Human body dynamic rehabilitation evaluation method and system for VR
By combining VR motion capture equipment and a human dynamic database with a VR rehabilitation assessment and analysis model, the accurate collection, screening, and optimization of human dynamic data have been achieved. This solves the problem of imperfect data processing in existing VR human dynamic rehabilitation assessment methods, improves the accuracy and efficiency of the assessment, and meets the needs of clinical rehabilitation treatment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- THE 958TH ARMY HOSPITAL OF THE CHINESE PEOPLES LIBERATION ARMY
- Filing Date
- 2025-12-31
- Publication Date
- 2026-04-21
AI Technical Summary
Existing VR-based human dynamic rehabilitation assessment methods suffer from inaccurate data collection, lack of effective data screening mechanisms, lack of multi-scenario adaptability for data analysis tasks, and imperfect data processing permission control, resulting in assessment accuracy, efficiency, and safety failing to meet clinical needs.
By collecting human dynamic data through VR motion capture devices and combining it with a human dynamic database and VR rehabilitation assessment and analysis model, the system can screen, transform, and optimize the target dynamic dataset, establish a data processing flow that adapts to multiple scenarios, and ensure data security and assessment accuracy.
It improves the accuracy and efficiency of VR-based dynamic human rehabilitation assessment, ensures data security, and meets the actual needs of clinical rehabilitation treatment.
Smart Images
Figure CN121890987A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of VR-based human dynamic rehabilitation assessment technology, and more specifically, to a VR-based human dynamic rehabilitation assessment method and system. Background Technology
[0002] With the rapid development of virtual reality (VR) technology, its application in the field of medical rehabilitation is becoming more and more widespread. Especially in the scenario of human dynamic rehabilitation assessment, VR technology, with its advantages such as immersive interaction and real-time motion capture, can provide richer motion data support for rehabilitation assessment, helping medical staff to accurately judge the recovery of patients' limb motor function.
[0003] Currently, existing methods for human dynamic rehabilitation assessment mostly rely on traditional motion capture equipment or manual observation and recording, which have many limitations: on the one hand, traditional assessment methods are difficult to collect detailed data on the human movement process, such as dynamic changes in joint angles and torsion, and real-time trajectories of limb displacement, resulting in insufficient accuracy and comprehensiveness of assessment results; on the other hand, existing methods lack effective integration and analysis of historical dynamic data, and cannot match corresponding standard data for comparative assessment based on the patient's age, movement habits, and other personalized characteristics, resulting in a weaker focus in the assessment process.
[0004] Meanwhile, existing VR rehabilitation assessment technologies have significant shortcomings in their data processing workflows: First, there is a lack of effective screening mechanisms for the collected human dynamic data, making it impossible to accurately identify target data that exceeds the preset time period and is in a pending state, resulting in data redundancy and affecting assessment efficiency. Second, the data optimization and cleaning logic is unclear, and a complete set of rules for the transformation and optimization of joint twisting state data sets has not been established. When the data scale is insufficient or the data quality is substandard, timely warnings and targeted processing cannot be carried out, thus affecting the reliability of rehabilitation assessments. Third, the generation of existing data optimization tasks mostly relies on single triggering conditions and lacks adaptation to multiple scenarios such as data integration set submission and file cleaning, resulting in poor flexibility and applicability of data processing.
[0005] Furthermore, existing methods lack sufficient coordination between data collected by VR motion capture devices and rehabilitation assessment models. The distribution of data analysis tasks and the feedback of results lack standardized processes, and the control mechanism for data processing permissions is inadequate, making data leakage or misoperation prone to occur. These shortcomings collectively result in existing VR human dynamic rehabilitation assessment methods failing to meet the actual needs of clinical rehabilitation treatment in terms of assessment accuracy, efficiency, and security. Therefore, there is an urgent need for a VR human dynamic rehabilitation assessment method that can achieve accurate data collection, intelligent data filtering and optimization, and multi-scenario adaptation to solve these problems. Summary of the Invention
[0006] To address the technical problems existing in related technologies, this application provides a method and system for dynamic human rehabilitation assessment using VR.
[0007] Firstly, a method for dynamic human rehabilitation assessment using VR is provided, the method comprising: Obtain several human dynamic standard data from the storage range of the human dynamic database, wherein the human dynamic standard data includes human age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data; Human age data and first joint angle torsion data are collected by VR motion capture device. Target dynamic dataset is determined in the human dynamic database based on the human age data and first joint angle torsion data. The upload time of the target dynamic dataset exceeds a preset time period. The first joint angle torsion data of the target dynamic dataset corresponds to the first joint twisting state. The first joint twisting state indicates that the target dynamic dataset is a joint twisting state to be processed. Based on the target limb displacement data and target kinematic dynamic data corresponding to the target dynamic dataset, a data analysis task is sent to the VR rehabilitation assessment and analysis model, and the analysis results fed back by the VR rehabilitation assessment and analysis model are received. When the analysis result indicates that there is a description record of the target data integration set corresponding to the target dynamic dataset, the joint torsion state data set of the target dynamic dataset is converted into the second joint torsion state, and the second joint torsion state is used as sample data. When the analysis results indicate that there is no description record for the target data integration set, the joint torsion state data set of the target dynamic dataset is converted into a third joint torsion state, and the third joint torsion state is further processed.
[0008] In this application, when the analysis result indicates the existence of a description record for the target data integration set corresponding to the target dynamic dataset, after converting the joint torsion state data set of the target dynamic dataset into a second joint torsion state, the method further includes: Obtain the first data size of the target dynamic dataset, and obtain the first limb motion state information and limb motion trajectory information corresponding to the limb displacement data; Calculate the comparison result between the first limb movement state information and limb movement trajectory information and the first data scale, and determine the optimized first limb movement state information and limb movement trajectory information based on the comparison result.
