User characteristic data effective acquisition and analysis method for lower limb hip and knee rehabilitation

Abstract

The invention relates to the field of lower limb hip and knee data analysis, and discloses a user feature data effective acquisition and analysis method for lower limb hip and knee rehabilitation, which comprises the following steps: acquiring related data for movement at the kitchen hip and knee of a user and performing data preprocessing; meanwhile, a convolutional neural network is used for conducting modeling analysis on the preprocessed related data, and a multi-channel sequence gait detection algorithm model is designed and constructed and used for presetting the next movement of the lower limbs, the hips and the knees of the user. And finally, performing model loss training optimization on the multi-channel sequence gait detection algorithm model, and introducing a scoring mechanism to enable the multi-channel sequence gait detection algorithm model to perform gait scoring on the target user. According to the invention, simultaneous input of multiple channel data can be realized, detection of the current state and gait of the human body can be realized, and the effect of still keeping recognition without switching modes under the condition that the channel data is missing can also be realized.

Description

Technical Field

[0001] The present invention relates to the field of lower limb hip and knee data analysis, and particularly to an effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation. Background Art

[0002] With the advent of population aging, the number of elderly people in China is currently increasing. As of 2024, the number of people aged 60 and above has approached 300 million. People's demand for sports health is becoming increasingly strong. However, neurological diseases such as stroke are seriously affecting people's health. With the improvement of people's living standards, the incidence of stroke has further increased and shows a trend of getting younger. Therefore, the prevention and treatment of stroke and the rehabilitation of motor function after stroke are particularly important, especially the motor rehabilitation of the lower limbs, which is crucial for people's ability to move freely and is even more prominent. At the same time, problems such as trauma and disability can also lead to motor dysfunction, further increasing people's demand for rehabilitation training. Traditional methods use manual techniques to achieve the limb rehabilitation of patients, but traditional techniques are inefficient and have poor effects. Therefore, rehabilitation robots have gradually emerged, with effects comparable to those of humans, higher efficiency, and greater comfort.

[0003] An effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation is proposed, which is used to achieve the simultaneous input of multi-channel data, detect the current state and gait of the human body, and can also adapt to remain recognized even in the case of missing channel data without the need to switch modes. Summary of the Invention

[0004] The present invention overcomes the deficiencies of the prior art and provides an effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation.

[0005] To achieve the above object, the technical solution adopted by the present invention is as follows: The first aspect of the present invention provides an effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation, including the following steps: Deploy sensors at the hip and knee of the user's lower limbs, and control the sensors to collect relevant data of the user's lower limb hip and knee. At the same time, perform data preprocessing on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data; Based on a convolutional neural network, in combination with the preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data, design a multi-channel sequence gait detection algorithm model; Perform model loss training optimization on the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism to enable the multi-channel sequence gait detection algorithm model to perform gait scoring on the target user.

[0006] Further, in a preferred embodiment of the present invention, sensors are deployed at the hip and knee of the user's lower limbs, and the sensors are controlled to collect relevant data of the user's lower limb hip and knee. Meanwhile, data preprocessing is performed on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data, specifically as follows: The user who needs lower limb hip and knee rehabilitation is designated as the target user, and the positions on the lower limb hip and knee of the target user where sensors need to be deployed are determined and designated as the target deployment positions. Among them, the sensors include pressure sensors and surface electromyography sensors, and the positions where the sensors are deployed include the hip joint, knee joint, and foot of the target user; A pressure sensor and a surface electromyography sensor are respectively installed at the target deployment positions, and a standard time for lower limb hip and knee rehabilitation testing is preset and designated as the target time; Within the target time, based on the pressure sensor and the surface electromyography sensor, pressure data and surface electromyography data at the target deployment positions of the target user are collected in real time. Among them, the pressure data includes the magnitude and distribution of the pressure, and the surface electromyography data includes the bioelectric current value, angular velocity, and acceleration of the muscles at the target deployment positions; When collecting the pressure data and surface electromyography data at the target deployment positions of the target user, the collection times of different pressure data and surface electromyography data are generated simultaneously. Based on the collection times of different pressure data and surface electromyography data, collection timestamps are generated; Based on the collection timestamps, the collected different pressure data and surface electromyography data are aligned in time consistency, and the collection frequencies of the pressure sensor and the surface electromyography sensor are controlled to be equal. Meanwhile, data preprocessing and data enhancement processing are performed on the collected pressure data and surface electromyography data.

