean sequence. The smaller the value, the closer the data is to the mean, and the larger the value, the more significant the deviation.

[0046] Then, the ratio of the data deviation ratio to the historical maximum data deviation ratio (which can be calculated based on historical record data) is calculated. This ratio reflects the degree of deviation between the current data and the historical data. Then, the ratio of the data deviation ratio to the historical maximum data deviation ratio is multiplied by K and rounded to obtain the number of plug-in selections P. P represents the number of plug-ins randomly selected from the K converged sports ability analysis plug-ins. The larger P is, the greater the degree of data deviation is, and more plug-ins are needed for multi-angle analysis. Then, according to the calculated P, P plug-ins are randomly selected from the K converged sports ability analysis plug-ins, and these plug-ins are used to analyze the multimodal rehabilitation data sequence of the target user. Each plug-in independently performs sports ability analysis based on the input multimodal data sequence, outputs a prediction result, that is, a predicted sports ability analysis report, and obtains P predicted sports ability analysis reports. The mean of the P predicted sports ability analysis reports is calculated to obtain the target sports ability analysis report.

[0047] Through deviation analysis based on multimodal data sequences and athletic ability analysis reports, the number of plug-in selections is dynamically calculated, and the final target athletic ability analysis report is generated based on the independent analysis results of multiple plug-ins. This method uses changes in data deviations to adjust the depth and accuracy of the analysis to ensure the scientificity and practicality of athletic ability assessment. At the same time, it can reduce unnecessary waste of computing resources while ensuring analysis accuracy and improve analysis efficiency.

[0048] Finally, an exercise rehabilitation plan for the next rehabilitation stage is formulated based on the target exercise capacity analysis report. That is, based on the patient's exercise capacity assessment results in the current rehabilitation stage, their deficiencies, potential areas, and types of exercise that need to be strengthened are analyzed to formulate a personalized and precise rehabilitation plan to ensure that each patient can receive the best training support at the appropriate rehabilitation stage and gradually improve their exercise capacity.

[0049] The embodiment of the present invention provides a method for analyzing sports ability by enhancing rehabilitation data, which has at least the following technical effects: By collecting the multimodal rehabilitation data sequence of the target user, taking the multimodal rehabilitation data sequence as a benchmark, a search is performed based on big data to obtain a similar user set, wherein the search includes data similarity comparison and data fluctuation similarity comparison; then the sports demand information is determined in combination with the target user's life scene, and the sports demand information is used as a benchmark to screen the first similar user set to obtain a secondary similar user set; further based on the secondary similar user set, a sports ability analysis report set is analyzed and obtained, and a sports ability analyzer is trained; then the sports ability analyzer is used to analyze according to the multimodal rehabilitation data sequence and output the target sports ability analysis report; finally, a sports rehabilitation plan for the next rehabilitation stage is formulated according to the target sports ability analysis report. In other words, by using artificial intelligence and big data technology, combined with the patient's life scene and sports needs to conduct sports ability analysis, the scientificity, accuracy and practicality of the patient's sports ability analysis can be significantly improved, so that a more accurate and effective sports rehabilitation plan can be formulated for the patient.

[0050] Although preferred embodiments of the present invention have been described, additional changes and modifications may occur to these embodiments once those skilled in the art understand the basic inventive concepts.

[0051] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention belong to the scope of the present invention and its equivalent technologies, the present invention is also intended to include these changes and variations.

Claims

1. A method for analyzing motor ability by enhancing rehabilitation data processing, characterized in that: include: Collecting a multimodal rehabilitation data sequence of a target user, taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data to obtain a similar user set, wherein the search includes a data similarity comparison and a data fluctuation similarity comparison; Determine the exercise demand information in combination with the target user's life scenario, and screen the primary similar user set based on the exercise demand information to obtain a secondary similar user set; Based on the secondary similar user set, analyzing and obtaining a set of sports ability analysis reports, and training a sports ability analyzer; The motor ability analyzer is used to analyze the multimodal rehabilitation data sequence, output a target motor ability analysis report, and formulate a motor rehabilitation plan for the next rehabilitation stage based on the target motor ability analysis report.

2. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 1, characterized in that: Collecting a multimodal rehabilitation data sequence of a target user, taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data, and obtaining a similar user set, including: Taking the end of the rehabilitation stage as the data collection node, the multimodal rehabilitation data of the target user is continuously recorded to construct a multimodal rehabilitation data sequence; Obtain the number of data collection nodes and configure a first similarity threshold, wherein the first similarity threshold decreases as the number of collection nodes increases; Taking the multimodal rehabilitation data sequence as a benchmark, performing a search based on big data in accordance with the order of data collection nodes, setting users whose data of all data collection nodes meet the first similarity threshold as initial similar users, and obtaining an initial similar user set; Perform rehabilitation volatility analysis on the multimodal rehabilitation data sequence, and screen the initial similar user set according to the multivariate rehabilitation volatility coefficient to obtain the primary similar user set.

3. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 2, characterized in that: The multimodal rehabilitation data at least includes physiological data, functional data, imaging data and rehabilitation treatment records.

4. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 2, characterized in that: Performing rehabilitation volatility analysis on the multimodal rehabilitation data sequence and screening the initial similar user set according to the multivariate rehabilitation volatility coefficient includes: Respectively calculating the data mean and data standard deviation of the multimodal rehabilitation data sequence, setting the ratio of the data mean to the data standard deviation as the rehabilitation fluctuation coefficient, and obtaining the multivariate rehabilitation fluctuation coefficient; Respectively calculating and obtaining a historical multivariate rehabilitation fluctuation coefficient set of the initial similar user set; According to the multivariate rehabilitation fluctuation coefficient, the historical multivariate rehabilitation fluctuation coefficient set is screened for similarity, and initial similar users whose multivariate rehabilitation fluctuation coefficients all meet the first similarity threshold are selected as primary similar users to obtain the primary similar user set.

5. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 1, characterized in that: Determining the exercise demand information in combination with the target user's life scenario, and screening the primary similar user set based on the exercise demand information to obtain the secondary similar user set, including: Determine the exercise demand information by analyzing the target user's life scenarios, wherein the exercise demand information includes a number of autonomous exercise types; Based on the several self-completed sports types, the first similar user set is screened, and the first similar users whose sports type intersection ratio is greater than a predetermined threshold are set as second similar users to obtain a second similar user set, wherein the sports type intersection ratio is the ratio of the number of sports type intersections to the total number of sports types of the first similar users, and the predetermined threshold is 80%.

6. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 5, characterized in that: Based on the secondary similar user set, a set of sports ability analysis reports is analyzed and obtained, and a sports ability analyzer is trained, including: Based on the big data, a plurality of sample multimodal rehabilitation data sequences of a plurality of secondary similar users in the secondary similar user set are obtained, and a plurality of sample sports ability analysis reports are obtained; Constructing K sports ability analysis plug-ins based on a feedforward neural network, where K is an integer greater than or equal to 10; The plurality of sample multimodal rehabilitation data sequences and the plurality of sample motor ability analysis reports are used as training data, and are equally divided into K parts, and are selected K times with replacement to construct a first training set, and are iteratively selected K times to obtain K training sets; The K training sets are used to perform supervised training and verification on the K athletic ability analysis plug-ins respectively until predetermined constraints are met, thereby obtaining K converged athletic ability analysis plug-ins, which are combined to obtain the athletic ability analyzer.

7. The method for analyzing sports ability by enhancing rehabilitation data processing according to claim 6, characterized in that: Get several sample athletic performance analysis reports, including: Count the proportion of exercise types completed independently by each secondary similar user within a preset time range after the end of treatment, and build a sample exercise ability analysis report based on the types of exercise completed independently; Multiple sample sports ability analysis reports of multiple secondary similar users are obtained by sequential analysis.

8. The method for analyzing motor ability by enhancing rehabilitation data processing according to claim 6, characterized in that: The motor ability analyzer is used to analyze the multimodal rehabilitation data sequence and output a target motor ability analysis report, including: Calculating the mean of the plurality of sample multimodal rehabilitation data sequences to obtain a sample multimodal rehabilitation data mean sequence; Performing deviation analysis on the multimodal rehabilitation data sequence and the sample multimodal rehabilitation data mean sequence to obtain a data deviation ratio; Calculate the ratio of the data deviation ratio to the historical maximum data deviation ratio, multiply by K and round up to obtain the number of plug-in selections P; P convergent motor ability analysis plug-ins are randomly selected from the K convergent motor ability analysis plug-ins of the motor ability analyzer, the multimodal rehabilitation data sequence is analyzed, P predicted motor ability analysis reports are output, and the target motor ability analysis report is obtained after mean calculation.

Images