Neural network-based cardiopulmonary endurance detection method, device and program product
Through neural network technology combined with exercise physiology, the BP neural network model is used to predict maximum oxygen uptake, which solves the problems of poor convenience and low accuracy of existing cardiopulmonary endurance detection methods, and achieves efficient and accurate cardiopulmonary endurance assessment, suitable for large-scale Application and individual health assessment.
Patent Information
- Application Number
- CN202410998490.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-12
AI Technical Summary
The existing cardiopulmonary endurance detection methods have problems such as poor convenience and low accuracy, especially indirect measurement methods rely on maximum load and physiological indicator measurements, and cannot be performed independently, and lack flexibility and accuracy.
Using a neural network-based cardiopulmonary endurance detection method, the peak heart rate and frequency of individual age, height, weight, and squat exercises are obtained, combined with the BP neural network model, the maximum oxygen uptake is predicted, thereby obtaining the cardiopulmonary endurance detection results and avoiding Reliance on expensive equipment improves convenience and accuracy.
It realizes the cardiopulmonary endurance assessment through simple incremental load squat exercise, which reduces costs and improves the accuracy and flexibility of the assessment. It is suitable for large-scale applications and can personalize the maximum oxygen intake and meets individual health needs.
Smart Images

Figure CN118924261A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the field of artificial intelligence technology, and specifically relates to a cardiopulmonary endurance detection method, device and program product based on neural network. Background Art
[0002] At present, there are two methods for measuring cardiorespiratory endurance: direct measurement and indirect measurement. Direct measurement is considered to be a more accurate method, but it requires the use of special testing equipment such as a treadmill or power bike, and requires participants to perform extreme intensity exercise. It has problems such as complex measurement process and cumbersome operation. Therefore, people have developed various indirect measurement methods to simplify the testing process.
[0003] However, most of the existing indirect measurement methods still rely on extreme loads. Although they do not require specific sports equipment, they still need to measure physiological indicators such as blood pressure and heart rate, and usually require the collaboration of multiple people, and cannot be performed independently by a single participant. In addition, these methods have specific requirements for the test site, which makes the test method lack flexibility and convenience. At the same time, at the algorithm level, most indirect measurement techniques use linear regression models to obtain test results (i.e., maximum oxygen uptake), but this method cannot fully fit the nonlinear relationship between input indicators and output indicators, and its accuracy is lacking. Based on this, how to provide a convenient and accurate cardiopulmonary endurance test method has become an urgent problem to be solved. Summary of the invention
[0004] The purpose of the present invention is to provide a cardiopulmonary endurance detection method, device and program product based on neural network to solve the problems of poor convenience and low accuracy in the prior art.
[0005] In order to achieve the above object, the present invention adopts the following technical solutions:
[0006] In a first aspect, a cardiopulmonary endurance detection method based on a neural network is provided, comprising:
[0007] Acquiring physiological index data of a target human body, wherein the physiological index data includes the age, weight, height of the target human body, and a peak heart rate and a total number of squats generated when the target human body performs several rounds of squats, the number of squats in each round of the several rounds of squats increases sequentially, the exercise time of each round of squats is the same, and the difference between the number of squats in two adjacent rounds of squats is a preset value;
[0008] Obtaining a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the maximum oxygen uptake prediction model adopts a neural network model;
[0009] Inputting the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body;
[0010] Based on the predicted value of the maximum oxygen uptake of the target human body, a cardiopulmonary endurance test result of the target human body is obtained.
[0011] Based on the above disclosed content, the present invention pre-trains a maximum oxygen uptake prediction model, wherein the model is trained with sample physiological indicator data of several sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the sample physiological indicator data of any sample human body includes the height, weight, age, and peak heart rate and total number of squats when performing several rounds of incremental squats of any sample human body; therefore, in actual application, it is only necessary to collect the physiological indicator data of the target human body and input the physiological indicator data into the aforementioned maximum oxygen uptake prediction model to obtain the maximum oxygen uptake of the target human body; finally, the cardiopulmonary endurance test result of the target human body can be determined based on the maximum oxygen uptake.
