User motion data analysis method, device and system and computing equipment
The user's movement data was analyzed through the double-layer nested LSTM model and self-attention mechanism, and the problem of inaccurate evaluation based on manual experience was solved, high-precision body function assessment and detailed movement data analysis were achieved, and detailed movement reports were generated.
Patent Information
- Application Number
- CN202510413326.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-03
- Publication Date
- 2025-08-15
AI Technical Summary
In the prior art, the user's physical function is inaccurate and accurate based on manual experience, resulting in safety hazards during the user's exercise.
The user's movement data was analyzed using a double-layer nested LSTM model. Combined with the long-term dependence capture ability and self-attention mechanism of LSTM, the blood pressure, heart rate, respiratory rate, blood oxygen saturation and position data of the user during the walking test, and a high-precision body function evaluation result was generated.
It realizes effective and accurate analysis of user exercise data without manual intervention, provides high-precision body function assessment, and generates detailed exercise data analysis reports, including rest duration and exercise prescriptions.
Smart Images
Figure CN120496730A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of medical information technology, and in particular to a method, device, system and computing equipment for analyzing user motion data. Background Art
[0002] With the improvement of economic level and the need to restore physical health after the epidemic, more and more users are participating in fitness and sports. However, for ordinary users, if they participate in strenuous exercise or exercise that does not match their body functions without understanding their own body functions, it is easy to cause serious consequences. Therefore, the collection and analysis of users' sports-related functional data is particularly important.
[0003] At present, conventional user motion data generally includes heart rate, blood oxygen saturation, respiratory rate and exercise load, etc., which can be monitored by corresponding sensors. Some smartphones and wearable devices have also added relevant sensors for monitoring. After being summarized, they are analyzed and identified based on the experience accumulated by doctors or professionals to determine the user's physical condition. However, evaluating the user's physical function based on manual experience is not accurate enough and the precision is not high. Summary of the Invention
[0004] The present invention provides a method, device, system and computing equipment for analyzing user motion data, which are used to solve the problem that the existing evaluation of user physical functions based on manual experience is not accurate and precise enough.
[0005] In a first aspect, the present invention provides a method for analyzing user motion data, comprising:
[0006] Preprocessing user motion data, including blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data of the user during a walking test, wherein the walking test is conducted according to a test route and test duration, and the user motion data is collected at preset time intervals;
[0007] The preprocessed user motion data is input into the motion data analysis model to obtain the physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
[0008] The above solution is based on a two-layer nested LSTM model to effectively and accurately analyze user motion data without human intervention, obtaining highly accurate physical function assessment results.
[0009] Optionally, the preprocessing includes filtering, completion, denoising and first normalization.
[0010] Optionally, also include:
[0011] Based on the preprocessed heart rate data, respiratory rate data, blood oxygen saturation data, and position data, plot a trend change graph of heart rate, respiratory rate, blood oxygen saturation, and acceleration with time as the horizontal axis and heart rate, respiratory rate, blood oxygen saturation, and acceleration as the vertical axis, respectively. The acceleration is calculated based on the preprocessed position data;
[0012] Splitting the preprocessed heart rate data, respiratory rate data, blood oxygen saturation data, and position data into the specific number of heart rate sets, respiratory rate sets, blood oxygen saturation sets, and position sets according to the unit time length, respectively, calculating the heart rate, respiratory rate, and blood oxygen saturation in each unit time length based on the heart rate sets, respiratory rate sets, and blood oxygen saturation sets in each unit time length, and calculating the walking speed and walking distance in each unit time length based on the position sets in each unit time length;
[0013] A user exercise data analysis report is generated based on the physical condition analysis results, the trend change graphs of the heart rate, respiratory rate, blood oxygen saturation and acceleration, the heart rate, respiratory rate and blood oxygen saturation in each unit time, the walking speed and walking distance in each unit time and the exercise prescription, where the exercise prescription is obtained based on the physical condition analysis results.
[0014] Optionally, the user motion data analysis report further includes a rest period, where the rest period is the time from when the user starts resting to when the user continues the test, and the rest period is determined based on the pre-processed position data, wherein:
[0015] The user starts to rest when the user's movement distance within the first specified time period is less than the first specified distance;
[0016] The user continues the test when the user's moving distance within the second specified time period is greater than a second specified distance.