[0007] Further, in a preferred embodiment of the present invention, the data preprocessing and data enhancement processing of the collected pressure data and surface electromyography data are specifically as follows: The collected pressure data and surface electromyography data are collectively referred to as lower limb hip and knee data. The Kalman filter algorithm is introduced to perform Kalman filtering on the lower limb hip and knee data, so that high-frequency noise is removed from the lower limb hip and knee data to obtain lower limb hip and knee filtered data; Gravity compensation separation is performed on the lower limb hip and knee filtered data to obtain the bioelectric current value, acceleration, and angular velocity of the surface electromyography data in the lower limb hip and knee filtered data, which are designated as preprocessed bioelectric current values, preprocessed acceleration, and preprocessed angular velocity. Meanwhile, pressure normalization processing is performed on the lower limb hip and knee filtered data to obtain preprocessed pressure data.

[0008] Further, in a preferred embodiment of the present invention, based on the convolutional neural network, combining the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data, a multi-channel sequence gait detection algorithm model is designed, specifically as follows: Introduce a convolutional neural network to construct a blank convolutional neural network model, where the blank convolutional neural network model includes a convolutional layer, a pooling layer, and an output layer; Collectively refer to the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data as target user feature data, import the target user feature data into the convolutional layer of the blank convolutional neural network model, and construct a multi-channel time series matrix for the target user feature data in the convolutional layer of the blank convolutional neural network model; Among them, the multi-channel time series matrix of the target user feature data is a matrix for spatio-temporal alignment of the target user feature data; In the convolutional layer of the blank convolutional neural network model, traverse and convolve the multi-channel time series matrix, and keep the time series length of the multi-channel time series matrix equal during the traversal convolution process; After the traversal convolution is completed, import the traversed and convolved multi-channel time series matrix into the pooling layer for global time series pooling to match the target user to adapt to different gaits, and at the same time introduce a multi-channel attention mechanism into the blank convolutional neural network model during the global time series pooling process; Among them, the multi-channel attention mechanism automatically focuses on the traversed and convolved multi-channel time series matrix during the global time series pooling process to prevent data loss; When the global time series pooling is completed, output the trained multi-channel time series matrix in the output layer to obtain a multi-channel sequence gait detection algorithm model; Among them, the multi-channel sequence gait detection algorithm model can predict the gait state of the target user within a specified time.

[0009] Further, in a preferred embodiment of the present invention, the multi-channel sequence gait detection algorithm model is trained and optimized for model loss, and at the same time a scoring mechanism is introduced to enable the multi-channel sequence gait detection algorithm model to score the gait of the target user, specifically as follows: Run the multi-channel sequence gait detection algorithm model, introduce smooth cross-entropy, and use the smooth cross-entropy as the loss function of the multi-channel sequence gait detection algorithm model. Through the loss function, perform loss training on the multi-channel sequence gait detection algorithm model, where the loss training is to perform class balance training on the prediction results using smooth cross-entropy; Introduce the cosine annealing algorithm. During loss training, calculate the cosine annealing learning rate of the prediction result in real time through the cosine annealing algorithm, and preset the standard cosine annealing learning rate threshold. If the cosine annealing learning rate of the prediction result reaches the standard cosine annealing learning rate threshold, stop the loss training to obtain the prediction result after loss training, which is calibrated as a type of predicted user feature data. At the same time, calibrate the multi-channel sequence gait detection algorithm model after loss training as the target multi-channel sequence gait detection algorithm model; Determine the timestamp corresponding to the type of predicted user feature data, which is calibrated as a type of timestamp, and collect in real time the real feature data corresponding to the target user at the type of timestamp, which is calibrated as a type of real user feature data; Among them, the type of predicted user feature data is the bioelectric current value, acceleration, angular velocity, and pressure data of the lower limb hip and knee of the predicted target user within a specific time, and the type of real user feature data is the real data of the bioelectric current value, acceleration, angular velocity, and pressure data of the lower limb hip and knee of the target user within a specific time; Calculate the Euclidean distance between the type of predicted user feature data and the type of real user feature data, which is calibrated as a type of Euclidean distance. If the type of Euclidean distance is maintained within the preset value, divide the target multi-channel sequence gait detection algorithm model into a qualified multi-channel sequence gait detection algorithm model; If the type of Euclidean distance is not maintained within the preset value, divide the target multi-channel sequence gait detection algorithm model into a model to be analyzed, perform edge computing optimization on the model to be analyzed, and introduce a scoring mechanism so that the qualified multi-channel sequence gait detection algorithm model can perform gait scoring on the target user.

