Exercise load monitoring method and device of intelligent wearable equipment, equipment and medium

Abstract

The invention relates to an exercise load monitoring method, device and equipment of intelligent wearable equipment and a medium. The method comprises the following steps: acquiring basic data of a user, and collecting breathing audio data, heart rate physiological data and motion posture data of the user; on the basis of the breathing audio data and the heart rate physiological data, combining user basic data to generate exercise intensity parameters; generating motion characteristic parameters based on the motion posture data; and inputting the user basic data, the exercise intensity parameters and the exercise characteristic parameters into an exercise load monitoring machine learning model to generate exercise type information, exercise load information and expected recovery time information. By adopting the method, the exercise monitoring precision can be improved, the dynamic adaptive capacity is enhanced, and the user exercise breathing performance is monitored.

Description

Technical Field

[0001] This application belongs to the field of exercise load monitoring, and in particular relates to a method, device, equipment and medium for exercising load monitoring of a smart wearable device. Background Technology

[0002] Wearable devices, as emerging technological products, are revolutionizing the field of sports training due to their unique advantages such as portability, real-time performance, and data-driven capabilities. Their application prospects are broad; they can not only provide real-time feedback on training data, helping users to precisely optimize their training plans, but also predict athletic performance through data analysis, providing a scientific basis for personalized training.

[0003] Traditional exercise load monitoring primarily relies on heart rate monitors, post-exercise lactate testing, and simple pedometers. Heart rate monitors measure heart rate by detecting the heart's electrical activity, lactate testing assesses exercise intensity by analyzing the lactate levels in the blood after exercise, and pedometers count steps using mechanical or electronic sensors. While these traditional methods provided some data support for exercise load assessment under the conditions at the time, the sheer number of devices involved and the cumbersome monitoring process presented numerous inconveniences. Furthermore, wearable device-based exercise load monitoring methods struggle to integrate multi-dimensional data for comprehensive and accurate exercise load assessment, failing to meet the demands for personalized and refined exercise load monitoring. Moreover, existing wearable device-based exercise load monitoring methods often neglect the user's respiratory characteristics, limiting the analytical dimensions. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, device, equipment, and medium for monitoring the exercise load of a smart wearable device that can improve the accuracy of motion monitoring, enhance dynamic adaptability, and monitor the user's exercise and respiratory performance, in order to address the above-mentioned technical problems.

[0005] Firstly, this application provides a method for monitoring exercise load using a smart wearable device, including...

[0006] Acquire basic user data, and collect user respiratory audio data, heart rate physiological data, and movement posture data;

[0007] Based on respiratory audio data and heart rate physiological data, combined with basic user data, exercise intensity parameters are generated;

[0008] Generate motion feature parameters based on motion posture data;

[0009] User basic data, exercise intensity parameters, and exercise characteristic parameters are input into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

[0010] In one embodiment, the exercise intensity parameter includes a respiratory intensity parameter, which is generated based on respiratory audio data and heart rate physiological data, combined with basic user data, and includes:

[0011] Respiratory frequency parameters and Mel spectra of respiratory audio data are extracted from respiratory audio data, and Mel spectrum entropy parameters and energy concentration parameters of Mel spectra are calculated.

[0012] The user's respiratory load parameters are calculated based on the respiratory rate parameter, Mel spectrum entropy parameter, and energy concentration parameter.

[0013] Respiratory intensity parameters are constructed based on respiratory rate parameters, Mel spectrum entropy parameters, energy concentration parameters, and respiratory load parameters;

[0014] The expressions for the Mel spectral entropy parameter and the energy concentration parameter are as follows:

[0015]

[0016] In the formula, Let S be the Mel spectral entropy parameter at time τ, M be the total number of Mel filters in the Mel filter bank, and S be the... τ (m) represents the energy value of the Mel spectral component of the m-th Mel filter in the audio frame of the breathing audio data at time τ. S is the energy concentration parameter at time τ. H This is the set of serial numbers for high-frequency Mel filters.

[0017] In one embodiment, the exercise intensity parameters include heart rate intensity parameters and cardiopulmonary coordination parameters. These parameters are generated based on respiratory audio data and heart rate physiological data, combined with basic user data.

[0018] The user's heart rate reserve percentage parameter is calculated based on heart rate physiological data, and the heart rate reserve percentage parameter is set as the heart rate intensity parameter;

[0019] Cardiopulmonary coordination parameters are calculated based on respiratory load parameters and heart rate reserve percentage parameters.

[0020] In one embodiment, the expressions for the respiratory load parameter and the cardiopulmonary coordination parameter are:

[0021]

[0022]

[0023] In the formula, Let ρ be the cardiopulmonary coordination parameter at time τ, ρ be the cardiopulmonary coordination index coefficient, and ||·||1 be the first norm. and These are the maximum and minimum preset respiratory load parameters, respectively. Let τ be the respiratory load parameter. The percentage of heart rate reserve at time τ. and These are the Mel spectrum entropy parameter, energy concentration parameter, and respiratory rate parameter at time τ, respectively.

[0024] Methods for monitoring exercise load using smart wearable devices also include:

[0025] If the cardiopulmonary coordination parameters are lower than the preset cardiopulmonary coordination threshold, an abnormal breathing pattern prompt message will be generated.

[0026] In one embodiment, the exercise load monitoring method for smart wearable devices further includes:

[0027] Obtain the standard posture feature parameters corresponding to the motion type information based on the motion type information;

[0028] Motion guidance information is generated by combining motion feature parameters and standard posture feature parameters;

[0029] Obtain standard respiratory intensity parameters corresponding to exercise load information based on exercise load information;

[0030] Breathing guidance information is generated by combining breathing intensity parameters and standard breathing parameters.

