Information processing method and device for monitoring rehabilitation effect, equipment and medium

Abstract

The application discloses an information processing method and device for monitoring rehabilitation effect, equipment and medium, relates to the technical field of artificial intelligence, and comprises the following steps: determining target cardiopulmonary endurance data of a target user; generating a target cardiopulmonary endurance feature vector based on the target cardiopulmonary endurance data by using a target cardiopulmonary endurance model; the target cardiopulmonary endurance model comprises a preset cardiopulmonary endurance common model, a preset individual long-term cardiopulmonary endurance model and a preset individual short-term cardiopulmonary endurance model; determining a predicted rehabilitation effect according to the target cardiopulmonary endurance feature vector, determining an actual rehabilitation effect according to a target rehabilitation exercise record of the target user; comparing the predicted rehabilitation effect and the actual rehabilitation effect, and adjusting the target cardiopulmonary endurance feature vector and / or a rehabilitation exercise scheme according to a comparison result, so as to redetermine the predicted rehabilitation effect based on the adjusted cardiopulmonary endurance feature vector and / or the adjusted rehabilitation exercise scheme. The application can monitor the rehabilitation exercise effect of inpatients in a precise manner according to the needs of the inpatients.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to information processing methods, devices, equipment and media for monitoring rehabilitation effects. Background Technology

[0002] While current wearable rehabilitation exercise monitoring technologies have made some progress in physiological signal acquisition and motion feature recognition, their adaptability to the specific rehabilitation needs of hospitalized patients remains significantly insufficient. These technologies suffer from deficiencies in multimodal exercise physiological data fusion and processing, and in constructing dynamic evaluation models for rehabilitation effects. Furthermore, they exhibit poor synergy with clinical medical systems and struggle to balance the monitoring stability of wearable devices with patient comfort. Due to the lack of hierarchical personalized model design, existing technologies cannot accurately capture individual differences in cardiopulmonary endurance among hospitalized patients, nor can they effectively extract short-term dynamic changes in physiological states during exercise, resulting in insufficient accuracy in predicting rehabilitation effects. Moreover, existing technologies often only achieve data acquisition and simple analysis, failing to establish a closed loop between model prediction and clinical efficacy verification. This disconnect between monitoring results and rehabilitation plan adjustments prevents the provision of accurate, efficient, and safe rehabilitation exercise recovery effect monitoring services for hospitalized patients, hindering the large-scale implementation and application of intelligent rehabilitation technologies in clinical hospital settings.

[0003] In conclusion, how to conduct precise monitoring of rehabilitation exercise effects tailored to the needs of hospitalized patients is a pressing technical problem that needs to be solved. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide an information processing method, apparatus, device, and medium for monitoring rehabilitation effects, capable of performing precise monitoring of rehabilitation exercise effects tailored to the needs of hospitalized patients. The specific solution is as follows: In a first aspect, this application provides an information processing method for monitoring rehabilitation effects, applied to a computer device, comprising: Receive initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device, and preprocess the initial cardiopulmonary endurance data to obtain target cardiopulmonary endurance data; Using a pre-trained target cardiorespiratory endurance model, a target cardiorespiratory endurance representation vector corresponding to the target user is generated based on the target cardiorespiratory endurance data; the target cardiorespiratory endurance model includes a preset common cardiorespiratory endurance model, a preset individual long-term cardiorespiratory endurance model, and a preset individual short-term cardiorespiratory endurance model; The predicted rehabilitation effect of the target user is determined based on the target cardiorespiratory endurance representation vector, and the actual rehabilitation effect of the target user is determined based on the target user's target rehabilitation exercise record. The predicted rehabilitation effect is compared with the actual rehabilitation effect, and the target cardiorespiratory endurance representation vector and / or the rehabilitation exercise program of the target user are adjusted according to the comparison results, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiorespiratory endurance representation vector and / or the adjusted rehabilitation exercise program.

[0005] Optionally, receiving the initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device, and preprocessing the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data, includes: When the target user is performing rehabilitation exercises, the initial cardiopulmonary endurance data of the target user is collected through the preset wearable device based on a preset collection frequency; the initial cardiopulmonary endurance data is in the form of time series data, and the initial cardiopulmonary endurance data includes the target user's heart rate, speed, location and exercise time; Outliers in the initial cardiopulmonary endurance data are removed to obtain the first cardiopulmonary endurance data, and the missing values ​​in the first cardiopulmonary endurance data are filled using the mean-filling method to obtain the second cardiopulmonary endurance data. The velocity data in the second cardiopulmonary endurance data is smoothed using Kalman filtering to obtain the third cardiopulmonary endurance data, and the third cardiopulmonary endurance data is then discretely standardized to obtain the target cardiopulmonary endurance data.

[0006] Optionally, the information processing method for monitoring rehabilitation effects further includes: Acquire first rehabilitation exercise records from several users and input the first rehabilitation exercise records into a first initial model; the first rehabilitation exercise records include first velocity sequence data and first time sequence data; The first rehabilitation exercise record is processed using the first initial model to output a corresponding first predicted heart rate sequence; wherein, the data flow of the first rehabilitation exercise record in the first initial model is a one-dimensional convolutional neural network, a preset multi-head attention mechanism, residual connections and layer normalization, a fully connected feedforward network, the residual connections and the layer normalization; The first actual heart rate sequence corresponding to the first rehabilitation exercise record is determined, and the first initial model is optimized based on the loss value between the first actual heart rate sequence and the first predicted heart rate sequence to obtain the preset cardiopulmonary endurance common model.

[0007] Optionally, the information processing method for monitoring rehabilitation effects further includes: The second rehabilitation exercise record of the target user is obtained, and the second rehabilitation exercise record is input into the second initial model and the preset cardiopulmonary endurance common model, respectively; the second rehabilitation exercise record includes second velocity sequence data and second time series data; A second predicted heart rate sequence is generated based on the second rehabilitation exercise record using the second initial model, and a third predicted heart rate sequence is generated based on the second rehabilitation exercise record using the preset cardiopulmonary endurance commonality model. The second and third predicted heart rate sequences are added together and averaged to obtain the fourth predicted heart rate sequence. The second actual heart rate sequence corresponding to the second rehabilitation exercise record is determined, and the second initial model is optimized based on the loss value between the second actual heart rate sequence and the fourth predicted heart rate sequence to obtain the preset individual long-term cardiopulmonary endurance model.

[0008] Optionally, the information processing method for monitoring rehabilitation effects further includes: Using the preset individual long-term cardiopulmonary endurance model, a corresponding fifth predicted heart rate sequence is generated based on the second rehabilitation exercise record; The second rehabilitation exercise record and the fifth predicted heart rate sequence are divided into corresponding slice data, and the second actual heart rate sequence is divided into corresponding actual heart rate sequence slices. The sliced ​​data is input into the third initial model, and the corresponding sixth predicted heart rate sequence is generated using the third initial model; The third initial model is optimized based on the loss value between the actual heart rate sequence slice and the sixth predicted heart rate sequence to obtain the preset individual short-term cardiopulmonary endurance model.