[0010] Furthermore, in a preferred embodiment of the present invention, the performing edge computing optimization on the model to be analyzed and introducing a scoring mechanism so that the qualified multi-channel sequence gait detection algorithm model can perform gait scoring on the target user is specifically as follows: Perform lightweight processing on the edge computing of the model to be analyzed. Among them, the lightweight processing of the edge computing of the model is to first perform channel pruning processing on the model to be analyzed, and introduce the singular value decomposition algorithm to perform singular value decomposition on the model to be analyzed after channel pruning processing to reduce the calculation amount and memory occupancy rate of the model to be analyzed and obtain the target model to be analyzed; Perform robustness verification on the target model to be analyzed. Among them, the robustness verification is to partially mask the feature data of the target user generated by the target model to be analyzed, and perform coincidence analysis on the feature data of the target user obtained after partial masking and the complete data; If the coincidence degree is greater than the preset value, output the target model to be analyzed, and classify it as a qualified multi-channel sequence gait detection algorithm model. If the coincidence degree is not greater than the preset value, continue to perform lightweight processing on the model edge calculation of the target model to be analyzed until a qualified multi-channel sequence gait detection algorithm model is generated; Introduce a real-time scoring mechanism into the qualified multi-channel sequence gait detection algorithm model, and construct a user gait scoring table, which indicates the scores corresponding to different user feature data; Output user feature data from the qualified multi-channel sequence gait detection algorithm model, label it as qualified user feature data, and perform gait state scoring on the qualified user feature data based on the user gait scoring table.

[0011] The second aspect of the present invention also provides a system for effectively collecting and analyzing user feature data for lower limb hip and knee rehabilitation. The system for effectively collecting and analyzing user feature data includes a memory and a processor. The memory stores a method for effectively collecting and analyzing user feature data. When the method for effectively collecting and analyzing user feature data is executed by the processor, the following steps are implemented: Deploy sensors at the hip and knee of the user's lower limbs, control the sensors to collect relevant data of the user's lower limb hip and knee, and at the same time perform data preprocessing on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed accelerations, preprocessed angular velocities, and preprocessed pressure data; Based on a convolutional neural network, combine the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data to design a multi-channel sequence gait detection algorithm model; Perform model loss training and optimization on the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user.

[0012] The present invention solves the technical defects existing in the background technology. The present invention has the following beneficial effects: collect relevant data for movement at the hip and knee of the user's lower limb and perform data preprocessing, and at the same time use a convolutional neural network to model and analyze the preprocessed relevant data, design and construct a multi-channel sequence gait detection algorithm model for the next action of the lower limb hip and knee of the preset user. Finally, perform model loss training and optimization on the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user. The present invention can realize the simultaneous input of multiple-channel data, realize the detection of the current state and gait of the human body, and can also realize the recognition even when channel data is missing without switching modes. Description of the Drawings

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0014] Figure 1 Shows a flowchart of an effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation; Figure 2 Shows a flowchart of a method for optimizing the model loss training of a multi-channel sequence gait detection algorithm model; Figure 3 Shows a program view of a system for effective acquisition and analysis of user characteristic data for lower limb hip and knee rehabilitation. Detailed implementation manners

[0015] In order to more clearly understand the above objects, features, and advantages of the present invention, the present invention will be further described in detail below in conjunction with the drawings and specific implementation manners. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0016] In the following description, many specific details are set forth in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited by the specific embodiments disclosed below.

[0017] Figure 1 Shows a flowchart of an effective acquisition and analysis method for user characteristic data for lower limb hip and knee rehabilitation, including the following steps: S102: Deploy sensors at the hip and knee of the user's lower limb, control the sensors to collect relevant data of the user's lower limb hip and knee, and at the same time perform data preprocessing on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data; S104: Based on a convolutional neural network, combine the preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data to design a multi-channel sequence gait detection algorithm model; S106: Perform model loss training optimization on the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user.