[0017] Optionally, the motion data analysis model further includes a self-attention mechanism, which is between the one LSTM network and the three fully connected layers.
[0018] The above solution combines the long-term dependency capture capability of LSTM and the dynamic focus adjustment capability of the self-attention mechanism, enabling the motion data analysis model to more effectively handle various complex sequence processing tasks.
[0019] Optionally, before preprocessing the user motion data:
[0020] performing a second normalization on the user motion data;
[0021] Inputting the second normalized user motion data into a motion data prediction model to obtain user motion prediction data, wherein the motion data prediction model includes a network composed of five 1DCNNs in parallel and a 1DCNN, wherein the 1DCNN includes two one-dimensional convolutional layers, a maximum pooling layer, a Flatten layer, and a fully connected layer;
[0022] The user motion data and the user motion prediction data are spliced in chronological order to obtain new user motion data.
[0023] The above solution can not only predict the user's motion data as a backup when the monitoring equipment fails, but also analyze the user's motion data and adapt to very long input sequences.
[0024] In a second aspect, the present invention provides a user motion data analysis device, comprising a preprocessing module and an analysis module, wherein:
[0025] The preprocessing module is used to preprocess the user motion data, wherein the user motion data includes blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data of the user during the walking test. The walking test is performed according to the test route and test duration, and the user motion data is collected at preset time intervals;
[0026] The analysis module is used to input the preprocessed user motion data into a motion data analysis model to obtain a physical condition analysis result. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test duration to the unit duration.
[0027] In a third aspect, the present invention provides a user motion data analysis system, comprising a blood pressure monitoring device, a heart rate monitoring device, a respiratory rate monitoring device, a blood oxygen saturation monitoring device, a position monitoring device, and a user motion data analysis device, wherein:
[0028] The blood pressure monitoring device is configured to start collecting blood pressure at a preset time interval upon receiving a first instruction, and stop collecting blood pressure data upon receiving a second instruction and upload the collected blood pressure data to the user motion data analysis device, wherein the first instruction and the second instruction are sent by the user motion data analysis device;
[0029] The heart rate monitoring device is configured to start collecting heart rate data at the preset time interval upon receiving the first instruction, and stop collecting heart rate data upon receiving the second instruction and upload the collected heart rate data to the user motion data analysis device;
[0030] The respiratory rate monitoring device is configured to start collecting respiratory rate data at the preset time interval upon receiving the first instruction, and stop collecting respiratory rate data upon receiving the second instruction and upload the collected respiratory rate data to the user motion data analysis device;
[0031] The blood oxygen saturation monitoring device is configured to start collecting blood oxygen saturation at the preset time interval upon receiving the first instruction, and stop collecting blood oxygen saturation upon receiving the second instruction and upload the collected blood oxygen saturation data to the user motion data analysis device;
[0032] The position monitoring device is configured to start collecting position data at the preset time interval upon receiving the first instruction, and stop collecting position data upon receiving the second instruction and upload the collected position data to the user motion data analysis device;
[0033] The user motion data analysis device is configured to set a test route and test duration for a walking test; simultaneously send the first instruction and the second instruction to the blood pressure monitoring device, the heart rate monitoring device, the respiratory rate monitoring device, the blood oxygen saturation monitoring device, and the position monitoring device; and perform the following operations upon receiving user motion data, wherein the user motion data includes the blood pressure data, the heart rate data, the respiratory rate data, the blood oxygen saturation data, and the position data:
[0034] Preprocessing the user motion data respectively;
[0035] The preprocessed user motion data is input into the motion data analysis model to obtain the physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
[0036] In a fourth aspect, the present invention provides a computing device, comprising:
[0037] a memory for storing program instructions;
[0038] The processor is configured to call the program instructions stored in the memory and execute any of the above methods according to the obtained program.
[0039] In a fifth aspect, the present invention provides a computer-readable non-volatile storage medium, comprising computer-readable instructions, which enables a computer to execute any of the above methods when the computer reads and executes the computer-readable instructions. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.
[0041] Figure 1 A schematic diagram of a flow chart of a method for analyzing user motion data provided by an embodiment of the present invention;
[0042] Figure 2 A schematic diagram of the structure of a user motion data analysis device provided by an embodiment of the present invention;
[0043] Figure 3 A schematic diagram of the structure of a user motion data analysis system provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0044] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.