[0031] In one embodiment, the exercise load monitoring machine learning model includes an exercise type feature recognition layer, an exercise load feature analysis layer, and a recovery time prediction layer. User basic data, exercise intensity parameters, and exercise feature parameters are input into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information, including:

[0032] User basic data, exercise intensity parameters, and exercise characteristic parameters are input into the exercise type feature recognition layer of the exercise load monitoring machine learning model to generate exercise type information;

[0033] Based on the exercise intensity parameters and exercise type information at the same moment of exercise, combined with user basic data, an exercise intensity feature vector is constructed. The exercise intensity feature vector is then input into the exercise load feature analysis layer of the exercise load monitoring machine learning model to generate exercise load information at the moment of exercise.

[0034] Based on the exercise load information, an exercise load time series is constructed, and the exercise load time series is input into the recovery time prediction layer of the exercise load monitoring machine learning model to generate expected recovery time information.

[0035] In one embodiment, the exercise load monitoring machine learning model is a hybrid machine learning model constructed based on an improved random forest machine learning model, an improved backpropagation neural network model, and an improved long short-term memory neural network model; the exercise type feature recognition layer is a machine learning model constructed based on an improved random forest machine learning model; the exercise load feature analysis layer is a machine information model constructed based on an improved backpropagation neural network model; and the recovery time prediction layer is a machine learning model constructed based on an improved long short-term memory neural network model. The exercise load monitoring method for smart wearable devices further includes:

[0036] Obtain user basic data training dataset, exercise intensity parameter training dataset, exercise feature parameter training dataset, exercise type label training dataset, metabolic equivalent label training dataset, and recovery time label training dataset;

[0037] The user basic data training dataset, the motion intensity parameter training dataset, and the motion feature parameter training dataset are used as the input to the motion type feature recognition layer, and the motion type label training dataset is used as the output to train and construct the motion type feature recognition layer.

[0038] The exercise load feature analysis layer is trained and constructed by using the exercise intensity parameter training dataset and the exercise type label training dataset as inputs and the metabolic equivalent label training dataset as outputs.

[0039] The recovery time prediction layer is trained by using the metabolic equivalent label training dataset as input and the recovery time label training dataset as output.

[0040] Secondly, this application also provides a motion load monitoring device for a smart wearable device, comprising:

[0041] The basic data acquisition module is used to acquire basic user data, including user respiratory audio data, heart rate physiological data, and movement posture data.

[0042] The exercise intensity analysis module is used to generate exercise intensity parameters based on respiratory audio data and heart rate physiological data, combined with the user's basic data;

[0043] The motion posture analysis module is used to generate motion feature parameters based on motion posture data;

[0044] The exercise load monitoring module is used to input user basic data, exercise intensity parameters, and exercise characteristic parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

[0045] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method as described in any of the first aspects of this application.

[0046] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method as described in any of the first aspects of this application.

[0047] The aforementioned smart wearable device's exercise load monitoring method, apparatus, equipment, and medium, by integrating multi-dimensional sensor data, can achieve multi-dimensional characterization of exercise load, improving the comprehensiveness of monitoring. It can effectively reduce the impact of environmental interference on the analysis structure of a single indicator, and improve the stability of monitoring results. By combining user baseline data for comprehensive analysis, it can adjust data weights based on user baseline data, making the exercise load monitoring results more closely match individual physiological characteristics. By constructing a personalized machine learning model based on user historical data, it can generate predictive parameters for exercise type, load level, and recovery time, enabling automatic identification of exercise type, quantification of exercise load, and prediction of expected recovery time. By acquiring respiratory audio data and generating exercise intensity parameters based on respiratory audio data and heart rate physiological data, combined with user baseline data, it can comprehensively consider the impact of the user's respiratory state on exercise state, improving the accuracy and applicability of exercise load monitoring. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 A schematic diagram illustrating the application environment of a method for monitoring the exercise load of a smart wearable device, provided in one embodiment of this application;

[0050] Figure 2 A flowchart illustrating a method for monitoring exercise load using a smart wearable device, provided as an embodiment of this application;

[0051] Figure 3 A flowchart illustrating another method for monitoring exercise load in a smart wearable device according to an embodiment of this application;

[0052] Figure 4 This is a schematic diagram of the structure of a motion load monitoring device for a smart wearable device provided in one embodiment of this application. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0054] The exercise load monitoring method for smart wearable devices provided in this application embodiment can be applied to, for example... Figure 1 The application environment is illustrated. The smart wearable device 101 can communicate with the smart terminal 102 via a communication channel, and the smart terminal 102 can communicate with the server 103 via a network. The smart wearable device 101 can obtain basic user data through the smart terminal 102, and can send the collected exercise load monitoring information to the smart terminal 102. The server 103 can update the data and applications in the smart terminal 102 and the smart wearable device 101. The database 104 can be used to store the data required for processing by the server 103. The database 104 can be integrated into the server 103, or deployed in the cloud or other network servers. The smart terminal 102 can be, but is not limited to, various smart terminal devices such as smartphones and tablets; the smart wearable device 101 can be, but is not limited to, smartwatches, smart bracelets, and head-mounted devices; the server 104 can be implemented using a standalone server or a server cluster composed of multiple servers.

[0055] In one exemplary embodiment, such as Figure 2 As shown, a method for monitoring exercise load in a smart wearable device is provided, which can be applied to... Figure 1 Taking the smart wearable device 101 as an example, the explanation includes the following steps S201 to S204. Wherein:

[0056] Step S201: Obtain basic user data and collect user respiratory audio data, heart rate physiological data and movement posture data.