[0012] Through the above design, the present invention obtains key physiological parameters such as the individual's age, height, weight, the number of squat exercises with continuous incremental load, and the peak heart rate recorded during the squat exercise, and combines it with a neural network deep learning algorithm to achieve accurate prediction of the individual's maximum oxygen uptake, so as to accurately derive the individual's cardiopulmonary endurance level based on this; in this way, compared with traditional technologies, the evaluation can be completed only through simple squat exercises with incremental loads, which not only avoids dependence on expensive measuring equipment and reduces costs, but also improves convenience and flexibility of use; at the same time, the use of neural networks for the evaluation of maximum oxygen uptake has a more accurate result; therefore, the present invention combines neural network technology with exercise physiology, which not only improves the accuracy and practicality of maximum oxygen uptake evaluation, but also brings an innovative solution to the field of sports health, and is therefore very suitable for large-scale application and promotion.
[0013] In a possible design, the maximum oxygen uptake prediction model adopts a BP neural network model, wherein the maximum oxygen uptake prediction model is trained using the following method;
[0014] Acquire a training data set, wherein the training data set includes sample physiological indicator data of a number of sample human bodies and label data of each sample human body, wherein the label data of any sample human body is the actual maximum oxygen uptake of any sample human body;
[0015] Performing standardization processing on the training data set to obtain a standardized training data set after the standardization processing;
[0016] Performing initialization assignment processing on the weights and intercepts of each neuron in the BP neural network model, so as to obtain an initial BP neural network model after the initialization assignment processing;
[0017] The standardized sample physiological index data of each sample human body in the standardized training data set is used as input, and the maximum oxygen uptake prediction value of each sample human body is used as output to train the initial BP neural network model. During the training process, the weights of each neuron in the initial BP neural network model are reversely adjusted according to the label value and output value of the input data of the initial BP neural network model. When the iteration stop condition is met, the training of the initial BP neural network model is completed, so that the trained initial BP neural network model is used as the maximum oxygen uptake prediction model.
[0018] In one possible design, the BP neural network model includes an input layer, an output layer and three hidden layers, wherein the input layer includes 5 input neurons, the output layer includes one output neuron, and each neuron in the BP neural network model corresponds to a nonlinear activation function.
[0019] In a possible design, during the training process, the input and output of any neuron in the initial BP neural network model are respectively:
[0020]
[0021] In formula (1), represents the input of any neuron, k represents the number of layers where any neuron is located, represents the weight of the pth first connected neuron in the first connected neuron set to any of the neurons, represents the output of the pth first connected neuron, represents the intercept of any one of the neurons, L is the total number of first connected neurons in the first connected neuron set, wherein each first connected neuron in the first connected neuron set is a neuron connected to any one of the neurons in the k-1th layer of the initial BP neural network model, when k is 1, the k-1th layer is the input layer, when k is the maximum number of layers of the initial BP neural network model, any one of the neurons is an output neuron, and when any one of the neurons is an output neuron, the input of any one of the neurons is equal to the output;
[0022] In formula (2), represents the output of any neuron, and sigmoid() represents the nonlinear activation function corresponding to any neuron.
[0023] In a possible design, the weights of each neuron in the initial BP neural network model are reversely adjusted according to the label value and output value of the input data of the initial BP neural network model, including:
[0024] According to the label value and output value of the input data of the initial BP neural network model, and using the gradient descent method, the weight of each neuron in the initial BP neural network model is reversely adjusted;
[0025] Wherein, for any of the neurons, the following formula (3) is used to obtain the weight correction amount of any of the neurons at the t+1th iteration;
[0026]
[0027] In the above formula (3), represents the weight correction of any neuron at the t+1th iteration, represents the weight correction of any neuron at the tth iteration, η represents the learning factor, α represents the momentum term, and δ represents the weight adjustment coefficient.