[0009] Optionally, generating the target cardiorespiratory endurance representation vector corresponding to the target user based on the target cardiorespiratory endurance data includes: A first vector is generated based on the target cardiopulmonary endurance data using the preset common cardiopulmonary endurance model, a second vector is generated based on the target cardiopulmonary endurance data using the preset individual long-term cardiopulmonary endurance model, and a third vector is generated based on the target cardiopulmonary endurance data using the preset individual short-term cardiopulmonary endurance model. Based on the first weight corresponding to the first vector, the second weight corresponding to the second vector, and the third weight corresponding to the third vector, the first vector, the second vector, and the third vector are weighted and fused to generate the target cardiopulmonary endurance characterization vector of the target user. Accordingly, adjusting the target cardiopulmonary endurance characterization vector based on the comparison results includes: If the comparison result indicates that the predicted rehabilitation effect and the actual rehabilitation effect are different, then the first weight, the second weight and the third weight are adjusted according to the comparison result, and the first vector, the second vector and the third vector are weighted and fused based on the adjusted first weight, the adjusted second weight and the adjusted third weight to obtain the adjusted cardiopulmonary endurance characterization vector.

[0010] Optionally, determining the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance representation vector includes: Obtain cardiorespiratory endurance representation vectors from several users, and cluster the cardiorespiratory endurance representation vectors based on preset rehabilitation effect labels; the preset rehabilitation effect labels include improving, maintaining, and deteriorating. Based on the clustering results, the target rehabilitation effect label corresponding to the target cardiopulmonary endurance representation vector is determined, and the predicted rehabilitation effect of the target user is determined based on the target rehabilitation effect label.

[0011] Secondly, this application provides an information processing device for monitoring rehabilitation effects, applied to a computer device, comprising: The data acquisition module is used to receive the initial cardiopulmonary endurance data of the target user transmitted by the preset wearable device, and to preprocess the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data. The vector generation module is used to generate a target cardiopulmonary endurance representation vector corresponding to the target user based on the target cardiopulmonary endurance data using a pre-trained target cardiopulmonary endurance model; the target cardiopulmonary endurance model includes a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model; The rehabilitation effect determination module is used to determine the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance characterization vector, and to determine the actual rehabilitation effect of the target user based on the target user's target rehabilitation exercise record; The rehabilitation effect comparison module is used to compare the predicted rehabilitation effect with the actual rehabilitation effect, and adjust the target cardiopulmonary endurance representation vector and / or the rehabilitation exercise program of the target user according to the comparison results, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise program.

[0012] Thirdly, this application provides an electronic device, comprising: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned information processing method for monitoring rehabilitation effects.

[0013] Fourthly, this application provides a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned information processing method for monitoring rehabilitation effects.

[0014] In this application, the initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device is first received and preprocessed to obtain target cardiopulmonary endurance data. Then, a target cardiopulmonary endurance representation vector corresponding to the target user is generated based on the target cardiopulmonary endurance data using a pre-trained target cardiopulmonary endurance model. The target cardiopulmonary endurance model includes a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model. Next, the predicted rehabilitation effect of the target user is determined based on the target cardiopulmonary endurance representation vector, and the actual rehabilitation effect of the target user is determined based on the target user's target rehabilitation exercise record. Finally, the predicted rehabilitation effect and the actual rehabilitation effect are compared, and the target cardiopulmonary endurance representation vector and / or the target user's rehabilitation exercise plan are adjusted based on the comparison results, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise plan. As can be seen from the above, this application first receives the initial cardiopulmonary endurance data of the target user collected by a preset wearable device, obtains the target cardiopulmonary endurance data after preprocessing, and then uses a pre-trained target cardiopulmonary endurance model composed of a preset cardiopulmonary endurance common model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model to generate the target cardiopulmonary endurance representation vector of the target user based on the target cardiopulmonary endurance data. Subsequently, the predicted rehabilitation effect of the target user is determined based on the target cardiopulmonary endurance representation vector, and the actual rehabilitation effect is determined by combining the target user's target rehabilitation exercise record. Finally, the predicted rehabilitation effect and the actual rehabilitation effect are compared, and the target cardiopulmonary endurance representation vector and / or the target user's rehabilitation exercise plan are adjusted in a targeted manner based on the comparison results. The predicted rehabilitation effect of the target user is then re-determined based on the adjusted data. In this way, this application can realize closed-loop management of the entire process from cardiopulmonary endurance data collection and processing to rehabilitation effect prediction, verification and optimization. By using the three sub-models of the target cardiopulmonary endurance model to accurately mine the user's cardiopulmonary endurance feature vector, and combining it with actual clinical rehabilitation records to verify rehabilitation effects and adjust rehabilitation exercise programs, it can effectively improve the accuracy of rehabilitation effect prediction and make rehabilitation exercise programs more in line with the user's rehabilitation needs. At the same time, it can realize deep collaboration between wearable device data collection and clinical rehabilitation assessment, promote the intelligent and personalized development of rehabilitation exercise monitoring, and provide scientific and efficient technical support for monitoring the rehabilitation exercise effects of hospitalized patients. Attached Figure Description

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

[0016] Figure 1 This application provides a flowchart of an information processing method for monitoring rehabilitation effects; Figure 2 A flowchart of an algorithm for a specific target cardiopulmonary endurance model provided in this application; Figure 3 A schematic diagram of a specific common model structure for cardiopulmonary endurance provided in this application; Figure 4 A schematic diagram of a specific multi-head attention mechanism structure is provided for this application; Figure 5 A schematic diagram of a specific Transformer encoder structure is provided in this application; Figure 6 A schematic diagram of a specific computational process for a pre-defined multi-head attention mechanism based on forward dependency provided in this application; Figure 7 A schematic diagram of a specific individual long-term cardiopulmonary endurance model provided in this application; Figure 8 A schematic diagram of a specific individual short-term cardiopulmonary endurance model provided in this application; Figure 9 A schematic diagram of a specific motion physiological representation model structure based on multimodal learning provided in this application; Figure 10 This application provides a flowchart of a specific information processing method for monitoring rehabilitation effects; Figure 11 This application provides a schematic diagram of an information processing device for monitoring rehabilitation effects. Figure 12 This application provides a structural diagram of an electronic device. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] While current wearable rehabilitation exercise monitoring technologies have made some progress in physiological signal acquisition and motion feature recognition, their adaptability to the specific rehabilitation needs of hospitalized patients remains significantly insufficient. These technologies suffer from deficiencies in multimodal exercise physiological data fusion and processing, and the construction of dynamic evaluation models for rehabilitation effects. Furthermore, they exhibit poor synergy with clinical medical systems and struggle to balance the monitoring stability of wearable devices with patient comfort. Due to the lack of hierarchical personalized model design, existing technologies cannot accurately capture individual cardiopulmonary endurance differences among hospitalized patients, nor can they effectively extract short-term dynamic changes in physiological states during exercise, resulting in insufficient accuracy in predicting rehabilitation effects. Moreover, existing technologies often only achieve data acquisition and simple analysis, failing to establish a closed loop between model prediction and clinical efficacy verification. This disconnect between monitoring results and rehabilitation plan adjustments prevents the provision of accurate, efficient, and safe rehabilitation exercise recovery effect monitoring services for hospitalized patients, hindering the large-scale implementation and application of intelligent rehabilitation technology in clinical hospital settings. Therefore, this application provides an information processing scheme for monitoring rehabilitation effects, enabling precise rehabilitation exercise effect monitoring tailored to the needs of hospitalized patients.