[0057] Specifically, the smart wearable device 101 can connect to a smart terminal via near-field communication protocols such as Bluetooth. After the smart wearable device 101 passes the authorization verification of the smart terminal, it can obtain the user's basic user data stored in the smart terminal. The smart wearable device 101 can also collect the user's respiratory audio data, heart rate physiological data, and movement posture data through the multimodal sensing devices mounted on the smart wearable device 101.

[0058] Optionally, user basic data may include, but is not limited to, the user's age, gender, weight, height, and maximum heart rate. resting heart rate Maximum value of respiratory load parameters and the minimum value of the preset respiratory load parameters

[0059] Optionally, the smart wearable device 101 can obtain the maximum value of the respiratory load parameter from the user's basic data stored in the smart terminal and calculated by the server from the server. and the minimum value of the preset respiratory load parameters

[0060] As an illustration, the server can generate maximum values ​​of respiratory load parameters for different age groups, genders, and physical conditions using respiratory test data from a large number of healthy individuals. and the minimum value of the preset respiratory load parameters The standard value.

[0061] Optionally, the smart wearable device 101 can collect the user's audio signal and obtain the time-domain audio waveform data of the audio signal through the audio sensor in the multimodal sensing device mounted on the smart wearable device 101.

[0062] Furthermore, the smart wearable device 101 can employ a preset filtering mechanism to remove environmental noise. Through wavelet transform and frequency domain analysis, it eliminates environmental noise interference while retaining features related to motion intensity, thereby separating external noise from the audio signal to obtain a denoised motion breathing rhythm sound signal. The preset filtering mechanism may include, but is not limited to, Kalman filtering and adaptive filtering.

[0063] Optionally, the smart wearable device 101 can acquire the user's heart rate physiological data through the photoplethysmography sensor in the multimodal sensing device mounted on the smart wearable device 101.

[0064] Optionally, the smart wearable device 101 can acquire the user's motion attitude data through the inertial measurement unit (IMU) and global positioning sensor unit in the multimodal sensing device on the smart wearable device 101. Among them, the inertial measurement unit (IMU) can integrate an accelerometer and a gyroscope.

[0065] Step S202: Based on respiratory audio data and heart rate physiological data, combined with user basic data, generate exercise intensity parameters.

[0066] Specifically, the smart wearable device 101 can perform frame-by-frame windowing processing on the denoised motion respiratory rhythm sound signal in the collected respiratory audio data, and apply a Hamming window function to each frame to obtain the framed motion respiratory rhythm sound signal. Based on the framed motion respiratory rhythm sound signal, the smart wearable device 101 can calculate the respiratory load parameters of each respiratory frame in the framed motion respiratory rhythm sound signal. The smart wearable device 101 can detect real-time heart rate based on physiological heart rate data. Combined with the user's basic data, the maximum heart rate and resting heart rate Calculate the user's heart rate reserve percentage parameter The smart wearable device 101 can be based on respiratory load parameters Heart rate reserve percentage parameter Build and generate the user's exercise intensity parameters.

[0067] Furthermore, the smart wearable device 101 can base its functions on the respiratory load parameters at the corresponding time of each respiratory frame. The maximum value of the respiratory load parameter in the user's basic data. and the minimum value of the preset respiratory load parameters The relative respiratory load parameters were calculated. Relative respiratory load parameters The expression can be: Relative respiratory load parameters It can be used to characterize the respiratory system's stress response to exercise.

[0068] Step S203: Generate motion feature parameters based on motion posture data.

[0069] Specifically, motion posture data may include, but is not limited to, positioning height data, planar position data, motion trajectory data, and motion amplitude data. Motion feature parameters may include, but are not limited to, height change feature parameters, planar position change feature parameters, and motion action feature parameters. The smart wearable device 101 can generate height change feature parameters based on the range and rate of change of positioning height data, generate planar position change feature parameters based on the range and rate of change of planar position data, and generate motion action feature parameters by combining motion trajectory data and motion amplitude data.

[0070] For example, taking running as an example, the motion characteristic parameters may include, but are not limited to, arm swing range parameters and trunk stability parameters.

[0071] Step S204: Input the user's basic data, exercise intensity parameters, and exercise characteristic parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

[0072] Specifically, the smart wearable device 101 can input the acquired user basic data, generated exercise intensity parameters, and generated exercise feature parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information. The exercise load monitoring machine learning model can be mounted on the smart wearable device 101, on a smart terminal, or on a server; there are no limitations on this.

[0073] For example, taking the exercise load monitoring machine learning model mounted on a smart wearable device 101 as an example, since the existing smart wearable device 101 typically has a RAM size of 512MB to 2GB, if the exercise load monitoring machine learning model needs to be mounted on the smart wearable device 101, the RAM occupied by the exercise load monitoring machine learning model should preferably not exceed one-fifth of the total RAM of the smart wearable device 101. By designing targeted models for exercise type information, exercise load information, and expected recovery time information respectively, the RAM occupied by the exercise load monitoring machine learning model mounted on the smart wearable device 101 can meet the corresponding technical specifications.

[0074] Optionally, the user basic data acquired by the smart wearable device 101 may include the user's historical exercise records and the user's average exercise interval corresponding to the user's historical exercise records. When the value of the expected recovery time interval corresponding to the expected recovery time information output by the exercise load monitoring machine learning model is greater than the value of the user's average exercise interval, the smart wearable device 101 can generate an exercise intensity exceeding the standard prompt information.