[0018] Further, in a preferred embodiment of the present invention, sensors are deployed at the hip and knee joints of the user's lower limbs, and the sensors are controlled to collect relevant data of the user's lower limb hip and knee. Meanwhile, data preprocessing is performed on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data. Specifically: The user who needs lower limb hip and knee rehabilitation is designated as the target user, and the positions on the lower limb hip and knee of the target user where sensors need to be deployed are determined and designated as the target deployment positions. Among them, the sensors include pressure sensors and surface electromyography sensors, and the positions where the sensors are deployed include the hip joint, knee joint, and foot of the target user; A pressure sensor and a surface electromyography sensor are respectively installed at the target deployment positions, and a standard time for lower limb hip and knee rehabilitation testing is preset and designated as the target time; Within the target time, based on the pressure sensor and the surface electromyography sensor, pressure data and surface electromyography data at the target deployment positions of the target user are collected in real time. Among them, the pressure data includes the magnitude and distribution of pressure, and the surface electromyography data includes the bioelectric current value, angular velocity, and acceleration of the muscles at the target deployment positions; When collecting the pressure data and surface electromyography data at the target deployment positions of the target user, the collection times of different pressure data and surface electromyography data are generated simultaneously. Based on the collection times of different pressure data and surface electromyography data, collection timestamps are generated; Based on the collection timestamps, the collected different pressure data and surface electromyography data are aligned in time consistency, and the collection frequencies of the pressure sensor and the surface electromyography sensor are controlled to be equal. Meanwhile, data preprocessing and data enhancement processing are performed on the collected pressure data and surface electromyography data.

[0019] It should be noted that lower limb hip and knee rehabilitation requires the assistance of a robotic arm. The robotic arm-assisted rehabilitation needs to collect human body data to assist the lower limb hip and knee movement to achieve rehabilitation. Human body data is collected and data prediction is performed to predict the next movement of the human body, and the robotic arm drives the lower limb hip and knee movement to achieve the role of assisted rehabilitation. First, data at the joints of the lower limb hip and knee are collected. Since the joints are the parts that drive the human body to move forward, pressure sensors and surface electromyography sensors are installed for data collection. Among them, the surface electromyography sensors are attached to key muscle groups such as the rectus femoris and biceps femoris to capture the muscle activation timing, etc., and the bioelectric current value, angular velocity, and acceleration of the muscles at the target deployment positions are collected, which are the necessary condition data for prediction. The pressure data is embedded in the insole or the ground force plate to record the plantar pressure distribution data. The purpose of time synchronization for all the collected data is to ensure the consistency and authenticity during data transmission and data analysis. At the same time, it is necessary to maintain the same sampling frequency to prevent adjusting the time synchronization at all times.

[0020] Furthermore, in a preferred embodiment of the present invention, the data preprocessing and data enhancement processing of the collected pressure data and surface electromyography data are specifically as follows: Collectively refer to the collected pressure data and surface electromyography data as lower limb hip-knee data, introduce the Kalman filter algorithm, perform Kalman filtering on the lower limb hip-knee data to remove high-frequency noise from the lower limb hip-knee data, and obtain lower limb hip-knee filtered data; Perform gravity compensation separation on the lower limb hip-knee filtered data to obtain the bioelectric current value, acceleration, and angular velocity of the surface electromyography data in the lower limb hip-knee filtered data, which are calibrated as the preprocessed bioelectric current value, preprocessed acceleration, and preprocessed angular velocity. At the same time, perform pressure normalization processing on the lower limb hip-knee filtered data to obtain preprocessed pressure data.

[0021] It should be noted that the purpose of filtering and enhancing the lower limb hip-knee data is to reduce the error rate of the data during the modeling process and improve the accuracy. The Kalman filter algorithm is an algorithm that can remove high-frequency noise from the data. After filtering, there may be intersections in the data, and the data needs to be redistributed. Therefore, the gravity compensation separation method is used to separate the bioelectric current value, acceleration, and angular velocity of the surface electromyography data.

[0022] Furthermore, in a preferred embodiment of the present invention, the multi-channel sequence gait detection algorithm model is designed based on the convolutional neural network, combined with the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data, specifically as follows: Introduce the convolutional neural network and construct a blank convolutional neural network model, where the blank convolutional neural network model includes a convolutional layer, a pooling layer, and an output layer; Collectively refer to the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data as target user feature data, import the target user feature data into the convolutional layer of the blank convolutional neural network model, and construct a multi-channel time series matrix for the target user feature data in the convolutional layer of the blank convolutional neural network model; Among them, the multi-channel time series matrix of the target user feature data is a matrix for spatio-temporal alignment of the target user feature data; In the convolutional layer of the blank convolutional neural network model, perform traversal convolution on the multi-channel time series matrix, and keep the time series length of the multi-channel time series matrix equal during the traversal convolution process; After the traversal convolution is completed, import the traversed multi-channel time series matrix into the pooling layer for global time series pooling to match different gaits of the target user. At the same time, introduce a multi-channel attention mechanism into the blank convolutional neural network model during the global time series pooling process; Among them, the multi-channel attention mechanism automatically focuses on traversing the multi-channel time series matrix after convolution during the global temporal pooling process to prevent data loss; When the global temporal pooling is completed, the trained multi-channel time series matrix is output at the output layer to obtain a multi-channel sequence gait detection algorithm model; Among them, the multi-channel sequence gait detection algorithm model can predict the gait state of the target user within a specified time.