[0019] See Figure 1 As shown, this embodiment of the invention discloses an information processing method for monitoring rehabilitation effects, applied to a computer device, and may include: Step S11: Receive the initial cardiopulmonary endurance data of the target user transmitted by the preset wearable device, and preprocess the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data.

[0020] In this embodiment, it is necessary to first receive the initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device, and preprocess the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data. The specific process may include: when the target user is performing rehabilitation exercises, firstly, the initial cardiopulmonary endurance data of the target user is collected through the preset wearable device based on a preset acquisition frequency; the initial cardiopulmonary endurance data is in the form of time series data, and the initial cardiopulmonary endurance data includes the target user's heart rate, speed, position, and exercise time; then, outliers in the initial cardiopulmonary endurance data are removed to obtain first cardiopulmonary endurance data, and missing values ​​in the first cardiopulmonary endurance data are filled using the mean-filling method to obtain second cardiopulmonary endurance data; finally, the speed data in the second cardiopulmonary endurance data is smoothed using Kalman filtering to obtain third cardiopulmonary endurance data, and the third cardiopulmonary endurance data is discretely standardized to obtain the target cardiopulmonary endurance data.

[0021] Specifically, initial cardiorespiratory endurance data is collected by patients wearing watches and starting a cardiorespiratory endurance exercise test on a mini-program. The mini-program provides voice prompts, suggesting acceleration when the speed is below the specified limit and deceleration when the speed exceeds the limit. The cardiorespiratory endurance exercise test rules are as follows: starting at 2 km / h, the speed increases by 2 km / h every 2 minutes until reaching 12 km / h, with data collected every second. During each exercise session, the heart rate, exercise speed, and exercise time are recorded. Therefore, the resulting initial cardiorespiratory endurance data is multivariate time-series data. The location series consists of timestamps and latitude / longitude, the speed series consists of timestamps and velocity, and the heart rate series consists of timestamps and heart rate values. As can be seen, this embodiment is specifically optimized for hospitalization scenarios: the exercise testing rules are set as a step-by-step scheme that gradually increases from 2km / h to 12km / h, with voice prompts to guide patients to exercise in a standardized manner, which is suitable for the exercise ability of hospitalized patients; the data acquisition frequency is set to 1 second, which ensures data density while avoiding excessive power consumption of the device affecting its use; the wearable device adopts the form of a watch, which takes into account both monitoring accuracy and wearing comfort, solving the problems of inconvenience of wearing and insufficient exercise guidance in hospitalization scenarios of previous technologies, and achieving a balance between monitoring effectiveness and ease of use.

[0022] Next, preprocessing is required before analyzing the initial cardiorespiratory endurance data. The specific process is as follows: (1) Removal of outliers: Due to limitations of the data acquisition equipment, some outliers may exist in speed and heart rate. Therefore, these outliers need to be removed first. For heart rate data uploaded by wearable devices, the maximum fluctuation threshold per second is used to determine whether there are any anomalies. Outliers are deleted directly. Similarly, a maximum fluctuation threshold can be set for speed information to remove speed data that exceeds the threshold range.

[0023] (2) Filling missing data with interpolation: After removing outliers from the exercise physiological data, some data may be missing in a certain dimension. Therefore, it is necessary to fill the missing values ​​with interpolation. In this embodiment, the mean fill method is used to fill the missing values. If there are consecutive missing values, the data segment and the data after it are deleted.

[0024] (3) Smoothing the velocity data using Kalman filtering: Since the velocity data contains a lot of noise, it is necessary to smooth the velocity data using Kalman filtering. Kalman filtering uses the estimate of the optimal state at the previous time step to predict the state at the current time step, and then corrects it with the value at the current time step to determine the estimate of the optimal state at the current time step. The formula is as follows: ; in, Let F represent the state variable at time k, and let F be the action of the state variable at time k. The state matrix on, B is the value of the control variable at time k, and B is the value acting on the controller vector. Input-control matrix on It's process noise. Let C represent the observed variables at time k, and let C be the observation matrix. It measures noise.

[0025] (4) Standardization: Before inputting cardiopulmonary endurance data into the cardiopulmonary endurance model, the input data needs to be standardized to eliminate the influence of dimensional differences. In this embodiment, min-max standardization, also known as discrete standardization, is used to ensure that the processed target cardiopulmonary endurance data is between [0,1]. For feature x, the standardization formula is as follows: ; in, This represents the minimum value among the features x. This represents the maximum value among the features x.

[0026] As can be seen, this embodiment effectively solves problems such as data noise, missing values, and inconsistent dimensions by using a complete data preprocessing process (outlier removal, mean interpolation filling, Kalman filtering smoothing, and maximum-minimum standardization), providing high-quality data support for model input.

[0027] Step S12: Using a pre-trained target cardiorespiratory endurance model, generate a target cardiorespiratory endurance representation vector corresponding to the target user based on the target cardiorespiratory endurance data; the target cardiorespiratory endurance model includes a preset common cardiorespiratory endurance model, a preset individual long-term cardiorespiratory endurance model, and a preset individual short-term cardiorespiratory endurance model.

[0028] In this embodiment, the target cardiopulmonary endurance model can mine the patient's cardiopulmonary endurance representation vector from large-scale cardiopulmonary endurance data. The target cardiopulmonary endurance model mainly consists of three sub-models: a preset common cardiopulmonary endurance model, i.e., a preset general cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model. The algorithm flow of the target cardiopulmonary endurance model is described in [link to algorithm description]. Figure 2 As shown.

[0029] It should be noted that the general cardiorespiratory endurance model refers to learning the cardiorespiratory endurance of different individuals at different exercise intensities from all rehabilitation exercise patients. This type of feature is helpful in providing a reference for the cardiorespiratory endurance of a particular rehabilitation exercise patient when exercise data for that individual is relatively scarce. The individual long-term cardiorespiratory endurance model extracts cardiorespiratory endurance performance at different exercise intensities from all exercise records of the same rehabilitation exercise patient, and can better represent the individual's cardiorespiratory endurance. The individual short-term cardiorespiratory endurance model requires first segmenting the rehabilitation exercise patient's single exercise data into 2-minute units, which can better extract the rehabilitation exercise patient's cardiorespiratory endurance performance at different time periods within the same type of exercise. The individual short-term cardiorespiratory endurance model differs from the general and long-term individual cardiorespiratory endurance models in that it not only incorporates the time and exercise intensity information of the current time period, but also considers the heart rate status information of previous time periods.