[0075] The aforementioned method for monitoring exercise load using smart wearable devices comprehensively analyzes user basic data, respiratory audio data, heart rate physiological data, and exercise posture data. This allows for the capture of the user's physiological and kinematic characteristics during exercise from multiple dimensions, avoiding the uncertainty of a single data source and significantly improving the completeness and reliability of the data. It can generate more accurate exercise intensity and characteristic parameters, thereby enhancing the reliability and effectiveness of the exercise load monitoring method. By comprehensively analyzing respiratory audio and heart rate physiological data, exercise intensity can be quantified more accurately, precisely reflecting the changing trend of exercise intensity during exercise, and enabling the detection and analysis of the user's breathing patterns. Furthermore, by constructing a machine learning model for exercise load monitoring, the complex relationships and patterns hidden in multi-source data can be deeply mined, generating interrelated and highly reliable information on exercise type, exercise load, and expected recovery time. This not only improves the accuracy of exercise load assessment but also enhances the scientific validity and rationality of the expected recovery time, further improving the safety and systematic nature of exercise training and preventing chronic fatigue and sports injuries.

[0076] In an optional embodiment of this application, the exercise intensity parameter may include the respiratory intensity parameter, please refer to... Figure 2 and Figure 3 Step S202, based on respiratory audio data and heart rate physiological data, combined with user baseline data, generates exercise intensity parameters, which may include:

[0077] Step S302: Extract the respiratory frequency parameter and the Mel spectrum of the respiratory audio data based on the respiratory audio data, and calculate the Mel spectrum entropy parameter and energy concentration parameter of the Mel spectrum.

[0078] Specifically, smart wearable devices can extract respiratory frequency parameters from respiratory audio data based on peak detection algorithms. They can also apply a Mel filter bank to map the linear spectrum of the respiratory audio data to a Mel scale based on the framed signal of the respiratory rhythm sound, obtaining the Mel spectrum of the respiratory audio data, and calculating the Mel spectral entropy parameter and energy concentration parameter of the Mel spectrum.

[0079] Step S303: Calculate the user's respiratory load parameters based on the respiratory rate parameter, Mel spectrum entropy parameter, and energy concentration parameter.

[0080] Specifically, smart wearable devices can calculate a user's respiratory load parameters based on the product of respiratory rate parameters, Mel spectrum entropy parameters, and energy concentration parameters.

[0081] Step S304: Construct respiratory intensity parameters based on respiratory rate parameters, Mel spectrum entropy parameters, energy concentration parameters, and respiratory load parameters.

[0082] Optionally, the expression for the respiratory intensity parameter can be:

[0083]

[0084] In the formula, Let τ be the respiratory intensity parameter. and These are the respiratory load parameter, the Mel spectrum entropy parameter, the energy concentration parameter, and the respiratory rate parameter at time τ, respectively.

[0085] The expressions for the Mel spectral entropy parameter and the energy concentration parameter can be:

[0086]

[0087]

[0088] In the formula, Let S be the Mel spectral entropy parameter at time τ, M be the total number of Mel filters in the Mel filter bank, and S be the...τ (m) represents the energy value of the Mel spectral component of the m-th Mel filter in the audio frame of the breathing audio data at time τ. S is the energy concentration parameter at time τ. H This is the set of serial numbers for high-frequency Mel filters.

[0089] Optional, Mel-spectral entropy parameter at time τ The complexity of the respiratory signal at time τ can be reflected by calculating the uncertainty of the Mel spectrum energy distribution. Mel spectrum entropy parameter. A higher value indicates a more irregular breathing pattern.

[0090] Optional, energy concentration parameter at time τ It can characterize the energy proportion of a high-frequency Mel filter, and the energy concentration parameter at time τ. A higher value indicates more rapid breathing.

[0091] In the above-mentioned method for monitoring exercise load in smart wearable devices, by introducing Mel spectrum entropy and energy concentration, the complex characteristics of breathing can be comprehensively characterized from different dimensions, thereby improving the accuracy of exercise intensity assessment.

[0092] In an optional embodiment of this application, the exercise intensity parameters may include heart rate intensity parameters and cardiopulmonary coordination parameters, please refer to... Figure 2 and Figure 3 Step S202, based on respiratory audio data and heart rate physiological data, combined with user baseline data, generates exercise intensity parameters, which may include:

[0093] Step S305: Calculate the user's heart rate reserve percentage parameter based on heart rate physiological data, and set the heart rate reserve percentage parameter as the heart rate intensity parameter.

[0094] Optional, percentage of heart rate reserve at time τ The expression can be:

[0095]

[0096] Step S306: Calculate cardiopulmonary coordination parameters based on respiratory load parameters and heart rate reserve percentage parameters.

[0097] The exercise load monitoring method of the aforementioned smart wearable device calculates cardiopulmonary coordination parameters based on respiratory load parameters and heart rate reserve percentage parameters. This method can comprehensively consider the load and interrelationship of the two key physiological systems of breathing and heart rate, thereby comprehensively assessing the coordination of cardiopulmonary function during exercise. This provides a more comprehensive and in-depth perspective for the overall assessment of exercise load, further improving the accuracy and scientific nature of exercise load monitoring.

[0098] In an optional embodiment of this application, the expressions for the respiratory load parameters and cardiopulmonary coordination parameters can be:

[0099]

[0100]

[0101] In the formula, Let ρ be the cardiopulmonary coordination parameter at time τ, ρ be the cardiopulmonary coordination index coefficient, and ||·||1 be the first norm. and These are the maximum and minimum preset respiratory load parameters, respectively. Let τ be the respiratory load parameter. The percentage of heart rate reserve at time τ. and These are the Mel spectrum entropy parameter, energy concentration parameter, and respiratory rate parameter at time τ, respectively.