[0045] It should be noted that the terms "first", "second" etc. involved in the documents of the present application are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable in appropriate circumstances, so that the embodiments of the present application described herein can be implemented in the order except for the content illustrated or described here. In addition, the terms "including" and "having" and any of their deformations are intended to cover non-exclusive inclusions, for example, the process, method, system, product or equipment comprising a series of steps or units need not be limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.
[0046] Figure 1 The present invention provides a method for analyzing user motion data in detail, including:
[0047] S101: Preprocess user motion data.
[0048] Specifically, the user motion data includes the user's blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data and location data during the walking test. The walking test is conducted according to the test route and test duration, and the user motion data is collected at preset time intervals.
[0049] In one example, the test duration is set to six minutes.
[0050] In one example, the preset time interval is set to every second.
[0051] During the walking test, the user starts the walking test according to the test route, updates the number of completions of the test route after completing a lap, records the test distance, and stops the walking test after the test time is reached.
[0052] In one example, preprocessing includes filtering, completion, denoising, and first normalization.
[0053] Filtering refers to filtering out data that exceeds the normal threshold range, and the normal threshold range is determined by the range specified by the medical indicator.
[0054] Completion refers to using data from adjacent time points to fill in missing data at the current time point to ensure data integrity and correctness. Data from the previous time point is prioritized for completion. For example, if the heart rate at the fifth second is missing, the heart rate at the fourth second is used to fill in the missing data.
[0055] Denoising refers to reducing the noise of data by using sliding average, median filtering, Fourier transform or wavelet transform, among which wavelet transform has the best effect.
[0056] The first normalization is performed according to the following formula:
[0057]
[0058] Where: x * is the data at a certain time point after the first normalization, x is the data at a certain time point, x μ is the mean of all data in each minute.
[0059] S102: Input the pre-processed user motion data into a motion data analysis model to obtain a physical condition analysis result.
[0060] Specifically, the motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series, where the specific number is the ratio of the test time to the unit time.
[0061] In one example, the unit duration is one minute. When the test duration is six minutes and the unit duration is one minute, the specific number is 6.
[0062] First, the two-layer nested LSTM model extracts features from the input time series samples (i.e., preprocessed user motion data), including:
[0063] Split the input time series samples into a specific number of time series sample slices according to the unit time length;
[0064] A specific number of time series sample slices are input into a specific number of LSTM networks connected in parallel. Each LSTM network extracts features from the input time series sample slices to obtain a specific number of feature sets.
[0065] After a specific number of feature sets are spliced into a feature set in chronological order, they are input into a series of LSTM networks. The LSTM network extracts the features between each time series sample fragment in the feature set to obtain the final features.
[0066] Then, three fully connected layers, two Batch Normalization layers, and one Softmax layer classify the extracted final features in the form of probability distribution, and the category with the highest probability is the physical condition analysis result.
[0067] For example, the final feature corresponds to the cardiovascular load level, and the classification result of the cardiovascular load level is {"pressure overload": 0.7, "volume overload": 0.1, "limited diastolic function": 0.1, "normal": 0.1}. The final output is: the cardiovascular load level shows excessive pressure load.
[0068] The two-layer nested LSTM model requires a large sample set for training. The sample set includes data from various dimensions of a large number of users performing walking tests and their physical condition analysis results. The specific training process is as follows:
[0069] The preprocessed five-dimensional time series samples are used as input, and the corresponding physical condition analysis results are used as output;
[0070] Set the model parameters (batch_size, time_step, input_size), where batch_size represents the amount of data for batch processing, i.e., the number of time series sample slices used for training once; time_step represents the time step, i.e., the number of sampling points in the time series sample slice. If sampling is performed once per second, time_step is set to 60; input_size represents the dimension of the input feature data for each time step. Since the motion data in the present invention has 5 dimensions, input_size is set to 5, the number of model training epochs is set to 2000, the model learning rate is set to 0.001, and the learning rate is tuned using the Adam algorithm; alternatively, the learning rate and batch size can be adjusted using the grid search method. The learning rate value range can be set to [0.001, 0.01, 0.1], and the batch size value range can be set to [32, 64, 128]. Construct all possible hyperparameter combinations according to the grid search method and try them one by one to find the best combination;
[0071] Use 60% of the sample set as the training set, 20% as the validation set, and 20% as the test set;
[0072] Input the training set into the double-layer nested LSTM model and train the model according to the given model parameters. The more times the better, the better. You can set it to about 2000.