[0009] In this application, after calculating the comparison result between the first limb movement state information and limb movement trajectory information and the first data scale, and determining the optimized first limb movement state information and limb movement trajectory information based on the comparison result, the method further includes: Obtain the description record of the target data integration set, and extract the second joint angle torsion data of the target dynamic dataset from the description record; When the second joint angle torsion data indicates that the target dynamic dataset has been cleaned, the joint torsion state data set of the target dynamic dataset is converted into the third joint torsion state; Based on the first data scale, the first limb movement state information and limb movement trajectory information are optimized.
[0010] In this application, obtaining the first limb movement state information and limb movement trajectory information corresponding to the limb displacement data includes: The data integration set corresponding to the limb displacement data is obtained to initiate the monitoring of the daily status information data of the object, and the normal movement description of the object is determined based on the daily status information data; wherein, the daily status information data corresponds to the information in the movement state, the first limb movement state information, and the limb movement trajectory information. Based on the limb displacement data, a real-time data analysis task is sent to the VR rehabilitation assessment and analysis model, and the VR rehabilitation assessment and analysis model sends feedback on several movement state folders corresponding to several second data scales. The normal motion descriptions of the remaining monitored objects are calculated based on the descriptions of the normal motion of the monitored objects and the several second data scales.
[0011] In this application, the step of calculating the comparison result between the first limb movement state information and limb movement trajectory information and the first data scale, and determining the optimized first limb movement state information and limb movement trajectory information based on the comparison result, includes: When the first limb movement state information and limb movement trajectory information are not less than the first data size, calculate the comparison result between the first limb movement state information and limb movement trajectory information and the first data size, and determine the optimized first limb movement state information and limb movement trajectory information based on the comparison result. When the first limb movement state information and limb movement trajectory information are less than the first data size, a warning data is sent to the data processing terminal of the monitoring object and the file upload monitoring object of the data integration set. The warning data indicates that the sample data of the movement state folder of the integration set is insufficient.
[0012] In this application, obtaining several sets of human dynamic standard data includes: Obtain historical human dynamic data; When the historical human dynamic data meets the triggering requirements of the pre-set joint twisting state data set management task, the human dynamic standard data of several data integration set motion state folders are obtained from the human dynamic database storage range.
[0013] In this application, obtaining several sets of human dynamic standard data includes: Receive a joint torsion state data set optimization task, wherein the joint torsion state data set optimization task has pre-defined management permissions; When the pre-set management permissions meet the management permission requirements, several individual human dynamic standard data are obtained.
[0014] In this application, the method further includes: The system receives a joint torsion state data set optimization task sent by the VR rehabilitation assessment and analysis model. The joint torsion state data set optimization task includes specified limb displacement data, specified motion trajectory data, and target optimization operation. The file to be optimized is determined based on the specified limb displacement data and the specified motion trajectory data; The target optimization operation is performed on the set of joint torsional state data of the file to be optimized.
[0015] In this application, the process of generating the joint torsional state data set optimization task includes the following steps: When a data integration set submission request is received, the specified limb displacement data and specified motion trajectory data covered in the integration set submission request are obtained; Perform file debugging on the file corresponding to the specified motion trajectory data to obtain the debugging results; Based on the debugging results, the target optimization operation is determined, and a joint twisting state data set optimization task is generated based on the specified limb displacement data, the specified motion trajectory data, and the target optimization operation.
[0016] In this application, determining the target optimization operation based on the debugging results includes: When the debugging result is qualified, the target optimization operation is determined to be to optimize the joint torsion state data set of the file corresponding to the specified motion trajectory data into the second joint torsion state; After performing the target optimization operation on the joint torsional state data set of the file to be optimized, the method further includes: The third data size required to obtain the file corresponding to the specified motion trajectory data, and the second limb motion state information and limb motion trajectory information corresponding to the specified limb displacement data; The second limb movement state information and limb movement trajectory information are optimized based on the comparison results between the second limb movement state information and the third data scale.
[0017] In this application, the process of generating the joint torsional state data set optimization task includes the following steps: When a file cleaning request is received, the specified limb displacement data and specified motion trajectory data covered in the file cleaning request are obtained; The file cleaning operation is identified as the target optimization operation, and a joint torsional state data set optimization task is generated based on the specified limb displacement data, the specified motion trajectory data, and the target optimization operation.
[0018] Secondly, a human dynamic rehabilitation assessment system for VR is provided, including a processor and a memory that communicate with each other, wherein the processor is used to read a computer program from the memory and execute it to implement the above-described method.
[0019] This application provides a VR-based human dynamic rehabilitation assessment method and system. It obtains several standard human dynamic data points stored in a human dynamic database. These standard data points include age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data. Based on the age data and the first joint angle torsion data, a target dynamic dataset is determined in the human dynamic database. The upload time of the target dynamic dataset exceeds a preset time period. The first joint angle torsion data corresponds to a first joint twisting state, indicating that the target dynamic dataset represents a joint twisting state to be processed. A data analysis task is sent to a VR rehabilitation assessment analysis model based on the limb displacement data and kinematic dynamic data, and the analysis results are received from the VR rehabilitation assessment analysis model. When the analysis results indicate the existence of a description record for a target data integration set corresponding to the target dynamic dataset, the joint twisting state data set of the target dynamic dataset is converted into a second joint twisting state. When the analysis results indicate the absence of a description record for a target data integration set, the joint twisting state data set of the target dynamic dataset is converted into a third joint twisting state, and the third joint twisting state is cleaned.
[0020] This embodiment of the disclosure obtains metadata from files in the integrated set of motion state folders management backend, and then selects target motion state folders based on the joint twisting status of the motion state folders in the metadata, identifying those whose upload time has reached a preset time but have not yet been stored on disk. It then requests the VR rehabilitation assessment and analysis model to submit the joint twisting status of the target motion state folders to the integrated set, and automatically optimizes the joint twisting status of the target motion state folders based on the submitted joint twisting status. This improves the optimization efficiency of the joint twisting status of the integrated set of motion state folders, thereby avoiding the problem of inaccurate assessment caused by low optimization efficiency of the joint twisting status of motion state folders. Attached Figure Description
[0021] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0022] Figure 1 This is a flowchart of a VR-based human dynamic rehabilitation assessment method provided in an embodiment of this application. Detailed Implementation
[0023] To better understand the above technical solutions, the technical solutions of this application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments of this application and the specific features in the embodiments are detailed descriptions of the technical solutions of this application, rather than limitations on the technical solutions of this application. In the absence of conflict, the embodiments of this application and the technical features in the embodiments can be combined with each other.