[0102] Optionally, respiratory load parameters at time τ. The maximum value of the preset respiratory load parameters and the minimum value of the preset respiratory load parameters The proportional relationship between them It can characterize the user's respiratory load. A higher value indicates a higher level of respiratory load on the user.

[0103] In the formula, Let τ be the heart rate parameter. This refers to the resting heart rate parameter. This is the maximum heart rate parameter.

[0104] Indicative of respiratory load parameters at time τ It can be used as input to a machine learning model for exercise load monitoring, thus allowing the respiratory load parameters at time τ to be obtained through the machine learning model for exercise load monitoring. The weighting coefficients, at this point, do not need to first adjust the respiratory rate parameter at time τ. Mel spectrum entropy parameter at time τ Energy concentration parameters at time τ The product is then weighted and converted.

[0105] Indicatively, as the user's exercise load increases, the user's breathing will first become irregular. At this time, the Mel-spectral entropy parameter... The value can increase. After the user gradually adapts to the exercise load, breathing will become even again; at this point, the Mel-spectral entropy parameter... The value can be lowered, but the increase in respiratory rate leads to a decrease in the Mel spectral entropy parameter. The value will still be higher than the resting Mel-frequency entropy parameter. The value of respiratory rate. As the user's exercise load increases, the user's respiratory rate can gradually increase; respiratory rate parameter The value can be gradually increased. As the user's exercise load increases, the energy proportion of the high-frequency band in the user's breathing audio can increase, and the energy concentration parameter... The value can be increased.

[0106] Please refer to Figure 3 The method for monitoring exercise load using smart wearable devices may also include:

[0107] Step S307: If the cardiopulmonary coordination parameters are lower than the preset cardiopulmonary coordination threshold, generate a breathing pattern abnormality prompt message.

[0108] Optionally, when the user's breathing is too rapid and irregular, the respiratory rate parameter... Values ​​of energy concentration parameters The value and Mel spectrum entropy parameter The value of can increase significantly, at which point the respiratory load parameter The value can increase significantly, which can lead to The value is significantly higher than the percentage heart rate reserve parameter. The value of can thus make the cardiopulmonary coordination parameter The breathing pattern falls below a preset cardiopulmonary coordination threshold. In this case, the smart wearable device can generate an abnormal breathing pattern alert.

[0109] Among the aforementioned methods for monitoring exercise load using smart wearable devices, the comprehensive consideration of cardiopulmonary coordination parameters can better reflect the actual working state and degree of cooperation of the cardiopulmonary system during exercise, providing more in-depth physiological function assessment indicators for exercise load monitoring. Abnormal breathing pattern alerts can promptly detect potential breathing problems during exercise, preventing abnormalities such as disordered breathing rhythm, insufficient breathing depth, or mismatch between breathing and heart rate. This helps users quickly adjust their breathing patterns, thus avoiding increased cardiopulmonary burden, reduced exercise efficiency, or even exercise injury caused by poor breathing habits. This enhances the real-time performance and interactivity of the exercise load monitoring system, improving exercise safety.

[0110] In an optional embodiment of this application, please refer to Figure 3 The method for monitoring exercise load using smart wearable devices may also include:

[0111] Step S310: Obtain the standard posture feature parameters corresponding to the motion type information based on the motion type information.

[0112] Step S311: Generate motion guidance information by combining motion feature parameters and standard posture feature parameters.

[0113] Step S312: Obtain the standard respiratory intensity parameters corresponding to the exercise load information based on the exercise load information.

[0114] Step S313: Generate breathing guidance information by combining breathing intensity parameters and standard breathing parameters.

[0115] In the above-mentioned method for monitoring exercise load in smart wearable devices, by combining exercise characteristic parameters to generate exercise guidance information, it is possible to accurately identify deviations in the user's exercise posture, thereby guiding the user to gradually correct incorrect exercise posture and improve exercise quality; by combining breathing intensity parameters to generate breathing guidance information, it is possible to gain a deeper understanding of the user's adaptability to breathing patterns under different exercise intensities, thereby reducing the burden on the user's respiratory system and improving the user's exercise performance.

[0116] In an optional embodiment of this application, the exercise load monitoring machine learning model may include an exercise type feature recognition layer, an exercise load feature analysis layer, and a recovery time prediction layer. User basic data, exercise intensity parameters, and exercise feature parameters are input into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information, which may include:

[0117] Specifically, smart wearable devices can input the acquired user basic data, generated exercise intensity parameters, and generated exercise feature parameters into the exercise type feature recognition layer of the exercise load monitoring machine learning model to generate exercise type information.

[0118] Specifically, smart wearable devices can construct a motion intensity feature vector based on motion intensity parameters and motion type information at the same moment of motion, combined with user basic data. The smart wearable device can then input the constructed motion intensity feature vector into the motion load feature analysis layer of the motion load monitoring machine learning model to generate motion load information at the moment of motion.

[0119] Specifically, smart wearable devices can construct a time series of exercise load based on exercise load information, and input the constructed time series of exercise load into the recovery time prediction layer of the exercise load monitoring machine learning model to generate expected recovery time information.

[0120] In the above-mentioned method for monitoring exercise load in smart wearable devices, by generating exercise type information based on the exercise type feature recognition layer, the current exercise type of the user can be accurately identified, thus improving the accuracy of exercise load monitoring; by generating exercise load information at the moment of exercise based on the exercise load feature analysis layer, the actual physical load of the user under a specific exercise type can be accurately reflected, thus improving the safety and effectiveness of exercise training; by constructing an exercise load time series and inputting it into the recovery time prediction layer to generate expected recovery time information, the impact of dynamic changes in exercise load on recovery time can be fully analyzed, thereby improving the systematicness and coherence of exercise training.