[0023] It should be noted that the multi-channel sequence gait detection algorithm model can simultaneously process the lower limb hip and knee data of the user in multiple channels to achieve the detection of the current state and gait of the human body, and can also adapt to remain recognized in the case of missing channel data without the need to switch modes. The convolutional neural network is a prediction algorithm used to construct a multi-channel sequence gait detection model. The multi-channel sequence gait detection model has a convolutional layer and a pooling layer. The convolutional layer is the training layer used to train the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data. The training purpose is to judge the gait of the user in the next period of time. The multi-channel time series matrix is constructed by the convolutional layer and is an important architecture in the convolutional layer, which contains all the preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity, and preprocessed pressure data. The purpose of keeping the temporal lengths of the multi-channel time series matrix equal is to ensure that there are no errors during the pooling process. The purpose of global temporal pooling is to adapt to the different walking speeds and gaits of the target user, and at the same time introduce a multi-channel attention mechanism to prevent data loss and achieve the purpose of automatically focusing on effective information.

[0024] Figure 2 The method flow chart for training and optimizing the model loss of the multi-channel sequence gait detection algorithm model is shown, including the following steps: S202: Train and optimize the model loss of the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user; S204: Perform edge computing optimization on the model to be analyzed and introduce a scoring mechanism so that the qualified multi-channel sequence gait detection algorithm model can perform gait scoring on the target user.

[0025] Furthermore, in a preferred embodiment of the present invention, the training and optimization of the model loss of the multi-channel sequence gait detection algorithm model, and at the same time introducing a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user, specifically: Run the multi-channel sequence gait detection algorithm model, introduce smoothed cross-entropy, and use the smoothed cross-entropy as the loss function of the multi-channel sequence gait detection algorithm model. Through the loss function, perform loss training on the multi-channel sequence gait detection algorithm model, where the loss training is to perform class balance training on the prediction results using smoothed cross-entropy; Introduce the cosine annealing algorithm. During the loss training, calculate the cosine annealing learning rate of the prediction results in real time through the cosine annealing algorithm, and preset the standard cosine annealing learning rate threshold. If the cosine annealing learning rate of the prediction results reaches the standard cosine annealing learning rate threshold, stop the loss training to obtain the prediction results after loss training, which are labeled as a type of predicted user feature data. At the same time, label the multi-channel sequence gait detection algorithm model after loss training as the target multi-channel sequence gait detection algorithm model; Determine the timestamp corresponding to a type of predicted user feature data, label it as a type of timestamp, and collect in real time the real feature data corresponding to the target user at the type of timestamp, which is labeled as a type of real user feature data; Among them, a type of predicted user feature data is the bioelectric current value, acceleration, angular velocity, and pressure data of the lower limb hip and knee of the predicted target user within a specific time, and a type of real user feature data is the real data of the bioelectric current value, acceleration, angular velocity, and pressure data of the lower limb hip and knee of the target user within a specific time; Calculate the Euclidean distance between a type of predicted user feature data and a type of real user feature data, label it as a type of Euclidean distance. If the type of Euclidean distance is maintained within the preset value, divide the target multi-channel sequence gait detection algorithm model into a qualified multi-channel sequence gait detection algorithm model; If the type of Euclidean distance is not maintained within the preset value, divide the target multi-channel sequence gait detection algorithm model into a model to be analyzed, perform edge computing optimization on the model to be analyzed, and introduce a scoring mechanism so that the qualified multi-channel sequence gait detection algorithm model can score the gait of the target user.

[0026] It should be noted that the smoothed cross-entropy is a loss function used to alleviate the class imbalance of all predicted data and achieve the purpose of smooth and real output of the data. The cosine annealing learning rate represents the accuracy of the predicted data. When the cosine annealing learning rate reaches the standard value, it proves that the accuracy of the predicted data reaches the standard value. At this time, the target multi-channel sequence gait detection algorithm model is output. The predicted data obtained from the target multi-channel sequence gait detection algorithm model, that is, the characteristic data predicted by the user at the hip and knee of the lower limbs, needs to be verified for authenticity. Authenticity can be judged by the data similarity between the actual value and the predicted value. The Euclidean distance inference method is an algorithm for judging data similarity. When the Euclidean distance is maintained within the standard value, it proves that the data similarity is high, and it also proves that the authenticity of the data predicted by the target multi-channel sequence gait detection algorithm model is valid and can be used for lower limb hip and knee rehabilitation. If the Euclidean distance is not maintained within the standard value, the model needs to be optimized.