[0024] Please see Figure 1 This paper presents a method for dynamic human rehabilitation assessment in VR, which may include the technical solutions described in steps 310-350.
[0025] Step 310: Obtain several human dynamic standard data from the storage range of the human dynamic database.
[0026] In this application, human dynamic standard data refers to dynamic data of different genders and ages under healthy conditions, which serves as sample data in the database.
[0027] This application demonstrates the technology using VR, which allows for accurate observation of human body dynamics.
[0028] For a motion state folder in the first joint torsion state, the joint torsion state data set management method provided in this disclosure can be used to optimize its joint torsion state, so as to convert its joint torsion state into a second joint torsion state or a cleaned third joint torsion state. Specifically, a second joint torsion state indicates that the integrated set corresponding to the motion state folder has been submitted to the VR rehabilitation assessment and analysis model, while a third joint torsion state indicates that the integrated set corresponding to the motion state folder has not been submitted to the VR rehabilitation assessment and analysis model.
[0029] In this application, the human dynamic standard data is understood as follows: First, the system retrieves several human dynamic standard data from the human dynamic database within the storage range. The human dynamic standard data are baseline data of healthy people or patients at different rehabilitation stages that have been clinically verified in advance. Its data dimensions include: human age data (divided into age groups in 10-year intervals, such as 20-30 years old, 31-40 years old, etc.), kinematic dynamic data (including limb movement speed, acceleration, angular velocity), first limb displacement data (a three-dimensional coordinate system is established with the center of the human trunk as the origin, and the coordinate change data of each limb endpoint), and first joint angle torsion data (including the torsion angle range and rate of change of key rehabilitation assessment joints such as shoulder joint, elbow joint, and knee joint).
[0030] The specific implementation of this step can be carried out in accordance with claim 6 or 7: If claim 6 is adopted, the system first acquires historical human dynamic data (more than 100,000 data points accumulated from clinical rehabilitation assessments over the past 3 years). When the age range and joint movement type covered by the historical human dynamic data meet the preset joint twisting state data set management task trigger requirements (such as covering more than 8 age ranges and including more than 5 core rehabilitation movements), the system extracts the corresponding data integration set movement state folder (each folder corresponds to an age range + a type of rehabilitation movement) of human dynamic standard data from the human dynamic database. If claim 7 is adopted, the system receives the joint twisting state data set optimization task sent by the medical staff terminal (this task carries the management permission identifier of the medical staff). After verifying that the pre-set management permission meets the system's preset "data retrieval permission requirements" (such as attending physician and above permissions) through the permission verification module, the system extracts the corresponding human dynamic standard data from the database.
[0031] In some possible implementations, several human dynamic standard data are obtained, including: Obtain historical human dynamic data; When historical human dynamic data meets the triggering requirements of the pre-set joint twisting state data set management task, the human dynamic standard data of several data integration sets motion state folders are obtained from the storage range of the human dynamic database.
[0032] In this embodiment of the disclosure, the optimized management of the joint twisting state in the motion state folder in the motion state folder management server can be triggered in the form of a timed task.
[0033] In some possible implementations, several human dynamic standard data are obtained, including: Receive joint torsion state data set optimization task, the joint torsion state data set optimization task has pre-defined management permissions; When the pre-set management permissions meet the management permission requirements, several individual human dynamic standard data are obtained.
[0034] In this embodiment, the triggering of the joint torsion state optimization task in the motion state folder can also be triggered based on the joint torsion state data set optimization task. This joint torsion state data set optimization task can originate from the VR rehabilitation assessment and analysis model, from within the motion state folder management server, or from other management servers. When the joint torsion state data set optimization task originates from the VR rehabilitation assessment and analysis model, specifically, the VR rehabilitation assessment and analysis model generates the joint torsion state data set optimization task after receiving the integration set submission request, and then sends the joint torsion state data set optimization task to the motion state folder management server.
[0035] Step 320: Collect human age data and first joint angle torsion data using VR motion capture device, and determine the target dynamic dataset in the human dynamic database based on the human age data and the first joint angle torsion data.
[0036] The determination of the target dynamic dataset includes: the patient's age data (manually input and verified by the patient's terminal) and the first joint angle torsion data (the patient wears the VR motion capture device to perform preset rehabilitation movements, such as raising the upper limb and flexing and extending the lower limb; the device collects the torsion angle data of key joints in real time, with a sampling frequency of 100Hz). The system uses the collected patient age data and the first joint angle torsion data as search criteria to match the corresponding target dynamic dataset in the human dynamic database.
[0037] The selection criteria for the target dynamic dataset are: 1) The upload time exceeds the preset time period (the preset time period is 3 months, that is, data that has been stored in the database for more than 3 months and has not undergone secondary optimization processing is selected); 2) The first joint angle torsion data of the dataset corresponds to the first joint torsion state, and the first joint torsion state indicates that the target dynamic dataset is a joint torsion state to be processed (i.e., non-standard data that needs further analysis and optimization) according to the system's preset state judgment rules (such as the joint torsion angle exceeding the standard range of the corresponding age group by ±15% and the torsion rate fluctuation being greater than 20%).
[0038] In this application, the first joint angle torsion value is understood as raw data, and the second joint angle torsion value is understood as real-time data.
[0039] When the joint torsion state optimization task in the motion state folder management server is triggered, the server obtains the human dynamic standard data for all motion state folders from the human dynamic database storage range. It then extracts human age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data from this standard data. Based on the human age data and first joint angle torsion data for each motion state folder, the server selects the target dynamic dataset for joint torsion state optimization from several motion state folders. Specifically, the upload time of the human age data in the target dynamic dataset for joint torsion state optimization must be within a pre-set time period relative to real-time.