[0121] In an optional embodiment of this application, the exercise load monitoring machine learning model can be a hybrid machine learning model constructed based on an improved random forest machine learning model, an improved backpropagation neural network model, and an improved long short-term memory neural network model; the exercise type feature recognition layer can be a machine learning model constructed based on an improved random forest machine learning model; the exercise load feature analysis layer can be a machine information model constructed based on an improved backpropagation neural network model; and the recovery time prediction layer can be a machine learning model constructed based on an improved long short-term memory neural network model. The exercise load monitoring method for smart wearable devices may further include:

[0122] Specifically, servers that can communicate with smart wearable devices can obtain user basic data training datasets, exercise intensity parameter training datasets, exercise feature parameter training datasets, exercise type label training datasets, metabolic equivalent label training datasets, and recovery time label training datasets.

[0123] Specifically, a server that can communicate with smart wearable devices can use the user basic data training dataset, the motion intensity parameter training dataset, and the motion feature parameter training dataset as inputs to the motion type feature recognition layer, and use the motion type label training dataset as the output of the motion type feature recognition layer to train and construct the motion type feature recognition layer.

[0124] Optionally, the motion type feature recognition layer can be built based on an improved random forest algorithm, which can solve the multimodal data classification problem. The user basic data training dataset may include, but is not limited to, age data, gender data, weight data, and maximum heart rate data. The exercise intensity parameter training dataset may include, but is not limited to, respiratory rate data, Mel spectrum entropy data, and heart rate reserve percentage data. The motion feature parameter training dataset may include, but is not limited to, acceleration variance data, gyroscope angular rate data, motion frequency data, altitude change data, and planar range data.

[0125] Specifically, a server that can communicate with smart wearable devices can use the exercise intensity parameter training dataset and the exercise type label training dataset as input to the exercise load feature analysis layer, and the metabolic equivalent label training dataset as output to train and construct the exercise load feature analysis layer.

[0126] Optionally, the exercise load feature analysis layer can be constructed based on an improved backpropagation neural network, which can establish a nonlinear mapping from exercise intensity features to metabolic equivalents (MET). The exercise load feature analysis layer can have three fully connected layers.

[0127] Specifically, a server that can communicate with smart wearable devices can use the metabolic equivalent label training dataset as the input to the recovery time prediction layer and the recovery time label training dataset as the output to train and construct the recovery time prediction layer.

[0128] Optionally, a recovery time prediction layer based on an improved Long Short-Term Memory (LSTM) neural network model can be constructed to specifically handle the temporal characteristics of exercise load. The forgetting gate in the improved LSTM neural network model can simulate the exercise recovery process.

[0129] As an illustration, the machine learning model for motion load monitoring in this embodiment adopts a three-layer cascaded hybrid architecture. Through division of labor and cooperation, it can achieve modular decoupling, feature reuse and enhancement, and computational efficiency optimization from motion type identification to load assessment and recovery prediction.

[0130] In the aforementioned method for monitoring exercise load in smart wearable devices, a machine learning model for exercise load monitoring is implemented through a hybrid model architecture based on improved random forest, improved backpropagation neural network, and improved long short-term memory neural network. At the exercise type feature recognition layer, it can efficiently process high-dimensional data, improving the accuracy and stability of exercise type recognition. At the exercise load feature analysis layer, by adjusting neuron weights, it accurately maps complex nonlinear relationships, improving the scientific rigor and accuracy of exercise load assessment. At the recovery time prediction layer, thanks to its unique memory unit structure, it selectively retains and utilizes historical information, accurately capturing the long-term dependencies of the exercise load time series, thus improving the accuracy of recovery time prediction.

[0131] In one exemplary embodiment of this application, as shown in the figure, a method for monitoring exercise load using a smart wearable device is provided, which may include:

[0132] Step S301: Obtain basic user data and collect user's respiratory audio data, heart rate physiological data, and movement posture data.

[0133] Step S302: Extract the respiratory frequency parameter and the Mel spectrum of the respiratory audio data based on the respiratory audio data, and calculate the Mel spectrum entropy parameter and energy concentration parameter of the Mel spectrum.

[0134] Step S303: Calculate the user's respiratory load parameters based on the respiratory rate parameter, Mel spectrum entropy parameter, and energy concentration parameter.

[0135] Step S304: Construct respiratory intensity parameters based on respiratory rate parameters, Mel spectrum entropy parameters, energy concentration parameters, and respiratory load parameters.

[0136] Step S305: Calculate the user's heart rate reserve percentage parameter based on heart rate physiological data, and set the heart rate reserve percentage parameter as the heart rate intensity parameter.

[0137] Step S306: Calculate cardiopulmonary coordination parameters based on respiratory load parameters and heart rate reserve percentage parameters.

[0138] Step S307: If the cardiopulmonary coordination parameters are lower than the preset cardiopulmonary coordination threshold, generate a breathing pattern abnormality prompt message.

[0139] Step S308: Generate motion feature parameters based on motion posture data.

[0140] Step S309: Input the user's basic data, exercise intensity parameters, and exercise characteristic parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

[0141] Step S310: Obtain the standard posture feature parameters corresponding to the motion type information based on the motion type information.

[0142] Step S311: Generate motion guidance information by combining motion feature parameters and standard posture feature parameters.

[0143] Step S312: Obtain the standard respiratory intensity parameters corresponding to the exercise load information based on the exercise load information.

[0144] Step S313: Generate breathing guidance information by combining breathing intensity parameters and standard breathing parameters.