[0030] In general, long-term individual cardiorespiratory endurance models can capture the relationship between exercise physiological state information in the current time period and exercise physiological state information in a longer previous time period; short-term individual cardiorespiratory endurance models focus on cardiorespiratory endurance performance in a short period of time.

[0031] It is understood that in this embodiment, it is necessary to first train and obtain a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model.

[0032] In a first specific implementation, to train a preset common cardiorespiratory endurance model, i.e., a preset general cardiorespiratory endurance model, the specific process may include: firstly, acquiring first rehabilitation exercise records from several users and inputting the first rehabilitation exercise records into a first initial model; the first rehabilitation exercise records include first velocity sequence data and first time sequence data; then, processing the first rehabilitation exercise records using the first initial model to output a corresponding first predicted heart rate sequence; wherein, the data flow of the first rehabilitation exercise records in the first initial model is a one-dimensional convolutional neural network, a preset multi-head attention mechanism, residual connections and layer normalization, a fully connected feedforward network, the residual connections and the layer normalization; finally, determining the first actual heart rate sequence corresponding to the first rehabilitation exercise record, and optimizing the first initial model based on the loss value between the first actual heart rate sequence and the first predicted heart rate sequence to obtain the preset common cardiorespiratory endurance model.

[0033] Specifically, in sports medicine, the human body reaches a certain stable heart rate at each level of exercise intensity, and there is a linear relationship between exercise heart rate and exercise load. In neural network model training, models often require a large amount of data to approximate the data distribution; insufficient training data can lead to inadequate model learning. Therefore, before training an individual long-term cardiorespiratory endurance model, a general cardiorespiratory endurance model needs to be trained first as a prerequisite. This reduces the data requirements during model training, thereby accelerating model convergence. For example... Figure 3 As shown. Due to the long sequence of input sample data, a one-dimensional convolutional neural network was set in the first initial model to shorten the sequence length while maintaining the original number of channels. The input of the general cardiopulmonary endurance model is the first rehabilitation exercise record of all rehabilitation exercise patients. The velocity sequence information and time sequence information in each exercise record are a sample, and the corresponding label data is the first actual heart rate sequence. The data flow of the first rehabilitation exercise record in the first initial model is a one-dimensional convolutional neural network, a pre-set multi-head attention mechanism Fd-MHAM based on forward dependency, then residual connections and layer normalization are performed on the data, then it passes through a fully connected feedforward network, and then residual connections and layer normalization are performed again, finally outputting the first predicted heart rate sequence. Then, the first initial model is optimized according to the loss value between the first actual heart rate sequence and the first predicted heart rate sequence, so as to obtain the common cardiopulmonary endurance model.

[0034] It should be noted that the pre-defined multi-head attention model based on forward dependency can be effectively trained in parallel and can also learn more general long-range dependencies. The Fd-MHAM approach is inspired by the Transformer model proposed by Google in 2017, which not only provides efficient parallel processing capabilities but also learns the global correlations of sequences. The encoder of the pre-defined multi-head attention model based on forward dependency employs a multi-head attention mechanism, such as... Figure 4 As shown, this allows the model to have more perspectives to model the dependencies between input signals, and also enables parallel training of the network. The encoder structure is as follows: Figure 5 As shown. It is understandable that the multi-head attention mechanism uses scaled dot product attention to calculate the attention value, as shown in the following formula: ; Where Q represents the query matrix, K represents the key matrix, and V represents the value matrix. The dimension of the key vector. This represents the scaling factor, which can suppress excessively large Q-K inner products. Multi-head attention mechanisms consist of multiple parallel scaled dot product attention mechanisms combined to perform multiple attention calculations, thereby representing key information of multi-dimensional learning sequences in different subspaces.

[0035] In this embodiment, Fd-MHAM is a modified version of the multi-head attention mechanism in Transformer, and the calculation diagram is shown below. Figure 6 As shown. Because exercise physiology time series data are unidirectionally dependent, the query vector Query for each input data can only match the key vector Key preceding the input sample, which is not the case in Transformer. Therefore, Fd-MHAM is expressed as follows: ; Where N represents the number of input signals, This represents the query vector for the nth input signal. This represents the matrix composed of the key vectors of the first to nth input signals. This represents the matrix composed of the value vectors of the first to nth input signals.

[0036] In a second specific implementation, to train a preset individual long-term cardiorespiratory endurance model, the specific process may include: firstly, acquiring the second rehabilitation exercise record of the target user, and inputting the second rehabilitation exercise record into the second initial model and the preset cardiorespiratory endurance common model respectively; the second rehabilitation exercise record includes second velocity sequence data and second time sequence data; then, using the second initial model to generate a second predicted heart rate sequence based on the second rehabilitation exercise record, and using the preset cardiorespiratory endurance common model to generate a third predicted heart rate sequence based on the second rehabilitation exercise record; subsequently, adding the second predicted heart rate sequence and the third predicted heart rate sequence and performing mean averaging to obtain a fourth predicted heart rate sequence; finally, determining the second actual heart rate sequence corresponding to the second rehabilitation exercise record, and optimizing the second initial model based on the loss value between the second actual heart rate sequence and the fourth predicted heart rate sequence to obtain the preset individual long-term cardiorespiratory endurance model.

[0037] Specifically, in cardiorespiratory endurance exercise tests, the exercise duration for each rehabilitation patient is not fixed, ranging from 2 to 12 minutes. Therefore, for most models, the longer the time series data, the more difficult it becomes to balance the relationship between local and global dependencies. Thus, individual long-term cardiorespiratory endurance models and individual short-term cardiorespiratory endurance models can be used to characterize the cardiorespiratory endurance status of complete exercise records and the changes in cardiorespiratory endurance over a short period. Individual long-term cardiorespiratory endurance models involve representation learning from all exercise records of a specific rehabilitation patient, similar to general cardiorespiratory endurance models, the only difference being that its input is the exercise records of the same rehabilitation patient. For an individual long-term cardiorespiratory endurance model, the exercise records of the same rehabilitation exercise patient need to be input into a general cardiorespiratory endurance model. The output third predicted heart rate sequence is used as part of the second initial model. The parameters in the general cardiorespiratory endurance model remain unchanged. Simultaneously, the sample data is input into the second initial model, which outputs a second predicted heart rate sequence. The second and third predicted heart rate sequences are then added together and averaged to obtain a fourth predicted heart rate sequence. This fourth predicted heart rate sequence is calculated and compared with the second actual heart rate sequence of the rehabilitation exercise patient to calculate the loss value and adjust the parameters of the second initial model, thus obtaining the preset individual long-term cardiorespiratory endurance model. See [link to relevant documentation]. Figure 7 As shown, the individual long-term cardiorespiratory endurance model is a parallel operation of the general cardiorespiratory endurance model and the individual long-term cardiorespiratory endurance model. Then, the predicted heart rate sequences output by each model are added together and averaged to output the final predicted heart rate sequence.