[0073] The above solution is based on a two-layer nested LSTM model to effectively and accurately analyze user motion data without human intervention, obtaining highly accurate physical function assessment results.
[0074] In one example, an embodiment of the present invention provides a method for analyzing user motion data, further comprising:
[0075] S103. Based on the pre-processed heart rate data, respiratory rate data, blood oxygen saturation data, and position data, a trend change graph of the heart rate, respiratory rate, blood oxygen saturation, and acceleration is drawn with time as the horizontal axis and heart rate, respiratory rate, blood oxygen saturation, and acceleration as the vertical axes, respectively.
[0076] Specifically, the acceleration is calculated based on the preprocessed position data, and the acceleration of the user during the walking test reflects the user's exercise intensity.
[0077] S104. Split the preprocessed heart rate data, respiratory rate data, blood oxygen saturation data, and position data into a specific number of heart rate sets, respiratory rate sets, blood oxygen saturation sets, and position sets according to unit time, calculate the heart rate, respiratory rate, and blood oxygen saturation in each unit time based on the heart rate sets, respiratory rate sets, and blood oxygen saturation sets in each unit time, and calculate the walking speed and walking distance in each unit time based on the position sets in each unit time.
[0078] It should be noted that the heart rate, respiratory rate, and blood oxygen saturation within each unit time are the means of the heart rate set, respiratory rate set, and blood oxygen saturation set within each unit time, respectively.
[0079] S105. Generate a user exercise data analysis report based on the physical condition analysis results, the trend change graph of heart rate, respiratory rate, blood oxygen saturation and acceleration, the heart rate, respiratory rate and blood oxygen saturation in each unit time, the walking speed and walking distance in each unit time and the exercise prescription. The exercise prescription is obtained based on the physical condition analysis results.
[0080] Exercise prescriptions can be given by doctors based on the results of physical condition analysis. The content of exercise prescriptions can include diagnostic suggestions and precautions.
[0081] During the walking test, if the user is in poor physical condition, he or she may need to take a break.
[0082] Furthermore, the user motion data analysis report also includes rest time, which is the time from the start of the user's rest to the continuation of the test. The rest time is determined based on the pre-processed position data, where:
[0083] The user starts to rest when the distance moved by the user within the first specified time period is less than the first specified distance;
[0084] The user continues the test when the user's moving distance within the second specified time period is greater than the second specified distance.
[0085] The first specified duration and the second specified duration are set based on actual conditions and can be the same or different. The first specified distance and the second specified distance are set based on actual conditions and can be the same or different. For example, a user may be determined to be in a resting state when the distance moved within three or four seconds does not exceed 0.5 meters.
[0086] Since the LSTM network cannot process long sequences and highlight important information well, it performs poorly in some cases. The self-attention mechanism simulates the characteristics of human visual attention mechanism and can solve the problems of the LSTM network well.
[0087] In one example, the motion data analysis model also includes a self-attention mechanism between an LSTM network and three fully connected layers.
[0088] The self-attention mechanism uses weight distribution to determine which parts of the input sequence to focus on. This allows the model to dynamically adjust its focus when generating output to better capture key information in the input sequence. For the large amount of motion data collected during user walking tests, the self-attention mechanism can assign different weights to information in the sequence, enhancing the contribution of motion data features at important moments to monitoring the current user's motion status. Combining the LSTM's ability to capture long-term dependencies with the self-attention mechanism's ability to dynamically adjust focus, the motion data analysis model can more effectively handle various complex sequence processing tasks.
[0089] The steps to build a self-attention mechanism are as follows:
[0090] Input sequence, the input of the self-attention mechanism consists of the output obtained after the double-layer nested LSTM layer extracts the time series features;
[0091] Create a query vector (Query), matching question (Key), and target vector (Value). The query, key, and value of the self-attention mechanism are obtained by linearly transforming the same input sequence X.
[0092] Define the encoder input sequence as X and the training weight (randomly generated weight matrix) as W Q 、W K and W V , multiply the input sequence by the weight matrix, and we get Q = XW Q , K=XW K , V=XW V ;
[0093] Calculate the dot product between the query vector Q and the matching vector K. To prevent the dot product from being too large, divide the result by the square root of the dimension of the matching vector K.