[0040] The failure of these motion status folders to be converted into disk-stored joint torsion states within the predetermined time period may be because the corresponding integrated set has not yet been submitted, or because the joint torsion state optimization task sent by the VR rehabilitation assessment and analysis model to the motion status folder management server when the corresponding integrated set was submitted was not effectively received. Therefore, in this embodiment of the disclosure, the motion status folder management server may further request the submitted joint torsion state data of the integrated set corresponding to the motion status folder from the VR rehabilitation assessment and analysis model, so as to optimize the joint torsion state of the motion status folder based on the submitted joint torsion state of the integrated set.
[0041] In some possible embodiments, the VR-based dynamic rehabilitation assessment method for humans provided in this application further includes: The system receives a joint torsion state data set optimization task sent by the VR rehabilitation assessment and analysis model. The joint torsion state data set optimization task includes specified limb displacement data, specified motion trajectory data, and target optimization operation. The file to be optimized is determined based on specified limb displacement data and specified motion trajectory data; Perform the target optimization operation on the set of joint torsion state data of the file to be optimized.
[0042] Furthermore, the VR rehabilitation assessment and analysis model can further generate a joint torsion state data set optimization task based on the metadata of the description records. This task optimizes the joint torsion states of the corresponding motion state folders in the motion state folder management server. The joint torsion state data set optimization task includes specified limb displacement data, specified motion trajectory data, and a target optimization operation. Unlike the timed joint torsion state data set optimization task of the motion state folder management server, which optimizes the joint torsion states of all motion state folders in all integrated sets, this task optimizes the joint torsion states of a specified motion state folder within a specified integrated set.
[0043] After receiving a joint torsional state data set optimization task that includes specified limb displacement data, specified motion trajectory data, and the target optimization operation, the motion state folder management server can first determine the files to be optimized from all motion state folders on the server based on the limb displacement data and the specified motion trajectory data. These files to be optimized are generally all motion state folders associated with kinematic dynamic data within the motion state folder corresponding to the limb displacement data. Once the files to be optimized are determined, the target optimization operation specified in the joint torsional state data set optimization task can be executed on the joint torsional state data set of these files.
[0044] In some possible implementations, the process of generating the joint torsional state dataset optimization task includes the following steps: When a data integration set submission request is received, the specified limb displacement data and specified motion trajectory data covered in the integration set submission request are obtained; Perform file debugging on the file corresponding to the specified motion trajectory data to obtain the debugging results; Based on the debugging results, the target optimization operation is determined, and the joint torsional state data set is generated to optimize the task based on the specified limb displacement data, the specified motion trajectory data, and the target optimization operation.
[0045] In some possible implementations, the target optimization operation is determined based on the debugging results, including: When the debugging result is qualified, the target optimization operation is determined to be to optimize the joint torsion state data set of the file corresponding to the specified motion trajectory data into the second joint torsion state; After performing the target optimization operation on the set of joint torsional state data of the file to be optimized, the following also includes: The third data size required to obtain the file corresponding to the specified motion trajectory data, and the second limb motion state information and limb motion trajectory information corresponding to the specified limb displacement data; The second limb movement state information and limb movement trajectory information are optimized based on the comparison results between the second limb movement state information and limb movement trajectory information and the third data scale.
[0046] Specifically, we can first obtain the required data size of the file corresponding to the specified motion trajectory data, which can be referred to here as the third data size, and then obtain the second limb motion state information and limb motion trajectory information corresponding to the specified limb displacement data. Then, we subtract the third data size from the second limb motion state information and limb motion trajectory information to obtain the optimized second limb motion state information and limb motion trajectory information. That is, we can calculate the comparison result between the second limb motion state information and limb motion trajectory information and the third data size, and optimize the second limb motion state information and limb motion trajectory information based on this comparison result.
[0047] In some possible implementations, the process of generating the joint torsional state dataset optimization task includes the following steps: When a file cleaning request is received, the specified limb displacement data and specified motion trajectory data covered in the file cleaning request are obtained; The file cleaning operation is identified as the target optimization operation, and the task is optimized by generating a set of joint torsional state data based on specified limb displacement data, specified motion trajectory data, and the target optimization operation.
[0048] When the motion state folder management server receives the joint torsion state data set optimization task, it can first find the motion state folder corresponding to the specified motion trajectory data in the motion state folder storage range, and then optimize the joint torsion state of the motion state folder from the second joint torsion state to the third joint torsion state.
[0049] Step 330: Send a data analysis task to the VR rehabilitation assessment and analysis model based on the target limb displacement data and target kinematic dynamic data corresponding to the target dynamic dataset, and receive the analysis results fed back by the VR rehabilitation assessment and analysis model.
[0050] The process of sending data analysis tasks and receiving analysis results includes: the system extracts the target limb displacement data (the three-dimensional coordinate change sequence of each limb endpoint when the patient completes rehabilitation movements) and target kinematic dynamic data (the velocity and acceleration change curves of the patient's limb movements) from the target dynamic dataset, encapsulates these data into data analysis tasks, and sends them to the VR rehabilitation assessment and analysis model (a CNN-LSTM fusion model based on deep learning, which is pre-trained and optimized using historical rehabilitation data).
[0051] The VR rehabilitation assessment and analysis model extracts features from the received data (extracting 128-dimensional feature vectors such as joint angle change features and limb displacement trajectory features), and queries the built-in target data integration set description record index library to determine whether there is a description record of the target data integration set that matches the current target dynamic dataset (the description record includes the ID of the data integration set, the corresponding rehabilitation action type, the data collection time, whether it has been cleaned and optimized, etc.), and feeds the judgment result back to the system as the analysis result.
[0052] Therefore, it is necessary to clean up the temporary motion status folders in a timely manner. Specifically, a joint torsion state optimization task can be initiated for motion status folders in the temporary storage area that have been stored for a longer period than a preset time. Then, these motion status folders can be cleaned up based on the optimized joint torsion state.