[0028] In a possible design, when k is the maximum number of layers of the initial BP neural network model;
[0029]
[0030] In formula (4), d represents the label value of the standardized sample physiological index data input into the initial BP neural network model at the tth iteration, represents the output of any neuron, and y represents the output value of the initial BP neural network model;
[0031] Wherein, when k is less than the maximum number of layers;
[0032]
[0033] In formula (5), E represents the network error of the initial BP neural network model at the tth iteration, represents the input of all neurons in the k+1th layer of the initial BP neural network model at the tth iteration, It represents the weight of the pth second connected neuron in the second connected neuron set to the qth neuron in the k+1th layer of the initial BP neural network model at the tth iteration, each second connected neuron in the second connected neuron set is a neuron in the kth layer of the initial BP neural network model connected to the qth neuron in the k+1th layer, and the ranking of any neuron in the kth layer is also q.
[0034] In a possible design, the training data set is standardized to obtain a standardized training data set after the standardization process, including:
[0035] Obtaining the maximum and minimum values of all sample physiological indicator data in the training data set;
[0036] According to the maximum value and the minimum value, each sample physiological indicator data in the training data set is standardized to obtain the standardized training data set after the standardization process.
[0037] In a second aspect, a cardiopulmonary endurance detection device based on a neural network is provided, comprising:
[0038] an acquisition unit, configured to acquire physiological index data of a target human body, wherein the physiological index data includes the age, weight, height of the target human body, and a peak heart rate and a total number of squats generated when the target human body performs several rounds of squats, the number of squats in each round of the several rounds of squats increases sequentially, the exercise time of each round of squats is the same, and the difference between the number of squats in two adjacent rounds of squats is a preset value;
[0039] A maximum oxygen uptake prediction unit, used to obtain a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the maximum oxygen uptake prediction model adopts a neural network model;
[0040] The maximum oxygen uptake prediction unit is further used to input the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body;
[0041] The detection unit is used to obtain a cardiopulmonary endurance test result of the target human body based on the predicted maximum oxygen uptake value of the target human body.
[0042] In a third aspect, another cardiopulmonary endurance detection device based on a neural network is provided. Taking the device as an electronic device as an example, the device includes a memory, a processor and a transceiver which are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the cardiopulmonary endurance detection method based on a neural network as in the first aspect or any possible design of the first aspect.
[0043] In a fourth aspect, a storage medium is provided, on which instructions are stored, and when the instructions are executed on a computer, the neural network-based cardiopulmonary endurance detection method as described in the first aspect or any possible design of the first aspect is executed.
[0044] In a fifth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, causes the computer to execute the neural network-based cardiopulmonary endurance testing method as described in the first aspect or any possible design of the first aspect.
[0045] Beneficial effects:
[0046] (1) The present invention achieves accurate prediction of the individual's maximum oxygen uptake by acquiring key physiological parameters such as the individual's age, height, weight, the number of squats with increasing loads, and the peak heart rate recorded during the squats, and combines them with a neural network deep learning algorithm, so as to accurately derive the individual's cardiopulmonary endurance level based on this. In this way, compared with traditional technologies, the evaluation can be completed only through simple squats with increasing loads, which not only avoids dependence on expensive measuring equipment and reduces costs, but also improves convenience and flexibility of use. At the same time, the use of neural networks to evaluate the maximum oxygen uptake makes the results more accurate. Therefore, the present invention combines neural network technology with sports physiology, which not only improves the accuracy and practicality of maximum oxygen uptake evaluation, but also brings an innovative solution to the field of sports health, and is therefore very suitable for large-scale application and promotion.