[0038] In a third specific implementation, to train a preset individual short-term cardiorespiratory endurance model, the specific process may include: first, using the preset individual long-term cardiorespiratory endurance model, generating a corresponding fifth predicted heart rate sequence based on the second rehabilitation exercise record; then, dividing the second rehabilitation exercise record and the fifth predicted heart rate sequence to obtain corresponding slice data, and dividing the second actual heart rate sequence to obtain corresponding actual heart rate sequence slices; next, inputting the slice data into a third initial model, and using the third initial model to generate a corresponding sixth predicted heart rate sequence; finally, optimizing the third initial model based on the loss value between the actual heart rate sequence slices and the sixth predicted heart rate sequence to obtain the preset individual short-term cardiorespiratory endurance model.

[0039] Specifically, individual short-term cardiorespiratory endurance models aim to characterize the physiological state of rehabilitation exercise patients within each time slice of each exercise session. The physiological state exhibited at different exercise times is inconsistent. Therefore, relying solely on individual long-term cardiorespiratory endurance models cannot adequately represent the short-term physiological state. The task of individual short-term cardiorespiratory endurance models is to represent the physiological representation of exercise within different time slices of a single exercise session, thereby representing the cardiorespiratory endurance representation vector of rehabilitation exercise patients. Its core is a multimodal learning-based exercise physiological representation model, the specific model representation of which is as follows: Figure 8 As shown. The input to the third initial model is a slice of the velocity sequence and time sequence from the same exercise, and the input of the unsliced ​​data to the preset individual long-term cardiorespiratory endurance model. The output is the fifth predicted heart rate sequence. The slice of the fifth predicted heart rate sequence is used as the input to the third initial model to obtain a cardiorespiratory endurance representation vector representing the exercise rehabilitation patient. Then, the cardiorespiratory endurance representation vector is input to the feedforward layer to obtain the sixth predicted heart rate sequence in a short time. The loss value is calculated by comparing the sixth predicted heart rate sequence with the actual heart rate sequence to train the third initial model to obtain the preset individual short-term cardiorespiratory endurance model.

[0040] It should be noted that this embodiment designs a multimodal learning-based exercise physiological representation model. For the cardiorespiratory endurance performance of exercise rehabilitation patients at different exercise intensities within the same exercise session, multimodal learning is used, which can better represent the cardiorespiratory endurance level of exercise rehabilitation patients at different time periods under different exercise intensities within the same exercise session. To better represent the cardiorespiratory endurance of exercise rehabilitation patients, this embodiment employs a multimodal learning-based exercise physiological representation model, the specific form of which is as follows: Figure 9 As shown, the architecture of the exercise physiological representation model based on multimodal learning consists of three parallel one-dimensional convolutional layers, one vector concatenation layer, two stacked unidirectional GRU (Gated Recurrent Unit) layers, and a feedforward network layer. The first slice in the figure is input into the model and generates exercise physiological representation 1 after two GRU layers. Exercise physiological representation 1 then passes through a feedforward layer to obtain the predicted heart rate sequence. Each exercise physiological representation in the figure only represents the exercise physiological state within the current time segment. It can be seen that this embodiment introduces an exercise physiological representation model based on forward-dependent multi-head attention mechanism (Fd-MHAM) and multimodal learning, efficiently fusing multivariate time series data such as speed, time, and heart rate, accurately capturing the long-distance dependence and short-term dynamic changes of exercise physiological signals. Compared with previous single-dimensional or simple fusion models, it significantly improves the completeness and accuracy of the cardiopulmonary endurance representation vector, laying a solid foundation for predicting rehabilitation effects.

[0041] In this embodiment, since the cardiorespiratory endurance test lasts 12 minutes, it can be divided into 1-minute segments. For the general cardiorespiratory endurance model, the large-scale cardiorespiratory endurance dataset is divided into training, validation, and test sets in an 8:1:1 ratio. For individual long-term and short-term cardiorespiratory endurance models, due to their smaller data volume, all data are used as the training set. All exercise data are preprocessed first, and then the preprocessed data is used to train the cardiorespiratory endurance model. After obtaining the trained target cardiorespiratory endurance model, the cardiorespiratory endurance test data of the rehabilitation patients are input into the model to obtain the cardiorespiratory endurance representation vector for each individual.

[0042] It should be noted that in this embodiment, a first vector can be generated based on the target cardiopulmonary endurance data using a preset common cardiopulmonary endurance model, a second vector can be generated based on the target cardiopulmonary endurance data using a preset individual long-term cardiopulmonary endurance model, and a third vector can be generated based on the target cardiopulmonary endurance data using a preset individual short-term cardiopulmonary endurance model. Then, the first vector, the second vector, and the third vector can be weighted and fused according to the first weight corresponding to the first vector, the second weight corresponding to the second vector, and the third weight corresponding to the third vector to generate the target user's target cardiopulmonary endurance representation vector.

[0043] As can be seen, this embodiment innovatively designs a three-tiered cardiopulmonary endurance sub-model architecture: a general model based on mining universal cardiopulmonary endurance patterns from massive patient data, providing a basic reference for individuals with insufficient data; a long-term individual model focusing on a single patient's complete exercise record to capture long-term rehabilitation trends; and a short-term individual model that can use 2-minute time slices, combined with multimodal learning, to capture dynamic changes in physiological state within a short period. The collaborative operation of the three models not only solves the problems of high data dependence and slow convergence of single models, but also achieves deep adaptation to individual differences and disease stages of hospitalized patients, filling the gap in previous technologies for refined individual monitoring.

[0044] Step S13: Determine the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance representation vector, and determine the actual rehabilitation effect of the target user based on the target user's target rehabilitation exercise record.

[0045] In this embodiment, the predicted rehabilitation effect of a target user is determined based on the target cardiopulmonary endurance representation vector. The specific process may include: first, obtaining cardiopulmonary endurance representation vectors of several users, and clustering the cardiopulmonary endurance representation vectors based on preset rehabilitation effect labels; the preset rehabilitation effect labels include improving, maintaining, and deteriorating; then, determining the target rehabilitation effect label corresponding to the target cardiopulmonary endurance representation vector based on the clustering results, and determining the predicted rehabilitation effect of the target user based on the target rehabilitation effect label.