[0053] Specifically, in this embodiment of the disclosure, the joint torsional state data set of the target dynamic dataset can be optimized based on whether the integrated set corresponding to the target dynamic dataset has been submitted. Thus, after determining several target dynamic datasets based on the upload time of the motion state folder and the joint torsional state of the motion state folder, the target limb displacement data and target kinematic dynamic data corresponding to each target dynamic dataset can be obtained sequentially. After obtaining the target limb displacement data and target kinematic dynamic data from the target dynamic dataset, an analysis task can be generated based on this data and sent to the VR rehabilitation assessment and analysis model. Upon receiving the target limb displacement data and target kinematic dynamic data, the VR rehabilitation assessment and analysis model first searches the integrated set description records based on the limb displacement data. If no description record corresponding to the target limb displacement data is found, it reports that no corresponding integrated set description record exists.
[0054] Step 340: When the analysis results indicate the existence of a description record of the target data integration set corresponding to the target dynamic dataset, the joint torsion state data set of the target dynamic dataset is converted into the second joint torsion state.
[0055] The conversion and processing of the joint torsion state data set includes: when the analysis results indicate the existence of a corresponding description record of the target data set, the system converts the joint torsion state data set of the target dynamic dataset (including joint angle torsion time series data and torsion rate data) into a second joint torsion state (a standardized sample data format that can be used for model training, such as normalizing angle data to the [0,1] interval and extracting time series data into a fixed-length sequence) through the data conversion module, and stores the second joint torsion state as sample data in the sample database for subsequent iterative optimization of the VR rehabilitation assessment and analysis model; When the analysis results indicate that there is no corresponding description record for the target data integration set, the system converts the joint torsion state data set of the target dynamic dataset into the third joint torsion state (data format to be cleaned, marked as "unintegrated data"), and triggers subsequent further processing procedures (including data denoising, outlier removal, and data completion).
[0056] After receiving the analysis results from the VR rehabilitation assessment and analysis model, the motion state folder management server can further optimize the joint torsion state data set of the target dynamic dataset based on the analysis results. Specifically, when the analysis results indicate the existence of a description record for the target data integration set corresponding to the target dynamic dataset, it means that the integration set corresponding to the target dynamic dataset has been submitted. At this time, the target dynamic dataset can be stored on the disc, that is, the joint torsion state data set of the target dynamic dataset can be converted into the second joint torsion state that has been stored on the disc.
[0057] In some possible implementations, when the analysis results indicate the existence of a description record for a target data integration set corresponding to the target dynamic dataset, after converting the joint torsion state data set of the target dynamic dataset into a second joint torsion state, the method further includes: Obtain the first data size of the target dynamic dataset, and obtain the first limb motion state information and limb motion trajectory information corresponding to the limb displacement data; The comparison results between the first limb motion state information and limb motion trajectory information and the first data scale are calculated, and the optimized first limb motion state information and limb motion trajectory information are determined based on the comparison results.
[0058] In this embodiment of the disclosure, when the analysis results fed back by the VR rehabilitation assessment and analysis model indicate the existence of a description record of a target data integration set corresponding to the target dynamic dataset, not only can the joint twisting state data set of the target dynamic dataset be converted into a second joint twisting state that has been stored on a disc, but the target dynamic dataset can also be stored from the file temporary storage area to the motion state folder storage area. Furthermore, the capacity of the integration set motion state folder can be optimized based on the file size of the target dynamic dataset. Specifically, a first data size of the target dynamic dataset can be obtained first, along with first limb motion state information and limb motion trajectory information corresponding to the limb displacement data. Then, a comparison result between the first limb motion state information and limb motion trajectory information and the first data size is calculated, and the optimized first limb motion state information and limb motion trajectory information are determined based on this comparison result.
[0059] In some possible implementation embodiments, after calculating the comparison result between the first limb motion state information and the limb motion trajectory information and the first data size, and determining the optimized first limb motion state information and limb motion trajectory information based on the comparison result, the method further includes: Obtain the description record of the target data integration set, and extract the second joint angle torsion data of the target dynamic dataset from the description record; When the second joint angle torsion data indicates that the target dynamic dataset has been cleaned, the joint torsion state data set of the target dynamic dataset is converted into the third joint torsion state. The first limb movement state information and limb movement trajectory information are optimized based on the first data scale.
[0060] The acquisition of data scale and limb movement-related information includes: after converting the target dynamic dataset into the second joint twisting state, the system calculates the first data scale of the target dynamic dataset (i.e., the number of valid data entries contained in the dataset, excluding invalid data during the sampling process, such as missing data caused by device signal interruption) through the data statistics module, and extracts the first limb movement state information (such as "upper limb raising movement - normal joint activity" and "lower limb flexion and extension movement - limited knee joint activity") and limb movement trajectory information (movement trajectory curves drawn with coordinate sequences in a three-dimensional coordinate system, marking the smoothness of the trajectory and whether there are any stuttering points).
[0061] The acquisition of the first limb movement state information and limb movement trajectory information is achieved in accordance with the method of claim 4: 1) The system acquires the daily status information data of the monitoring object (i.e., the patient to be evaluated) corresponding to the data integration set of the limb displacement data (including the patient's daily exercise habits, past medical history, and current rehabilitation progress, which are entered into the system by medical staff). Based on these daily status information data, the system determines the description of the monitoring object's normal movement state (e.g., "a small amount of daily upper limb movement, limited range of motion of the shoulder joint") through a fuzzy matching algorithm. The daily status information data corresponds one-to-one with the first limb movement state information and limb movement trajectory information (e.g., less daily upper limb movement corresponds to a smaller range of upper limb movement trajectory). ); 2) Based on limb displacement data, send real-time data analysis tasks to the VR rehabilitation assessment and analysis model. The model returns several second data scales corresponding to several movement status folders (such as "normal upper limb raising movement for 20-30 years old" and "limited upper limb raising movement for 20-30 years old"). (The number of valid data entries in each folder is 1200 or 800). 3) By comparing the normal movement description of the monitored subjects with several second data scales, use interpolation algorithms to calculate the normal movement description of other monitored subjects (such as other patients with similar age and rehabilitation stage to the patient to be assessed) (such as "the average range of upper limb movement trajectory for patients in the same age group who have been rehabilitated for 2 months is ±30cm").