[0094] Use the Softmax function to normalize the dot product result to get the attention score;
[0095] The target vector V is weightedly added to the attention score, which is the output of self-attention. The specific formula is as follows:
[0096]
[0097] User movement is a dynamic process, and movement data is constantly changing. During the test, users may encounter monitoring equipment failures, resulting in the inability to pay attention to changes in user movement data in real time.
[0098] In one example, before pre-processing the user motion data:
[0099] Performing a second normalization on the user motion data;
[0100] Inputting the second normalized user motion data into the motion data prediction model to obtain user motion prediction data. The motion data prediction model includes a network composed of five 1DCNNs in parallel and a 1DCNN. The 1DCNN includes two one-dimensional convolutional layers, a maximum pooling layer, a Flatten layer, and a fully connected layer.
[0101] The user motion data and the user motion prediction data are spliced in chronological order to obtain new user motion data.
[0102] Furthermore, the second normalization is maximum and minimum value normalization, and the result is between 0 and 1.
[0103] The convolution layer uses nonlinear activation functions, namely the ReLU activation function and the Sigmoid activation function, to learn nonlinear features. The expression of the ReLU activation function is R(x)=Max(0,x). The feature extraction formula of the one-dimensional convolution layer for time series data is Y=σ(W·X+b), where Y is the extracted feature, σ is the Sigmoid activation function, W is the weight matrix, X is the time series data, and b is the bias vector.
[0104] The calculation formula of 1DCNN is i is the serial number of the convolution kernel, Y(i) is the result of the i-th convolution operation, x is the input time series data, w is the corresponding weight, c is the bias of the convolution kernel, and n=5 is the dimension of the input data.
[0105] The Flatten layer is added by the add function and plays a conversion role, which is used to flatten the data.
[0106] 1DCNN (one-dimensional convolutional neural network) is a widely used network architecture in deep learning. 1DCNN interprets input sequences into blocks or subsequences, and a two-layer nested LSTM model combines the solutions of these subsequences. 1DCNN can process high-dimensional data and automatically extract features. The two-layer nested LSTM model can solve the problem of vanishing gradients and is a good nonlinear model. To address the uncertainty of device changes during user movement, combining the two can not only predict user movement data as a backup in the event of monitoring equipment failure, but also analyze user movement data and adapt to very long input sequences.
[0107] The changes in motion data of different users cannot be predicted using the same motion data prediction model. A motion data prediction model must be established for each user's motion data for learning and training. The training process is as follows:
[0108] Divide the dataset into a training set and a test set, and determine the input and output of the motion data prediction model. For example, for six minutes of user motion data, use the first five minutes of user motion data as input and the last minute of user motion data as output.
[0109] After the second normalization of the dataset, the divided training set is input into the motion data prediction model in chronological order, where: the convolution kernels are 128 and 64 respectively, the convolution kernel size is 2, and the sliding window in the pooling layer is set to 3.
[0110] The input time series data (i.e., user motion data) is time series data with 5 variables (i.e., dimensions) and a time step of 1 minute. Each input variable is first subjected to feature extraction by its own 1DCNN to obtain the corresponding feature vector. Then, all the feature vectors are connected into one vector through the serial 1DCNN to obtain the user motion prediction data.