[0046] Specifically, in this embodiment, cardiorespiratory endurance analysis and cluster analysis can be performed on the exercise dataset. To achieve effective clustering, this embodiment uses K-Means to cluster the cardiorespiratory endurance representation vectors of several users, with preset rehabilitation effect labels including improving, maintaining, and deteriorating. After clustering, this embodiment uses FMI (Fowlkes-Mallows index) to evaluate the clustering effect. Based on the current patients' exercise rehabilitation records, the actual exercise rehabilitation effect is extracted, and deteriorating, maintaining, and improving are used as the true cluster labels, dividing rehabilitation exercise patients into three categories. Therefore, for the clustering of cardiorespiratory endurance representation vectors, the K value is set to 3. Assuming the clustering result A1 (cardiorespiratory endurance clustering result) and the actual classification result (effect classification) are A2, the FMI formula is as follows: ; Wherein, TP indicates that the sample pair is a cluster in A1 and also a cluster in A2; FP indicates that the sample pair is a cluster in A2 but not a cluster in A1; FN indicates that the sample pair is a cluster in A1 but not a cluster in A2; TN indicates that the sample pair is not a cluster in A1 and not a cluster in A2.

[0047] In this way, the target rehabilitation effect label corresponding to the target cardiopulmonary endurance representation vector can be determined based on the clustering results, and the predicted rehabilitation effect for the target user can be determined based on the target rehabilitation effect label. It should be noted that in this embodiment, the exercise rehabilitation records of the target user can be extracted from actual inpatient HIS data to obtain the actual rehabilitation effect.

[0048] Step S14: Compare the predicted rehabilitation effect with the actual rehabilitation effect, and adjust the target cardiopulmonary endurance representation vector and / or the rehabilitation exercise program of the target user according to the comparison result, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise program.

[0049] In this embodiment, the rehabilitation effect predicted by cardiorespiratory endurance data can be compared with the actual rehabilitation effect. By comparing the results, the target cardiorespiratory endurance representation vector and / or the rehabilitation exercise plan of the target user can be manually adjusted to iterate and optimize the target cardiorespiratory endurance model repeatedly, so as to realize the automated exercise rehabilitation monitoring capability.

[0050] It should be noted that adjusting the target cardiopulmonary endurance representation vector based on the comparison results can specifically include the following process: if the predicted rehabilitation effect differs from the actual rehabilitation effect, the first, second, and third weights are adjusted according to the comparison results. Then, the first, second, and third vectors are weighted and fused based on the adjusted first, second, and third weights to obtain the adjusted cardiopulmonary endurance representation vector. As can be seen, this embodiment uses K-Means clustering analysis to classify rehabilitation effects into three categories: worsening, maintaining, and improving. The FMI index is used to accurately evaluate the clustering effect, and bidirectional verification is performed using actual rehabilitation records from the inpatient HIS system, constructing a closed-loop mechanism of "model prediction - clinical validation - weight adjustment / program optimization." This dynamic adjustment mode breaks through the limitations of previous technologies' "disconnect between monitoring and intervention," enabling timely matching of differences between manual evaluation and model prediction results. By iteratively optimizing representation weights or rehabilitation exercise programs, the scientific rigor and effectiveness of rehabilitation training for inpatients are significantly improved, promoting deep collaboration between intelligent rehabilitation technology and the clinical medical system.

[0051] In one specific implementation, see Figure 10 As shown, firstly, rehabilitation exercise data of patients is collected through wearable devices, and in-hospital rehabilitation exercise records are collected simultaneously. The rehabilitation exercise data collected by the wearable devices is input into a cardiopulmonary endurance model, which outputs a corresponding representation vector. Simultaneously, clinical characteristics such as gender, age, and diagnosis are extracted from the in-hospital rehabilitation exercise records to determine the actual rehabilitation exercise effect of the patients. Cluster analysis is performed based on the cardiopulmonary endurance representation vector to predict the patients' rehabilitation effect. The predicted rehabilitation effect is compared with the actual rehabilitation exercise effect to determine if there is a difference. If a difference exists, medical staff adjust the weights of each feature of the cardiopulmonary endurance representation vector and / or the patients' rehabilitation exercise plan to re-predict the rehabilitation effect based on the adjusted content.

[0052] As can be seen from the above, in this embodiment, the initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device is first received, and the initial cardiopulmonary endurance data is preprocessed to obtain target cardiopulmonary endurance data. Then, using a pre-trained target cardiopulmonary endurance model, a target cardiopulmonary endurance representation vector corresponding to the target user is generated based on the target cardiopulmonary endurance data. The target cardiopulmonary endurance model includes a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model. Next, the predicted rehabilitation effect of the target user is determined according to the target cardiopulmonary endurance representation vector, and the actual rehabilitation effect of the target user is determined according to the target user's target rehabilitation exercise record. Finally, the predicted rehabilitation effect and the actual rehabilitation effect are compared, and the target cardiopulmonary endurance representation vector and / or the target user's rehabilitation exercise plan are adjusted according to the comparison result, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise plan. As can be seen from the above, in this embodiment, the initial cardiopulmonary endurance data of the target user collected by the preset wearable device is first received, and the target cardiopulmonary endurance data is obtained after preprocessing. Then, the target cardiopulmonary endurance model, which is composed of the preset cardiopulmonary endurance common model, the preset individual long-term cardiopulmonary endurance model, and the preset individual short-term cardiopulmonary endurance model, is used to generate the target cardiopulmonary endurance representation vector of the target user based on the target cardiopulmonary endurance data. Subsequently, the predicted rehabilitation effect of the target user is determined according to the target cardiopulmonary endurance representation vector. At the same time, the actual rehabilitation effect is determined by combining the target user's target rehabilitation exercise record. Finally, the predicted rehabilitation effect and the actual rehabilitation effect are compared. Based on the comparison results, the target cardiopulmonary endurance representation vector and / or the target user's rehabilitation exercise plan are adjusted in a targeted manner, and the predicted rehabilitation effect of the target user is re-determined based on the adjusted data. In this way, this embodiment can realize closed-loop management of the entire process from cardiopulmonary endurance data collection and processing to rehabilitation effect prediction, verification and optimization. By using the three sub-models of the target cardiopulmonary endurance model to accurately mine the user's cardiopulmonary endurance feature vector, and combining it with actual clinical rehabilitation records to verify rehabilitation effects and adjust rehabilitation exercise programs, the accuracy of rehabilitation effect prediction is effectively improved, making rehabilitation exercise programs more in line with the user's rehabilitation needs. At the same time, it realizes deep collaboration between wearable device data collection and clinical rehabilitation assessment, promotes the intelligent and personalized development of rehabilitation exercise monitoring, and provides scientific and efficient technical support for monitoring the rehabilitation exercise effects of hospitalized patients.

[0053] Accordingly, see Figure 11 As shown in the illustration, this application also provides an information processing device for monitoring rehabilitation effects, applied to a computer device, and may include: The data acquisition module 11 is used to receive the initial cardiopulmonary endurance data of the target user transmitted by the preset wearable device, and to preprocess the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data. The vector generation module 12 is used to generate a target cardiopulmonary endurance representation vector corresponding to the target user based on the target cardiopulmonary endurance data using a pre-trained target cardiopulmonary endurance model; the target cardiopulmonary endurance model includes a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model; The rehabilitation effect determination module 13 is used to determine the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance characterization vector, and to determine the actual rehabilitation effect of the target user based on the target user's target rehabilitation exercise record; The rehabilitation effect comparison module 14 is used to compare the predicted rehabilitation effect with the actual rehabilitation effect, and adjust the target cardiopulmonary endurance representation vector and / or the rehabilitation exercise program of the target user according to the comparison result, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise program.