[0062] In this embodiment of the disclosure, after the analysis result indicates the existence of a description record corresponding to the target dynamic dataset, a set of joint torsion state data of the target dynamic dataset can be further obtained from the description record, which can be referred to here as the second joint angle torsion data. The second joint angle torsion data indicates the joint torsion state of the target dynamic dataset in the VR rehabilitation assessment analysis model, specifically divided into normal joint torsion state and cleaned joint torsion state. Specifically, the second joint angle torsion data can be determined by debugging the motion state folder covered in the data integration set when submitting the integration set, and based on the debugging results, if the debugging result of the motion state folder is qualified, the second joint angle torsion data can be determined to be normal; if the debugging result of the motion state folder is unqualified, the motion state folder needs to be cleaned, thus determining that the second joint angle torsion data in the motion state folder is a cleaned joint torsion state.
[0063] Thus, when the analysis results indicate the existence of a description record corresponding to the target dynamic dataset, the description record of the target data integration set can be further obtained, and the second joint angle torsion data of the target dynamic dataset can be extracted from the description record. If the second joint angle torsion data is a normal joint torsion state, then it is determined that the joint torsion state data set of the target dynamic dataset will be converted into the second joint torsion state, and the first data scale corresponding to the target dynamic dataset will be subtracted from the first limb motion state information and the limb motion trajectory information.
[0064] In some possible implementations, obtaining the first limb motion state information and limb motion trajectory information corresponding to the limb displacement data includes: The data collection corresponding to the limb displacement data is used to initiate the monitoring of the daily status information data of the monitored object, and the description of the normal movement of the monitored object is determined based on the daily status information data. Based on limb displacement data, a real-time data analysis task is sent to the VR rehabilitation assessment and analysis model, and several second data scales corresponding to several motion state folders are received from the VR rehabilitation assessment and analysis model. The normal motion descriptions of the remaining monitored objects are calculated based on the descriptions of the normal motion of the monitored objects and several second data scales.
[0065] Furthermore, real-time data analysis tasks can be sent to the VR rehabilitation assessment and analysis model based on limb displacement data to analyze all real-time data corresponding to the limb displacement data. All real-time data corresponding to the limb displacement data includes data from the movement status folders in the movement status folders where the joint twisting state is normal, uploaded by all participants who filled in the integration set. The real-time data covers the second data scale of the movement status folders.
[0066] After obtaining the normal motion description of the monitored object and the second data scale of several motion status folders, the comparison result of the normal motion description of the monitored object and the sum of several second data scales can be calculated to obtain the normal motion description of the remaining monitored objects.
[0067] In some possible implementation embodiments, the comparison result between the first limb motion state information and limb motion trajectory information and the first data size is calculated, and the optimized first limb motion state information and limb motion trajectory information are determined based on the comparison result, including: When the first limb movement state information and limb movement trajectory information are not less than the first data size, calculate the comparison result between the first limb movement state information and limb movement trajectory information and the first data size, and determine the optimized first limb movement state information and limb movement trajectory information based on the comparison result. When the first limb movement status information and limb movement trajectory information are less than the first data size, an early warning data is sent to the data processing terminal of the monitoring object and the file upload monitoring object of the data integration set. The early warning data indicates that the sample data of the movement status folder of the integration set is insufficient.
[0068] The data scale comparison and optimization includes: the system calculates the comparison results between the first limb movement state information and limb movement trajectory information and the first data scale according to the logic of claim 5, specifically: the threshold of the first data scale is set to 500 (preset based on clinical assessment experience to ensure statistical significance of the data); when the number of valid data points corresponding to the first limb movement state information and limb movement trajectory information is not less than 500, the difference between the two is calculated (e.g., 600 - 500 = 100); the optimization strategy is determined based on the difference (data downsampling is performed when the difference is greater than 100, retaining 500 representative data points). According to the data, when the difference is between 0 and 100, the original data is directly retained, and the optimized first limb movement status information and limb movement trajectory information are obtained. When the number of valid data corresponding to the first limb movement status information and limb movement trajectory information is less than 500, the system sends warning data to the patient terminal of the monitoring object (patient to be evaluated) and the medical terminal of the monitoring object (medical staff) through the message push module. The warning information is "Insufficient sample data in the movement status folder of the integration set (current valid data is 320, 180 need to be added), please complete 1 set of rehabilitation action collection again".
[0069] In this embodiment of the disclosure, when the analysis result fed back by the VR rehabilitation assessment and analysis model indicates that the description record of the target data integration set corresponding to the target dynamic dataset can be analyzed, and the joint twisting state data set of the target dynamic dataset is converted from the first joint twisting state to the second joint twisting state, when deducting the capacity of the motion state folder based on the first data size corresponding to the target dynamic dataset, that is, when optimizing the first limb motion state information and limb motion trajectory information, the first data size can be compared with the size of the first limb motion state information and limb motion trajectory information, and then different processing schemes can be determined according to the comparison result. Specifically, when the first limb motion state information and limb motion trajectory information are not less than the first data size, the comparison result between the first limb motion state information and limb motion trajectory information and the first data size can be calculated, and then the optimized first limb motion state information and limb motion trajectory information can be determined according to the comparison result. When the first limb motion state information and limb motion trajectory information are less than the first data size, early warning data can be sent to the data processing terminal of the monitoring object of the data integration set and the file upload monitoring object. The early warning data indicates that the sample data of the motion state folder of the integration set is insufficient.
[0070] Step 350: When the analysis results indicate that there is no description record for the target data integration set, the joint torsion state data set of the target dynamic dataset is converted into the third joint torsion state, and the third joint torsion state is cleaned.