[0111] Based on the same technical concept, Figure 2 The structure of a user motion data analysis device provided by this embodiment is shown, including a preprocessing module and an analysis module, wherein:
[0112] The preprocessing module is used to preprocess the user's motion data, which includes the user's blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data during the walking test. The walking test is conducted according to the test route and test duration, and the user's motion data is collected at preset time intervals;
[0113] The analysis module is used to input the preprocessed user motion data into the motion data analysis model to obtain physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
[0114] Based on the same technical concept, Figure 3 The structure of a user motion data analysis system provided by this embodiment is shown, including a blood pressure monitoring device, a heart rate monitoring device, a respiratory rate monitoring device, a blood oxygen saturation monitoring device, a position monitoring device, and a user motion data analysis device, wherein:
[0115] The blood pressure monitoring device is configured to start collecting blood pressure data at a preset time interval upon receiving a first instruction, and stop collecting blood pressure data upon receiving a second instruction and upload the collected blood pressure data to the user motion data analysis device, wherein the first instruction and the second instruction are sent by the user motion data analysis device;
[0116] The heart rate monitoring device is configured to start collecting heart rate data at a preset time interval upon receiving a first instruction, and stop collecting heart rate data upon receiving a second instruction and upload the collected heart rate data to a user motion data analysis device;
[0117] The respiratory rate monitoring device is configured to start collecting respiratory rate at a preset time interval upon receiving a first instruction, and stop collecting respiratory rate upon receiving a second instruction and upload the collected respiratory rate data to the user motion data analysis device;
[0118] The blood oxygen saturation monitoring device is configured to start collecting blood oxygen saturation at a preset time interval upon receiving a first instruction, and stop collecting blood oxygen saturation upon receiving a second instruction and upload the collected blood oxygen saturation data to the user motion data analysis device;
[0119] The position monitoring device is configured to start collecting position data at a preset time interval upon receiving a first instruction, and stop collecting position data upon receiving a second instruction and upload the collected position data to the user motion data analysis device;
[0120] The user motion data analysis device is used to set a test route and test duration for a walking test; simultaneously send a first instruction and a second instruction to a blood pressure monitoring device, a heart rate monitoring device, a respiratory rate monitoring device, a blood oxygen saturation monitoring device, and a location monitoring device; and perform the following operations upon receiving user motion data, where the user motion data includes blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data:
[0121] Preprocess user motion data separately;
[0122] The preprocessed user motion data is input into the motion data analysis model to obtain the physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers, and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
[0123] The blood pressure monitoring device, heart rate monitoring device, respiratory rate monitoring device, blood oxygen saturation monitoring device, and trajectory information monitoring device can be connected to the user's motion data analysis device via wired or wireless means. Each monitoring device can be a different sensor, and the heart rate detection device and respiratory rate monitoring device can also be the same sensor. The blood pressure monitoring device is set on the user's upper arm, the heart rate / respiratory rate monitoring device is set in a preset area around the user's heart, the blood oxygen saturation monitoring device is set on the user's fingertips, and the trajectory information monitoring device can be worn on either arm.
[0124] To improve user experience, the user motion data analysis device plays a first voice prompting the user to start a walking test when sending a first instruction, and plays a second voice prompting the user to end the walking test when sending a second instruction.
[0125] Based on the same technical concept, an embodiment of the present invention provides a computing device, including:
[0126] a memory for storing program instructions;
[0127] The processor is used to call the program instructions stored in the memory and execute the above method according to the obtained program.
[0128] Based on the same technical concept, an embodiment of the present invention provides a computer-readable non-volatile storage medium, including computer-readable instructions. When a computer reads and executes the computer-readable instructions, the computer executes the above method.
[0129] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0130] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.
[0131] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0132] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.
[0133] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.
Claims
1. A method for analyzing user motion data, characterized in that: include: Preprocessing user motion data, including blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data of the user during a walking test, wherein the walking test is conducted according to a test route and test duration, and the user motion data is collected at preset time intervals; The preprocessed user motion data is input into the motion data analysis model to obtain the physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
2. The method according to claim 1, characterized in that The preprocessing includes filtering, completion, denoising and first normalization.
3. The method according to claim 1, characterized in that Also includes: Based on the preprocessed heart rate data, respiratory rate data, blood oxygen saturation data, and position data, plot a trend change graph of heart rate, respiratory rate, blood oxygen saturation, and acceleration with time as the horizontal axis and heart rate, respiratory rate, blood oxygen saturation, and acceleration as the vertical axis, respectively. The acceleration is calculated based on the preprocessed position data; Splitting the preprocessed heart rate data, respiratory rate data, blood oxygen saturation data, and position data into the specific number of heart rate sets, respiratory rate sets, blood oxygen saturation sets, and position sets according to the unit time length, respectively, calculating the heart rate, respiratory rate, and blood oxygen saturation in each unit time length based on the heart rate sets, respiratory rate sets, and blood oxygen saturation sets in each unit time length, and calculating the walking speed and walking distance in each unit time length based on the position sets in each unit time length; A user exercise data analysis report is generated based on the physical condition analysis results, the trend change graphs of the heart rate, respiratory rate, blood oxygen saturation and acceleration, the heart rate, respiratory rate and blood oxygen saturation in each unit time, the walking speed and walking distance in each unit time and the exercise prescription, where the exercise prescription is obtained based on the physical condition analysis results.