[0145] The aforementioned method for monitoring exercise load using smart wearable devices achieves multi-dimensional and comprehensive accurate monitoring of exercise load; by generating abnormal breathing pattern alerts, it can promptly detect and warn of abnormal breathing patterns during exercise, enabling users to quickly recognize potential breathing problems; through exercise guidance and breathing guidance information, it can provide users with personalized and precise exercise and breathing guidance plans, helping users to standardize exercise movements and improve breathing patterns, thereby enhancing exercise quality; through exercise load monitoring machine learning models, it can achieve intelligent and efficient exercise load assessment, lowering the barrier for users to obtain professional exercise load assessments and promoting the application of exercise load monitoring technology in daily exercise and health management.

[0146] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0147] Based on the same inventive concept, this application also provides a motion load monitoring device for a smart wearable device to implement the motion load monitoring method for the smart wearable device described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the motion load monitoring device for a smart wearable device provided below can be found in the limitations of the motion load monitoring method for smart wearable devices described above, and will not be repeated here.

[0148] In one exemplary embodiment, such as Figure 4 As shown, a smart wearable device for monitoring exercise load 400 is provided, comprising:

[0149] The basic data acquisition module 401 can be used to acquire basic user data and collect user respiratory audio data, heart rate physiological data and movement posture data.

[0150] The exercise intensity analysis module 402 can be used to generate exercise intensity parameters based on respiratory audio data and heart rate physiological data, combined with user basic data.

[0151] The motion posture analysis module 403 can be used to generate motion feature parameters based on motion posture data.

[0152] The exercise load monitoring module 404 can be used to input user basic data, exercise intensity parameters and exercise characteristic parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information and expected recovery time information.

[0153] In an optional embodiment of this application, the exercise intensity analysis module 402 can also be used for:

[0154] Respiratory frequency parameters and Mel spectra of respiratory audio data are extracted from respiratory audio data, and Mel spectrum entropy parameters and energy concentration parameters of Mel spectra are calculated.

[0155] The user's respiratory load parameters are calculated based on the respiratory rate parameter, Mel spectrum entropy parameter, and energy concentration parameter.

[0156] Respiratory intensity parameters are constructed based on respiratory rate parameters, Mel spectrum entropy parameters, energy concentration parameters, and respiratory load parameters.

[0157] In an optional embodiment of this application, the exercise intensity analysis module 402 can also be used for:

[0158] The user's heart rate reserve percentage parameter is calculated based on heart rate physiological data, and the heart rate reserve percentage parameter is set as the heart rate intensity parameter.

[0159] Cardiopulmonary coordination parameters are calculated based on respiratory load parameters and heart rate reserve percentage parameters.

[0160] In an optional embodiment of this application, the exercise load monitoring device 400 of the smart wearable device can also be used for:

[0161] If the cardiopulmonary coordination parameters are lower than the preset cardiopulmonary coordination threshold, an abnormal breathing pattern prompt message will be generated.

[0162] In an optional embodiment of this application, the exercise load monitoring method for smart wearable devices further includes:

[0163] Obtain the standard posture feature parameters corresponding to the motion type information based on the motion type information.

[0164] Motion guidance information is generated by combining motion characteristic parameters and standard posture characteristic parameters.

[0165] Obtain the standard respiratory intensity parameters corresponding to the exercise load information based on the exercise load information.

[0166] Breathing guidance information is generated by combining breathing intensity parameters and standard breathing parameters.

[0167] In an optional embodiment of this application, the exercise load monitoring module 404 can also be used for:

[0168] User basic data, exercise intensity parameters, and exercise characteristic parameters are input into the exercise type feature recognition layer of the exercise load monitoring machine learning model to generate exercise type information.

[0169] Based on the exercise intensity parameters and exercise type information at the same moment of exercise, and combined with user basic data, an exercise intensity feature vector is constructed. The exercise intensity feature vector is then input into the exercise load feature analysis layer of the exercise load monitoring machine learning model to generate exercise load information at the moment of exercise.

[0170] Based on the exercise load information, an exercise load time series is constructed, and the exercise load time series is input into the recovery time prediction layer of the exercise load monitoring machine learning model to generate expected recovery time information.

[0171] In an optional embodiment of this application, the exercise load monitoring device 400 of the smart wearable device can also be used for:

[0172] Obtain user basic data training dataset, exercise intensity parameter training dataset, exercise feature parameter training dataset, exercise type label training dataset, metabolic equivalent label training dataset, and recovery time label training dataset;

[0173] The user basic data training dataset, the motion intensity parameter training dataset, and the motion feature parameter training dataset are used as the input to the motion type feature recognition layer, and the motion type label training dataset is used as the output to train and construct the motion type feature recognition layer.

[0174] The exercise load feature analysis layer is trained and constructed by using the exercise intensity parameter training dataset and the exercise type label training dataset as inputs and the metabolic equivalent label training dataset as outputs.

[0175] The recovery time prediction layer is trained by using the metabolic equivalent label training dataset as input and the recovery time label training dataset as output.

[0176] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, the processor executing the computer program to implement the steps of the above-described method for monitoring the motion load of a smart wearable device.