[0071] Specifically, when the analysis results from the VR rehabilitation assessment and analysis model indicate that there is no description record of the target data integration set in the VR rehabilitation assessment and analysis model, that is, after the target dynamic dataset has been uploaded to the motion state folder management server for more than a pre-set time period, the file upload monitoring object of the target dynamic dataset has still not submitted the data integration set to the VR rehabilitation assessment and analysis model, then the joint torsion state data set of the target dynamic dataset can be converted from the unprocessed joint torsion state to the cleaned joint torsion state. At the same time, the temporarily stored target dynamic dataset can be cleaned to avoid long-term occupation of the file temporary storage space.
[0072] In this embodiment, the target dynamic dataset can be one or several. When there are several target dynamic datasets, the target limb displacement data and target kinematic dynamic data corresponding to each target dynamic dataset can be obtained one by one. Then, based on the target limb displacement data and target kinematic dynamic data corresponding to each target dynamic dataset, a data analysis task is sent to the VR rehabilitation assessment and analysis model to obtain the analysis results corresponding to each target dynamic dataset fed back by the VR rehabilitation assessment and analysis model. Then, the joint torsional state data set of each target dynamic dataset can be optimized based on the analysis results corresponding to each target dynamic dataset.
[0073] The VR-based human dynamic rehabilitation assessment method provided in this embodiment obtains several human dynamic standard data from a human dynamic database. These standard data include human age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data. A target dynamic dataset is determined in the human dynamic database based on the human age data and the first joint angle torsion data. The upload time of the target dynamic dataset exceeds a preset time period. The first joint angle torsion data corresponds to a first joint twisting state, and the first joint twisting state indicates that the target dynamic dataset is a joint twisting state to be processed. A data analysis task is sent to a VR rehabilitation assessment analysis model based on the limb displacement data and kinematic dynamic data, and the analysis results fed back by the VR rehabilitation assessment analysis model are received. When the analysis result indicates the existence of a description record for a target data integration set corresponding to the target dynamic dataset, the joint twisting state data set of the target dynamic dataset is converted into a second joint twisting state. When the analysis result indicates the absence of a description record for a target data integration set, the joint twisting state data set of the target dynamic dataset is converted into a third joint twisting state, and the third joint twisting state is cleaned.
[0074] This embodiment of the disclosure obtains metadata from the files in the integrated set of motion state folders management backend, and then selects target motion state folders based on the joint twisting status of the motion state folders in the metadata, which have reached the upload time of a preset time but have not yet been stored on disk. Then, it requests the integrated set of joint twisting status of the target motion state folders from the VR rehabilitation assessment and analysis model, and automatically optimizes the joint twisting status of the target motion state folders based on the submitted joint twisting status of the integrated set. This improves the optimization efficiency of the joint twisting status of the integrated set of motion state folders, thereby avoiding the problem of inaccurate assessment caused by low optimization efficiency of the joint twisting status of motion state folders.
[0075] Based on the above, a human dynamic rehabilitation assessment system for VR is shown, including a processor and a memory that communicate with each other. The processor is used to read computer programs from the memory and execute them to implement the above-described method.
[0076] Based on the above, a computer-readable storage medium is also provided, on which a computer program stored implements the above method during runtime.
[0077] In summary, based on the above scheme, several standard human dynamic data are obtained from the storage range of the human dynamic database. These standard data include human age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data. A target dynamic dataset is determined in the human dynamic database based on the human age data and the first joint angle torsion data. The upload time of the target dynamic dataset exceeds a preset time period. The first joint angle torsion data corresponds to the first joint twisting state, and the first joint twisting state indicates that the target dynamic dataset is a joint twisting state to be processed. A data analysis task is sent to the VR rehabilitation assessment and analysis model based on the limb displacement data and kinematic dynamic data, and the analysis results fed back by the VR rehabilitation assessment and analysis model are received. When the analysis results indicate the existence of a description record for the target data integration set corresponding to the target dynamic dataset, the joint twisting state data set of the target dynamic dataset is converted into a second joint twisting state. When the analysis results indicate the absence of a description record for the target data integration set, the joint twisting state data set of the target dynamic dataset is converted into a third joint twisting state, and the third joint twisting state is cleaned.
[0078] This embodiment of the disclosure obtains metadata from files in the integrated set of motion state folders management backend, and then selects target motion state folders based on the joint twisting status of the motion state folders in the metadata, identifying those whose upload time has reached a preset time but have not yet been stored on disk. It then requests the VR rehabilitation assessment and analysis model to submit the joint twisting status of the target motion state folders to the integrated set, and automatically optimizes the joint twisting status of the target motion state folders based on the submitted joint twisting status. This improves the optimization efficiency of the joint twisting status of the integrated set of motion state folders, thereby avoiding the problem of inaccurate assessment caused by low optimization efficiency of the joint twisting status of motion state folders.
[0079] It should be understood that the systems and modules described above can be implemented in various ways. For example, in some embodiments, the systems and modules can be implemented by hardware, software, or a combination of both. The hardware portion can be implemented using dedicated logic; the software portion can be stored in memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated-design hardware. Those skilled in the art will understand that the methods and systems described above can be implemented using computer-executable instructions and / or included in processor control code, for example, such code provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The systems and modules of this application can be implemented not only by hardware circuits such as very large-scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field-programmable gate arrays, programmable logic devices, etc., but also by software executed by various types of processors, or by a combination of the aforementioned hardware circuits and software (e.g., firmware).
[0080] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.