[0047] (2) The present invention can perform personalized maximum oxygen uptake prediction based on individual differences (such as age, weight, etc.), so the results are more in line with individual health and training needs. BRIEF DESCRIPTION OF THE DRAWINGS
[0048] Figure 1 A schematic diagram of the steps of a neural network-based cardiopulmonary endurance testing method provided in an embodiment of the present invention;
[0049] Figure 2 A schematic diagram of the structure of a maximum oxygen uptake prediction model provided by an embodiment of the present invention;
[0050] Figure 3 A schematic diagram of the structure of a cardiopulmonary endurance detection device based on a neural network provided in an embodiment of the present invention;
[0051] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION
[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the present invention will be briefly introduced below in combination with the drawings and the description of the embodiments or the prior art. Obviously, the following description of the structure of the drawings is only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. It should be noted that the description of these embodiments is used to help understand the present invention, but does not constitute a limitation of the present invention.
[0053] It should be understood that although the terms first, second, etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another unit. For example, a first unit can be referred to as a second unit, and similarly, a second unit can be referred to as a first unit without departing from the scope of the exemplary embodiments of the present invention.
[0054] It should be understood that the term "and / or" that may appear in this article is merely a description of the association relationship between associated objects, indicating that three relationships may exist. For example, A and / or B can represent three situations: A exists alone, B exists alone, and A and B exist at the same time. The term " / and" that may appear in this article describes another type of association object relationship, indicating that two relationships may exist. For example, A / and B can represent two situations: A exists alone, and A and B exist alone. In addition, the character " / " that may appear in this article generally indicates that the previous and next associated objects are in an "or" relationship.
[0055] Example:
[0056] See also Figure 1As shown, the cardiopulmonary endurance detection method based on neural network provided in this embodiment predicts the maximum oxygen uptake of the individual through the individual's physiological parameters and the peak heart rate and total number of squats generated by the squat exercise with increasing load, combined with the BP neural network; then, the cardiopulmonary endurance detection result of the individual can be obtained based on the aforementioned maximum oxygen uptake; thus, compared with the traditional technology, the method only needs to complete the evaluation through a simple squat exercise with increasing load, which not only reduces the cost but also improves the convenience and flexibility of use; at the same time, the neural network algorithm is used for individual evaluation, which has higher accuracy; therefore, the method combines the neural network technology with sports physiology, which not only improves the accuracy and practicality of the maximum oxygen uptake evaluation, but also brings an innovative solution to the field of sports health, thus, it is very suitable for large-scale application and promotion; wherein, for example, the method can be but not limited to running on the cardiopulmonary endurance detection terminal side, optionally, the cardiopulmonary endurance detection terminal can be but not limited to a personal computer (PC), a tablet computer or a smart phone, it can be understood that the aforementioned execution subject does not constitute a limitation on the embodiment of the present application, and accordingly, the operation steps of the method can be but not limited to the following steps S1 to S4.
[0057] S1. Acquire physiological index data of the target human body, wherein the physiological index data includes the age, weight, height of the target human body, and the peak heart rate and total number of squats generated when the target human body performs several rounds of squats, the number of squats in each round of the several rounds of squats increases successively, the exercise time of each round of squats is the same, and the difference between the number of squats in two adjacent rounds is a preset value; in this embodiment, the aforementioned physiological index data can be obtained by measurement and transmitted to the cardiopulmonary endurance detection end after measurement; wherein the aforementioned several rounds of squats are performed according to a signal sound, the signal sound is sounded once, the subject (i.e., the target human body) performs a squat, and the frequency of the signal sound increases by level; optionally, the load time of each level of the load used in this embodiment is 30s, the number of squats in the first level is 10 times, and then each level increases by 5 times (i.e., 10 times in the first round, 15 times in the second round, and then 5 times in each round); of course, the specific squat time and number can be specifically set according to actual use, and are not limited to the aforementioned example.
[0058] At the same time, the calculation process of the total number of squats is: the total number of squats is consistent with the number of signal sounds, so the number of signal sounds is the total number of squats; specifically, the number of squats can be calculated based on the squat time, and the calculation formula is:
[0059]
[0060] In this formula, T is the squat time (in seconds), a1 is the number of squats in the first 30 seconds, and D is the number of increments between consecutive levels; in this way, the total number of squats can be calculated using the above formula.