[0027] Further, in a preferred embodiment of the present invention, the edge computing of the model to be analyzed is optimized and a scoring mechanism is introduced, so that the qualified multi-channel sequence gait detection algorithm model can perform gait scoring on the target user, specifically: Perform lightweight processing on the edge computing of the model to be analyzed. Among them, the lightweight processing of the edge computing of the model is to first perform channel pruning on the model to be analyzed, and introduce the singular value decomposition algorithm to perform singular value decomposition on the model to be analyzed after channel pruning, reduce the computational amount and memory occupancy rate of the model to be analyzed, and obtain the target model to be analyzed; Perform robustness verification on the target model to be analyzed. Among them, the robustness verification is to partially mask the characteristic data of the target user generated by the target model to be analyzed, and perform coincidence analysis on the characteristic data of the target user obtained after partial masking and the complete data; If the coincidence degree is greater than the preset value, output the target model to be analyzed and classify it as a qualified multi-channel sequence gait detection algorithm model. If the coincidence degree is not greater than the preset value, continue to perform lightweight processing on the edge computing of the target model to be analyzed until a qualified multi-channel sequence gait detection algorithm model is generated; Introduce a real-time scoring mechanism into the qualified multi-channel sequence gait detection algorithm model and construct a user gait scoring table, and mark the scores corresponding to different user characteristic data in the user gait scoring table; Output the user characteristic data by the qualified multi-channel sequence gait detection algorithm model, label it as qualified user characteristic data, and perform gait state scoring on the qualified user characteristic data based on the user gait scoring table.

[0028] It should be noted that the edge computing optimization of the model, that is, the lightweight processing of the model, can reduce the computational complexity of the model and improve the computational accuracy of the model. Pruning is to remove redundant data in the model, and the singular value decomposition algorithm is an algorithm for reducing computational complexity. The target model to be analyzed obtained after edge computing optimization needs to be verified for robustness. By randomly masking about 30% of the channel data manually and comparing the coincidence degree under the complete data, if it is proved that the robustness is qualified within the preset range, it is a qualified multi-channel sequence gait detection algorithm model. If it is unqualified, continue the lightweight processing until it is qualified. The purpose of introducing a real-time scoring mechanism for gait scoring in the qualified multi-channel sequence gait detection algorithm model is that if the predicted next action of the lower limb hip and knee is a gait scoring qualified action, the robotic arm does not need to intervene to assist in rehabilitation, but the user moves by himself, so the rehabilitation effect is better. If the predicted next action gait score is unqualified, it proves that the user's lower limb hip and knee will have difficulty performing the next action and needs the robotic arm to assist in rehabilitation. Therefore, according to the gait score, it is judged whether the robotic arm intervenes in the rehabilitation work.

[0029] As Figure 3 shown, the second aspect of the present invention also provides a system for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation. The system for effectively collecting and analyzing user characteristic data includes a memory 31 and a processor 32. The memory 31 stores a method for effectively collecting and analyzing user characteristic data. When the method for effectively collecting and analyzing user characteristic data is executed by the processor 32, the following steps are implemented: Deploy sensors at the user's lower limb hip and knee, control the sensors to collect relevant data of the user's lower limb hip and knee, and at the same time perform data preprocessing on the relevant data of the user's lower limb hip and knee to obtain preprocessed bioelectric current values, preprocessed accelerations, preprocessed angular velocities, and preprocessed pressure data; Based on a convolutional neural network, combined with the preprocessed bioelectric current values, preprocessed accelerations, preprocessed angular velocities, and preprocessed pressure data, design a multi-channel sequence gait detection algorithm model; Perform model loss training optimization on the multi-channel sequence gait detection algorithm model, and at the same time introduce a scoring mechanism so that the multi-channel sequence gait detection algorithm model can perform gait scoring on the target user.