Claims
1. A method for dynamic human rehabilitation assessment using VR, characterized in that, The method includes: Obtain several human dynamic standard data from the storage range of the human dynamic database, wherein the human dynamic standard data includes human age data, kinematic dynamic data, first limb displacement data, and first joint angle torsion data; Human age data and first joint angle torsion data are collected by VR motion capture device. Target dynamic dataset is determined in the human dynamic database based on the human age data and first joint angle torsion data. The upload time of the target dynamic dataset exceeds a preset time period. The first joint angle torsion data of the target dynamic dataset corresponds to the first joint twisting state. The first joint twisting state indicates that the target dynamic dataset is a joint twisting state to be processed. Based on the target limb displacement data and target kinematic dynamic data corresponding to the target dynamic dataset, a data analysis task is sent to the VR rehabilitation assessment and analysis model, and the analysis results fed back by the VR rehabilitation assessment and analysis model are received. When the analysis result indicates that there is a description record of the target data integration set corresponding to the target dynamic dataset, the joint torsion state data set of the target dynamic dataset is converted into the second joint torsion state, and the second joint torsion state is used as sample data. When the analysis results indicate that there is no description record for the target data integration set, the joint torsion state data set of the target dynamic dataset is converted into a third joint torsion state, and the third joint torsion state is further processed.
2. The method as described in claim 1, characterized in that, When the analysis result indicates the existence of a description record for the target data integration set corresponding to the target dynamic dataset, after converting the joint torsion state data set of the target dynamic dataset into a second joint torsion state, the method further includes: Obtain the first data size of the target dynamic dataset, and obtain the first limb motion state information and limb motion trajectory information corresponding to the limb displacement data; Calculate the comparison result between the first limb movement state information and limb movement trajectory information and the first data scale, and determine the optimized first limb movement state information and limb movement trajectory information based on the comparison result.
3. The method as described in claim 2, characterized in that, After calculating the comparison result between the first limb movement state information and limb movement trajectory information and the first data scale, and determining the optimized first limb movement state information and limb movement trajectory information based on the comparison result, the method further includes: Obtain the description record of the target data integration set, and extract the second joint angle torsion data of the target dynamic dataset from the description record; When the second joint angle torsion data indicates that the target dynamic dataset has been cleaned, the joint torsion state data set of the target dynamic dataset is converted into the third joint torsion state; The first limb movement state information and limb movement trajectory information are optimized based on the first data scale.
4. The method as described in claim 2, characterized in that, The step of obtaining the first limb movement state information and limb movement trajectory information corresponding to the limb displacement data includes: The data integration set corresponding to the limb displacement data is obtained to initiate the monitoring of the daily status information data of the object, and the normal movement description of the object is determined based on the daily status information data; wherein, the daily status information data corresponds to the information in the movement state, the first limb movement state information, and the limb movement trajectory information. Based on the limb displacement data, a real-time data analysis task is sent to the VR rehabilitation assessment and analysis model, and the VR rehabilitation assessment and analysis model sends feedback on several movement state folders corresponding to several second data scales. The normal motion descriptions of the remaining monitored objects are calculated based on the descriptions of the normal motion of the monitored objects and the several second data scales.
5. The method as described in claim 2, characterized in that, The step of calculating the comparison result between the first limb movement state information and limb movement trajectory information and the first data size, and determining the optimized first limb movement state information and limb movement trajectory information based on the comparison result, includes: When the first limb movement state information and limb movement trajectory information are not less than the first data size, calculate the comparison result between the first limb movement state information and limb movement trajectory information and the first data size, and determine the optimized first limb movement state information and limb movement trajectory information based on the comparison result. When the first limb movement state information and limb movement trajectory information are less than the first data size, a warning data is sent to the data processing terminal of the monitoring object and the file upload monitoring object of the data integration set. The warning data indicates that the sample data of the movement state folder of the integration set is insufficient.
6. The method as described in claim 1, characterized in that, The acquisition of several individual human dynamic standard data includes: Obtain historical human dynamic data; When the historical human dynamic data meets the triggering requirements of the pre-set joint twisting state data set management task, the human dynamic standard data of several data integration set motion state folders are obtained from the human dynamic database storage range.
7. The method as described in claim 1, characterized in that, The acquisition of several individual human dynamic standard data includes: Receive a joint torsion state data set optimization task, wherein the joint torsion state data set optimization task has pre-defined management permissions; When the pre-set management permissions meet the management permission requirements, several individual human dynamic standard data are obtained.
8. The method as described in claim 1, characterized in that, The method further includes: The system receives a joint torsion state data set optimization task sent by the VR rehabilitation assessment and analysis model. The joint torsion state data set optimization task includes specified limb displacement data, specified motion trajectory data, and target optimization operation. The file to be optimized is determined based on the specified limb displacement data and the specified motion trajectory data; The target optimization operation is performed on the set of joint torsional state data of the file to be optimized.
9. The method as described in claim 8, characterized in that, The process of generating the joint torsional state data set optimization task includes the following steps: When a data integration set submission request is received, the specified limb displacement data and specified motion trajectory data covered in the integration set submission request are obtained; Perform file debugging on the file corresponding to the specified motion trajectory data to obtain the debugging results; The target optimization operation is determined based on the debugging results, and a joint twisting state data set optimization task is generated based on the specified limb displacement data, the specified motion trajectory data, and the target optimization operation. The step of determining the target optimization operation based on the debugging results includes: When the debugging result is qualified, the target optimization operation is determined to be to optimize the joint torsion state data set of the file corresponding to the specified motion trajectory data into the second joint torsion state; After performing the target optimization operation on the joint torsional state data set of the file to be optimized, the method further includes: The third data size required to obtain the file corresponding to the specified motion trajectory data, and the second limb motion state information and limb motion trajectory information corresponding to the specified limb displacement data; The second limb movement state information and limb movement trajectory information are optimized based on the comparison results between the second limb movement state information and limb movement trajectory information and the third data scale. The process of generating the joint torsional state data set optimization task includes the following steps: When a file cleaning request is received, the specified limb displacement data and specified motion trajectory data covered in the file cleaning request are obtained; The file cleaning operation is identified as the target optimization operation, and a joint torsional state data set optimization task is generated based on the specified limb displacement data, the specified motion trajectory data, and the target optimization operation.
10. A human dynamic rehabilitation assessment system for VR, characterized in that, The method includes a processor and a memory that communicate with each other, the processor being configured to read a computer program from the memory and execute it to implement the method of any one of claims 1-9.