[0177] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0178] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0179] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A method for monitoring exercise load in a smart wearable device, characterized in that, The method includes: Acquire basic user data, and collect user respiratory audio data, heart rate physiological data, and movement posture data; Based on the respiratory audio data and the heart rate physiological data, combined with the user's basic data, exercise intensity parameters are generated; Motion feature parameters are generated based on the motion posture data; The user's basic data, the exercise intensity parameters, and the exercise characteristic parameters are input into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

2. The method according to claim 1, characterized in that, The exercise intensity parameters include respiratory intensity parameters. The generation of exercise intensity parameters based on the respiratory audio data and the heart rate physiological data, combined with the user's basic data, includes: Based on the respiratory audio data, the respiratory frequency parameter and the Mel spectrum of the respiratory audio data are extracted, and the Mel spectrum entropy parameter and energy concentration parameter of the Mel spectrum are calculated. The user's respiratory load parameters are calculated based on the respiratory rate parameter, the Mel spectrum entropy parameter, and the energy concentration parameter. The respiratory intensity parameter is constructed based on the respiratory rate parameter, the Mel spectrum entropy parameter, the energy concentration parameter, and the respiratory load parameter; The expressions for the Mel spectrum entropy parameter and the energy concentration parameter are as follows: In the formula, Let S be the Mel spectral entropy parameter at time τ, M be the total number of Mel filters in the Mel filter bank, and S be the... τ (m) represents the energy value of the Mel spectral component of the m-th Mel filter in the audio frame of the breathing audio data at time τ. Let S be the energy concentration parameter at time τ. H This is the set of serial numbers for high-frequency Mel filters.

3. The method according to claim 2, characterized in that, The exercise intensity parameters include heart rate intensity parameters and cardiopulmonary coordination parameters. The generation of exercise intensity parameters based on the respiratory audio data and the heart rate physiological data, combined with the user's basic data, includes: The heart rate reserve percentage parameter of the user is calculated based on the heart rate physiological data, and the heart rate reserve percentage parameter is set as the heart rate intensity parameter. The cardiopulmonary coordination parameters are calculated based on the respiratory load parameters and the percentage heart rate reserve parameters.

4. The method according to claim 3, characterized in that, The expressions for the respiratory load parameters and the cardiopulmonary coordination parameters are as follows: In the formula, Let ρ be the cardiopulmonary coordination parameter at time τ, ρ be the cardiopulmonary coordination index coefficient, and ||·||1 be the first norm. and These are the preset maximum and minimum values ​​of the respiratory load parameters, respectively. Let be the respiratory load parameter at time τ. Let τ be the percentage of heart rate reserve at time τ. and These are the Mel spectrum entropy parameter, the energy concentration parameter, and the breathing frequency parameter at time τ, respectively. The method further includes: If the cardiopulmonary coordination parameters are lower than the preset cardiopulmonary coordination threshold, an abnormal breathing pattern prompt message is generated.

5. The method according to claim 2, characterized in that, The method further includes: Based on the motion type information, obtain the standard posture feature parameters corresponding to the motion type information; Motion guidance information is generated by combining the motion feature parameters and the standard posture feature parameters; Based on the exercise load information, obtain the standard respiratory intensity parameters corresponding to the exercise load information; Breathing guidance information is generated by combining the breathing intensity parameters and the standard breathing parameters.

6. The method according to any one of claims 1 to 5, characterized in that, The exercise load monitoring machine learning model includes an exercise type feature recognition layer, an exercise load feature analysis layer, and a recovery time prediction layer. The process of inputting the user's basic data, the exercise intensity parameters, and the exercise feature parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information includes: The user basic data, the exercise intensity parameters, and the exercise feature parameters are input into the exercise type feature recognition layer in the exercise load monitoring machine learning model to generate the exercise type information; Based on the exercise intensity parameters and exercise type information at the same moment of exercise, and combined with the user basic data, an exercise intensity feature vector is constructed. The exercise intensity feature vector is then input into the exercise load feature analysis layer of the exercise load monitoring machine learning model to generate the exercise load information at the moment of exercise. Based on the exercise load information, an exercise load time series is constructed, and the exercise load time series is input into the recovery time prediction layer of the exercise load monitoring machine learning model to generate the expected recovery time information.

7. The method according to claim 6, characterized in that, The exercise load monitoring machine learning model is a hybrid machine learning model constructed based on an improved random forest machine learning model, an improved backpropagation neural network model, and an improved long short-term memory neural network model. The exercise type feature recognition layer is a machine learning model constructed based on the improved random forest machine learning model. The exercise load feature analysis layer is a machine information model constructed based on the improved backpropagation neural network model. The recovery time prediction layer is a machine learning model constructed based on the improved long short-term memory neural network model. The method further includes: Obtain user basic data training dataset, exercise intensity parameter training dataset, exercise feature parameter training dataset, exercise type label training dataset, metabolic equivalent label training dataset, and recovery time label training dataset; The user basic data training dataset, the exercise intensity parameter training dataset, and the exercise feature parameter training dataset are used as inputs to the exercise type feature recognition layer, and the exercise type label training dataset is used as the output of the exercise type feature recognition layer to train and construct the exercise type feature recognition layer. The exercise intensity parameter training dataset and the exercise type label training dataset are used as inputs to the exercise load feature analysis layer, and the metabolic equivalent label training dataset is used as the output of the exercise load feature analysis layer to train and construct the exercise load feature analysis layer. The recovery time prediction layer is trained and constructed by using the metabolic equivalent label training dataset as the input and the recovery time label training dataset as the output.

8. A motion load monitoring device for a smart wearable device, characterized in that, The device includes: The basic data acquisition module is used to acquire basic user data, including user respiratory audio data, heart rate physiological data, and movement posture data. The exercise intensity analysis module is used to generate exercise intensity parameters based on the respiratory audio data and the heart rate physiological data, combined with the user's basic data; The motion posture analysis module is used to generate motion feature parameters based on the motion posture data. The exercise load monitoring module is used to input the user's basic data, the exercise intensity parameters, and the exercise characteristic parameters into the exercise load monitoring machine learning model to generate exercise type information, exercise load information, and expected recovery time information.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the method of any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images