[0054] In some specific embodiments, the data acquisition module 11 may include: The data acquisition unit is used to acquire the initial cardiopulmonary endurance data of the target user through the preset wearable device based on a preset acquisition frequency when the target user is performing rehabilitation exercises; the initial cardiopulmonary endurance data is in the form of time series data, and the initial cardiopulmonary endurance data includes the target user's heart rate, speed, location and exercise time; The missing value filling unit is used to remove outliers from the initial cardiopulmonary endurance data to obtain the first cardiopulmonary endurance data, and to fill the missing values ​​in the first cardiopulmonary endurance data using the mean filling method to obtain the second cardiopulmonary endurance data. The smoothing unit is used to smooth the velocity data in the second cardiopulmonary endurance data using Kalman filtering to obtain the third cardiopulmonary endurance data, and to perform discrete normalization processing on the third cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data.

[0055] In some specific embodiments, the information processing device for monitoring rehabilitation effects may further include: The first rehabilitation exercise record acquisition unit is used to acquire the first rehabilitation exercise records of several users and input the first rehabilitation exercise records into the first initial model; the first rehabilitation exercise record includes first velocity sequence data and first time sequence data; The first predicted heart rate sequence output unit is used to process the first rehabilitation exercise record using the first initial model to output a corresponding first predicted heart rate sequence; wherein, the data flow of the first rehabilitation exercise record in the first initial model is a one-dimensional convolutional neural network, a preset multi-head attention mechanism, residual connections and layer normalization, a fully connected feedforward network, the residual connections and the layer normalization; The first initial model optimization unit is used to determine the first actual heart rate sequence corresponding to the first rehabilitation exercise record, and optimize the first initial model based on the loss value between the first actual heart rate sequence and the first predicted heart rate sequence to obtain the preset cardiopulmonary endurance common model.

[0056] In some specific embodiments, the information processing device for monitoring rehabilitation effects may further include: The second rehabilitation exercise record acquisition unit is used to acquire the second rehabilitation exercise record of the target user and input the second rehabilitation exercise record into the second initial model and the preset cardiopulmonary endurance common model respectively; the second rehabilitation exercise record includes second velocity sequence data and second time sequence data; The third predicted heart rate sequence generation unit is used to generate a second predicted heart rate sequence based on the second rehabilitation exercise record using the second initial model, and to generate a third predicted heart rate sequence based on the second rehabilitation exercise record using the preset cardiopulmonary endurance commonality model. The fourth predicted heart rate sequence determination unit is used to add the second predicted heart rate sequence and the third predicted heart rate sequence and perform mean averaging to obtain the fourth predicted heart rate sequence. The second initial model optimization unit is used to determine the second actual heart rate sequence corresponding to the second rehabilitation exercise record, and optimize the second initial model based on the loss value between the second actual heart rate sequence and the fourth predicted heart rate sequence to obtain the preset individual long-term cardiopulmonary endurance model.

[0057] In some specific embodiments, the information processing device for monitoring rehabilitation effects may further include: The fifth predicted heart rate sequence generation unit is used to generate a corresponding fifth predicted heart rate sequence based on the second rehabilitation exercise record using the preset individual long-term cardiopulmonary endurance model. The slice data determination unit is used to divide the second rehabilitation exercise record and the fifth predicted heart rate sequence to obtain corresponding slice data, and to divide the second actual heart rate sequence to obtain corresponding actual heart rate sequence slices. The sixth predicted heart rate sequence generation unit is used to input the slice data into the third initial model and use the third initial model to generate the corresponding sixth predicted heart rate sequence. The third initial model optimization unit is used to optimize the third initial model based on the loss value between the actual heart rate sequence slice and the sixth predicted heart rate sequence to obtain the preset individual short-term cardiopulmonary endurance model.

[0058] In some specific embodiments, the vector generation module 12 may include: The vector generation unit is used to generate a first vector based on the target cardiopulmonary endurance data using the preset cardiopulmonary endurance commonality model, generate a second vector based on the target cardiopulmonary endurance data using the preset individual long-term cardiopulmonary endurance model, and generate a third vector based on the target cardiopulmonary endurance data using the preset individual short-term cardiopulmonary endurance model. The vector fusion unit is used to perform weighted fusion of the first vector, the second vector, and the third vector according to the first weight corresponding to the first vector, the second weight corresponding to the second vector, and the third weight corresponding to the third vector, so as to generate the target cardiopulmonary endurance characterization vector of the target user. Accordingly, the rehabilitation effect comparison module 14 may include: The weight adjustment unit is used to adjust the first weight, the second weight, and the third weight according to the comparison result if the comparison result represents a difference between the predicted rehabilitation effect and the actual rehabilitation effect. The unit then performs weighted fusion on the first vector, the second vector, and the third vector based on the adjusted first weight, the adjusted second weight, and the adjusted third weight to obtain the adjusted cardiopulmonary endurance characterization vector.

[0059] In some specific embodiments, the rehabilitation effect determination module 13 may include: A vector clustering unit is used to obtain cardiopulmonary endurance representation vectors of several users and cluster the cardiopulmonary endurance representation vectors based on preset rehabilitation effect labels; the preset rehabilitation effect labels include improving, maintaining, and deteriorating. The rehabilitation effect determination unit is used to determine the target rehabilitation effect label corresponding to the target cardiopulmonary endurance characterization vector based on the clustering results, and to determine the predicted rehabilitation effect of the target user based on the target rehabilitation effect label.

[0060] Furthermore, embodiments of this application also disclose an electronic device, Figure 12This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the information processing method for monitoring rehabilitation effects disclosed in any of the foregoing embodiments. Furthermore, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0061] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0062] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0063] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the information processing method for monitoring rehabilitation effects executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0064] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned information processing method for monitoring rehabilitation effects. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0065] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0066] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0067] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0068] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0069] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. An information processing method for monitoring rehabilitation effects, characterized in that, Applied to computer devices, including: Receive initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device, and preprocess the initial cardiopulmonary endurance data to obtain target cardiopulmonary endurance data; Using a pre-trained target cardiorespiratory endurance model, a target cardiorespiratory endurance representation vector corresponding to the target user is generated based on the target cardiorespiratory endurance data; the target cardiorespiratory endurance model includes a preset common cardiorespiratory endurance model, a preset individual long-term cardiorespiratory endurance model, and a preset individual short-term cardiorespiratory endurance model; The predicted rehabilitation effect of the target user is determined based on the target cardiorespiratory endurance representation vector, and the actual rehabilitation effect of the target user is determined based on the target user's target rehabilitation exercise record. The predicted rehabilitation effect is compared with the actual rehabilitation effect, and the target cardiorespiratory endurance representation vector and / or the rehabilitation exercise program of the target user are adjusted according to the comparison results, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiorespiratory endurance representation vector and / or the adjusted rehabilitation exercise program.