[0030] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An effective method for collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation, characterized in that: The following steps are involved: Deploy sensors at the hip and knee of the user's lower limbs, and control the sensors to collect relevant data of the hip and knee of the user's lower limbs, and preprocess the relevant data of the hip and knee of the user's lower limbs to obtain preprocessed biocurrent value, preprocessed acceleration, preprocessed angular velocity and preprocessed pressure data; Based on convolutional neural network, combined with preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity and preprocessed pressure data, a multi-channel sequence gait detection algorithm model is designed; The model loss training of the multi-channel sequence gait detection algorithm model is optimized, and a scoring mechanism is introduced to enable the multi-channel sequence gait detection algorithm model to score the gait of the target user.

2. The method for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation according to claim 1, characterized in that: The sensors are deployed at the hip and knee of the user's lower limbs, and the sensors are controlled to collect relevant data of the user's lower limbs hip and knee, and the relevant data of the user's lower limbs hip and knee are preprocessed to obtain preprocessed biocurrent value, preprocessed acceleration, preprocessed angular velocity and preprocessed pressure data, specifically: The user who needs lower limb hip and knee rehabilitation is calibrated as the target user, and the position where the sensor needs to be deployed on the target user's lower limb hip and knee is determined and calibrated as the target deployment position, wherein the sensor includes a pressure sensor and a surface electromyography sensor, and the sensor deployment position includes the target user's hip joint, knee joint and foot; A pressure sensor and a surface electromyography sensor are respectively installed at the target deployment position, and a standard time for performing a lower limb hip and knee rehabilitation test is preset and calibrated as a target time; Within the target time, based on the pressure sensor and the surface electromyography sensor, the pressure data and the surface electromyography data at the target deployment position of the target user are collected in real time, wherein the pressure data includes the pressure magnitude and the pressure distribution, and the surface electromyography data includes the biocurrent value, angular velocity and acceleration of the muscle at the target deployment position; When collecting pressure data and surface electromyography data at the target deployment position of the target user, different collection times of the pressure data and the surface electromyography data are generated simultaneously, and a collection timestamp is generated based on the different collection times of the pressure data and the surface electromyography data; Based on the acquisition timestamp, the different pressure data and surface electromyography data collected are aligned for time consistency, and the acquisition frequencies of the pressure sensor and the surface electromyography sensor are controlled to be equal. At the same time, the collected pressure data and surface electromyography data are preprocessed and enhanced.

3. The method for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation according to claim 2, characterized in that: The data preprocessing and data enhancement processing of the collected pressure data and surface electromyography data are specifically as follows: The collected pressure data and surface electromyography data are collectively referred to as lower limb hip and knee data, and a Kalman filter algorithm is introduced to perform Kalman filtering on the lower limb hip and knee data to remove high-frequency noise from the lower limb hip and knee data to obtain lower limb hip and knee filtered data; The lower limb hip and knee filter data are subjected to gravity compensation separation to obtain the biocurrent value, acceleration and angular velocity of the surface electromyography data in the lower limb hip and knee filter data, which are calibrated as preprocessed biocurrent value, preprocessed acceleration and preprocessed angular velocity. At the same time, the lower limb hip and knee filter data are subjected to pressure normalization processing to obtain preprocessed pressure data.

4. The method for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation according to claim 1, characterized in that: The multi-channel sequence gait detection algorithm model is designed based on the convolutional neural network, combined with the pre-processed bioelectric current value, pre-processed acceleration, pre-processed angular velocity and pre-processed pressure data, specifically: Introducing a convolutional neural network and constructing a convolutional neural network blank model, wherein the convolutional neural network blank model includes a convolution layer, a pooling layer, and an output layer; The preprocessed bioelectric current value, preprocessed acceleration, preprocessed angular velocity and preprocessed pressure data are collectively referred to as target user feature data, the target user feature data are imported into the convolution layer of the convolutional neural network blank model, and a multi-channel time series matrix is ​​constructed for the target user feature data in the convolution layer of the convolutional neural network blank model; The multi-channel time series matrix of the target user feature data is a matrix for performing spatiotemporal alignment of the target user feature data; In the convolution layer of the blank model of the convolutional neural network, the multi-channel time series matrix is ​​traversed and convolved, and the time series lengths of the multi-channel time series matrix are kept equal during the traversal and convolution process; After the convolution is completed, the multi-channel time series matrix after the convolution is imported into the pooling layer for global time series pooling to match the target user to adapt to different gaits. At the same time, a multi-channel attention mechanism is introduced into the blank model of the convolutional neural network during the global time series pooling process. The multi-channel attention mechanism automatically focuses on the multi-channel time series matrix after convolution during the global time series pooling process to prevent data loss. When the global time series pooling is completed, the trained multi-channel time series matrix is ​​output in the output layer to obtain the multi-channel sequence gait detection algorithm model; Among them, the multi-channel sequence gait detection algorithm model can predict the gait state of the target user within a specified time.