4. The method according to claim 3, characterized in that The user motion data analysis report also includes a rest period, which is the time from the start of the user's rest to the continuation of the test. The rest period is determined based on the pre-processed position data, wherein: The user starts to rest when the user's movement distance within the first specified time period is less than the first specified distance; The user continues the test when the user's moving distance within the second specified time period is greater than a second specified distance.
5. The method according to claim 1, characterized in that: The motion data analysis model also includes a self-attention mechanism, which is between the one LSTM network and the three fully connected layers.
6. The method according to claim 1, characterized in that Before preprocessing the user motion data: performing a second normalization on the user motion data; Inputting the second normalized user motion data into a motion data prediction model to obtain user motion prediction data, wherein the motion data prediction model includes a network composed of five 1DCNNs in parallel and a 1DCNN, wherein the 1DCNN includes two one-dimensional convolutional layers, a maximum pooling layer, a Flatten layer, and a fully connected layer; The user motion data and the user motion prediction data are spliced in chronological order to obtain new user motion data.
7. A user motion data analysis device, characterized in that: It includes pre-processing module and analysis module, among which: The preprocessing module is used to preprocess the user motion data, wherein the user motion data includes blood pressure data, heart rate data, respiratory rate data, blood oxygen saturation data, and location data of the user during the walking test. The walking test is performed according to the test route and test duration, and the user motion data is collected at preset time intervals; The analysis module is used to input the preprocessed user motion data into a motion data analysis model to obtain a physical condition analysis result. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test duration to the unit duration.
8. A user motion data analysis system, characterized in that: Including blood pressure monitoring equipment, heart rate monitoring equipment, respiratory rate monitoring equipment, blood oxygen saturation monitoring equipment, position monitoring equipment and user motion data analysis equipment, including: The blood pressure monitoring device is configured to start collecting blood pressure at a preset time interval upon receiving a first instruction, and stop collecting blood pressure data upon receiving a second instruction and upload the collected blood pressure data to the user motion data analysis device, wherein the first instruction and the second instruction are sent by the user motion data analysis device; The heart rate monitoring device is configured to start collecting heart rate data at the preset time interval upon receiving the first instruction, and stop collecting heart rate data upon receiving the second instruction and upload the collected heart rate data to the user motion data analysis device; The respiratory rate monitoring device is configured to start collecting respiratory rate data at the preset time interval upon receiving the first instruction, and stop collecting respiratory rate data upon receiving the second instruction and upload the collected respiratory rate data to the user motion data analysis device; The blood oxygen saturation monitoring device is configured to start collecting blood oxygen saturation at the preset time interval upon receiving the first instruction, and stop collecting blood oxygen saturation upon receiving the second instruction and upload the collected blood oxygen saturation data to the user motion data analysis device; The position monitoring device is configured to start collecting position data at the preset time interval upon receiving the first instruction, and stop collecting position data upon receiving the second instruction and upload the collected position data to the user motion data analysis device; The user motion data analysis device is configured to set a test route and test duration for a walking test; simultaneously send the first instruction and the second instruction to the blood pressure monitoring device, the heart rate monitoring device, the respiratory rate monitoring device, the blood oxygen saturation monitoring device, and the position monitoring device; and perform the following operations upon receiving user motion data, wherein the user motion data includes the blood pressure data, the heart rate data, the respiratory rate data, the blood oxygen saturation data, and the position data: Preprocessing the user motion data respectively; The preprocessed user motion data is input into the motion data analysis model to obtain the physical condition analysis results. The motion data analysis model includes a two-layer nested LSTM model consisting of an input layer, a network consisting of a specific number of LSTM networks in parallel, an LSTM network, three fully connected layers, two Batch Normalization layers and a Softmax layer in series. The specific number is the ratio of the test time to the unit time.
9. A computing device, characterized in that include: a memory for storing program instructions; A processor is configured to call the program instructions stored in the memory and execute the method according to any one of claims 1 to 6 according to the obtained program.
10. A computer-readable non-volatile storage medium, characterized in that: The method comprises computer-readable instructions, and when a computer reads and executes the computer-readable instructions, the computer is caused to execute the method according to any one of claims 1 to 6.