2. The information processing method for monitoring rehabilitation effects according to claim 1, characterized in that, The process of receiving initial cardiopulmonary endurance data of the target user transmitted by a preset wearable device and preprocessing the initial cardiopulmonary endurance data to obtain target cardiopulmonary endurance data includes: When the target user is performing rehabilitation exercises, the initial cardiopulmonary endurance data of the target user is collected through the preset wearable device based on a preset collection frequency; the initial cardiopulmonary endurance data is in the form of time series data, and the initial cardiopulmonary endurance data includes the target user's heart rate, speed, location and exercise time; Outliers in the initial cardiopulmonary endurance data are removed to obtain the first cardiopulmonary endurance data, and the missing values ​​in the first cardiopulmonary endurance data are filled using the mean-filling method to obtain the second cardiopulmonary endurance data. The velocity data in the second cardiopulmonary endurance data is smoothed using Kalman filtering to obtain the third cardiopulmonary endurance data, and the third cardiopulmonary endurance data is then discretely standardized to obtain the target cardiopulmonary endurance data.

3. The information processing method for monitoring rehabilitation effects according to claim 1, characterized in that, Also includes: Obtain the first rehabilitation exercise records of several users and input the first rehabilitation exercise records into the first initial model; The first rehabilitation exercise record includes first velocity sequence data and first time sequence data; The first rehabilitation exercise record is processed using the first initial model to output a corresponding first predicted heart rate sequence; wherein, the data flow of the first rehabilitation exercise record in the first initial model is a one-dimensional convolutional neural network, a preset multi-head attention mechanism, residual connections and layer normalization, a fully connected feedforward network, the residual connections and the layer normalization; The first actual heart rate sequence corresponding to the first rehabilitation exercise record is determined, and the first initial model is optimized based on the loss value between the first actual heart rate sequence and the first predicted heart rate sequence to obtain the preset cardiopulmonary endurance common model.

4. The information processing method for monitoring rehabilitation effects according to claim 3, characterized in that, Also includes: The second rehabilitation exercise record of the target user is obtained, and the second rehabilitation exercise record is input into the second initial model and the preset cardiopulmonary endurance common model, respectively; the second rehabilitation exercise record includes second velocity sequence data and second time series data; A second predicted heart rate sequence is generated based on the second rehabilitation exercise record using the second initial model, and a third predicted heart rate sequence is generated based on the second rehabilitation exercise record using the preset cardiopulmonary endurance commonality model. The second and third predicted heart rate sequences are added together and averaged to obtain the fourth predicted heart rate sequence. The second actual heart rate sequence corresponding to the second rehabilitation exercise record is determined, and the second initial model is optimized based on the loss value between the second actual heart rate sequence and the fourth predicted heart rate sequence to obtain the preset individual long-term cardiopulmonary endurance model.

5. The information processing method for monitoring rehabilitation effects according to claim 4, characterized in that, Also includes: Using the preset individual long-term cardiopulmonary endurance model, a corresponding fifth predicted heart rate sequence is generated based on the second rehabilitation exercise record; The second rehabilitation exercise record and the fifth predicted heart rate sequence are divided into corresponding slice data, and the second actual heart rate sequence is divided into corresponding actual heart rate sequence slices. The sliced ​​data is input into the third initial model, and the corresponding sixth predicted heart rate sequence is generated using the third initial model; The third initial model is optimized based on the loss value between the actual heart rate sequence slice and the sixth predicted heart rate sequence to obtain the preset individual short-term cardiopulmonary endurance model.

6. The information processing method for monitoring rehabilitation effects according to claim 1, characterized in that, The step of generating the target cardiopulmonary endurance representation vector corresponding to the target user based on the target cardiopulmonary endurance data includes: A first vector is generated based on the target cardiopulmonary endurance data using the preset common cardiopulmonary endurance model, a second vector is generated based on the target cardiopulmonary endurance data using the preset individual long-term cardiopulmonary endurance model, and a third vector is generated based on the target cardiopulmonary endurance data using the preset individual short-term cardiopulmonary endurance model. Based on the first weight corresponding to the first vector, the second weight corresponding to the second vector, and the third weight corresponding to the third vector, the first vector, the second vector, and the third vector are weighted and fused to generate the target cardiopulmonary endurance characterization vector of the target user. Accordingly, adjusting the target cardiopulmonary endurance characterization vector based on the comparison results includes: If the comparison result indicates that the predicted rehabilitation effect and the actual rehabilitation effect are different, then the first weight, the second weight and the third weight are adjusted according to the comparison result, and the first vector, the second vector and the third vector are weighted and fused based on the adjusted first weight, the adjusted second weight and the adjusted third weight to obtain the adjusted cardiopulmonary endurance characterization vector.

7. The information processing method for monitoring rehabilitation effects according to any one of claims 1 to 6, characterized in that, The step of determining the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance representation vector includes: Obtain cardiorespiratory endurance representation vectors from several users, and cluster the cardiorespiratory endurance representation vectors based on preset rehabilitation effect labels; the preset rehabilitation effect labels include improving, maintaining, and deteriorating. Based on the clustering results, the target rehabilitation effect label corresponding to the target cardiopulmonary endurance representation vector is determined, and the predicted rehabilitation effect of the target user is determined based on the target rehabilitation effect label.

8. An information processing device for monitoring rehabilitation effects, characterized in that, Applied to computer devices, including: The data acquisition module is used to receive the initial cardiopulmonary endurance data of the target user transmitted by the preset wearable device, and to preprocess the initial cardiopulmonary endurance data to obtain the target cardiopulmonary endurance data. The vector generation module is used to generate a target cardiopulmonary endurance representation vector corresponding to the target user based on the target cardiopulmonary endurance data using a pre-trained target cardiopulmonary endurance model; the target cardiopulmonary endurance model includes a preset common cardiopulmonary endurance model, a preset individual long-term cardiopulmonary endurance model, and a preset individual short-term cardiopulmonary endurance model; The rehabilitation effect determination module is used to determine the predicted rehabilitation effect of the target user based on the target cardiopulmonary endurance characterization vector, and to determine the actual rehabilitation effect of the target user based on the target user's target rehabilitation exercise record; The rehabilitation effect comparison module is used to compare the predicted rehabilitation effect with the actual rehabilitation effect, and adjust the target cardiopulmonary endurance representation vector and / or the rehabilitation exercise program of the target user according to the comparison results, so as to redetermine the predicted rehabilitation effect of the target user based on the adjusted cardiopulmonary endurance representation vector and / or the adjusted rehabilitation exercise program.

9. An electronic device, characterized in that, The electronic device includes a processor and a memory; wherein the memory is used to store a computer program, which is loaded and executed by the processor to implement the information processing method for monitoring rehabilitation effects as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the information processing method for monitoring rehabilitation effects as described in any one of claims 1 to 7.

Images