5. The method for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation according to claim 1, characterized in that: The multi-channel sequence gait detection algorithm model is trained and optimized for model loss, and a scoring mechanism is introduced so that the multi-channel sequence gait detection algorithm model can score the gait of the target user, specifically: Running a multi-channel sequence gait detection algorithm model, introducing smoothed cross entropy, and using the smoothed cross entropy as a loss function of the multi-channel sequence gait detection algorithm model, and performing loss training on the multi-channel sequence gait detection algorithm model through the loss function, wherein the loss training is to use the smoothed cross entropy to perform category balance training on the prediction results; The cosine annealing algorithm is introduced to calculate the cosine annealing learning rate of the prediction result in real time during the loss training, and the standard cosine annealing learning rate threshold is preset. If the cosine annealing learning rate of the prediction result reaches the standard cosine annealing learning rate threshold, the loss training is stopped, and the prediction result after the loss training is obtained and calibrated as a class of predicted user feature data. At the same time, the multi-channel sequence gait detection algorithm model after the loss training is calibrated as the target multi-channel sequence gait detection algorithm model; Determine a timestamp corresponding to a type of predicted user feature data, mark it as a type of timestamp, and collect real feature data corresponding to the type of timestamp of the target user in real time, mark it as a type of real user feature data; Among them, one type of predicted user characteristic data is the predicted bioelectric current value, acceleration, angular velocity and pressure data of the lower limb hip and knee of the target user within a specific time, and one type of real user characteristic data is the real data of the bioelectric current value, acceleration, angular velocity and pressure data of the lower limb hip and knee of the target user within a specific time; Calculate the Euclidean distance between a class of predicted user feature data and a class of real user feature data, calibrate it as a class of Euclidean distance, and if the class of Euclidean distance is maintained within a preset value, classify the target multi-channel sequence gait detection algorithm model as a qualified multi-channel sequence gait detection algorithm model; If a type of Euclidean distance is not maintained at the preset value, the target multi-channel sequence gait detection algorithm model is divided into models to be analyzed, edge computing optimization is performed on the models to be analyzed, and a scoring mechanism is introduced so that the qualified multi-channel sequence gait detection algorithm model can score the gait of the target user.

6. The method for effectively collecting and analyzing user characteristic data for lower limb hip and knee rehabilitation according to claim 5, characterized in that: The edge computing optimization of the model to be analyzed and the introduction of a scoring mechanism enable the qualified multi-channel sequence gait detection algorithm model to score the gait of the target user, specifically: The model to be analyzed is subjected to model edge computing lightweight processing, wherein the model edge computing lightweight processing is to first perform channel pruning processing on the model to be analyzed, and introduce a singular value decomposition algorithm to perform singular value decomposition on the model to be analyzed after the channel pruning processing, so as to reduce the computational complexity and memory occupancy rate of the model to be analyzed, and obtain the target model to be analyzed; Performing robustness verification on the target model to be analyzed, wherein the robustness verification is to partially mask the feature data of the target user generated by the target model to be analyzed, and performing overlap analysis on the feature data of the target user obtained after the partial masking and the complete data; If the overlap is greater than the preset value, the target model to be analyzed is output and classified as a qualified multi-channel sequence gait detection algorithm model. If the overlap is not greater than the preset value, the target model to be analyzed continues to be processed by edge computing lightweight processing until a qualified multi-channel sequence gait detection algorithm model is generated. A real-time scoring mechanism is introduced into the qualified multi-channel sequence gait detection algorithm model, and a user gait scoring table is constructed, wherein the scores corresponding to different user feature data are indicated in the user gait scoring table; The user characteristic data output by the qualified multi-channel sequence gait detection algorithm model is calibrated as qualified user characteristic data, and the gait state of the qualified user characteristic data is scored based on the user gait scoring table.

7. An effective user characteristic data collection and analysis system for lower limb hip and knee rehabilitation, characterized in that: The user characteristic data effective collection and analysis system includes a memory and a processor, and the memory stores a user characteristic data effective collection and analysis method program. When the user characteristic data effective collection and analysis method program is executed by the processor, the user characteristic data effective collection and analysis method steps as described in any one of claims 1-6 are implemented.

Images