[0061] At the same time, when the target person is doing squatting exercise, the squatting action should meet the sports standards (such as the preparation posture: the body is upright, the feet are shoulder-width apart, and the eyes are looking forward; when squatting, the movement is natural and smooth, the knees are in the same direction as the toes, the knees should not exceed the toes, and should not bend inward, the palms are facing each other, the arms are raised horizontally in front of the body, and the thighs are approximately parallel to the ground at the lowest point, and then stand up to return to the original position, with the waist and back straight throughout the process; and inhale when squatting and exhale when standing up). Further, the stopping standards may include but are not limited to: (1) more than 3 consecutive movements fail to keep up with the signal rhythm; (2) continuous (1) The exercise is performed more than 3 times without the standard; (2) The exercise is unable to continue due to lack of physical strength; (3) The exercise is unable to continue due to lack of physical strength; (4) The symptoms such as dyspnea, nausea, dizziness, tinnitus, chest pain, extreme fatigue, etc. occur; In this case, the exercise can be stopped if any of the above conditions is met; Of course, the above content is the reference standard when performing squat exercises. When performing cardiopulmonary endurance assessment, this embodiment only needs to obtain the physiological indicator data obtained during squat exercises; In addition, the peak heart rate is collected when the squat exercise is stopped, and can be obtained using a heart rate acquisition instrument and transmitted to the cardiopulmonary endurance detection terminal.
[0062] After obtaining the physiological index data of the target human body, the present embodiment pre-trains a maximum oxygen uptake prediction model for the target human body; therefore, the target human body's cardiopulmonary endurance assessment can be completed by only inputting the physiological index data of the target human body into the maximum oxygen uptake prediction model obtained by the aforementioned training; specifically, the process of obtaining the maximum oxygen uptake prediction model can be, but is not limited to, as shown in the following step S2.
[0063] S2. Obtain a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the maximum oxygen uptake prediction model adopts a neural network model; in specific implementation, the maximum oxygen uptake prediction model can be, but is not limited to, a BP neural network model, wherein, see Figure 2As shown, for example, the BP neural network model may include but is not limited to an input layer, an output layer and three hidden layers, wherein the input layer includes 5 input neurons (i.e., corresponding to the height, weight, age, total number of squats and peak heart rate in the physiological indicator data respectively), the output layer includes an output neuron, and each neuron in the BP neural network model corresponds to a nonlinear activation function; optionally, for example, the nonlinear activation function corresponding to each neuron may be but is not limited to a sigmoid function, and the numbers of hidden neurons in the three hidden layers are l, i and m respectively. Of course, the number of hidden neurons can be specifically set according to actual use and is not specifically limited here.
[0064] Furthermore, the following provides one of the training methods for the aforementioned BP neural network model, which may be but is not limited to the following steps S21 to S24.
[0065] S21. Obtain a training data set, wherein the training data set includes sample physiological indicator data of several sample human bodies and label data of each sample human body, wherein the label data of any sample human body is the actual maximum oxygen uptake of any sample human body; in this embodiment, the process of obtaining each sample physiological indicator data is the same as the process of obtaining the physiological indicator data of the target human body, which will not be elaborated here; at the same time, the label data of any sample human body can be obtained in the following way: during the squat exercise, the actual maximum oxygen uptake of any sample human body is measured by wearing a breathing mask and connecting to a gas analyzer; in this way, the aforementioned method can provide an accurate reference standard for this embodiment, ensuring the effectiveness and scientificity of the maximum oxygen uptake prediction method based on incremental load squat exercise.
[0066] After obtaining the training data set, in order to remove the unit restrictions of each data in the training data set, it is also necessary to standardize the training data set so as to uniformly map each training data to between [0,1]. The standardization process may be but is not limited to the following step S22.
[0067] S22. Standardize the training data set to obtain a standardized training data set after the standardization process. In specific implementation, for example, but not limited to, first obtain the maximum and minimum values of all sample physiological indicator data in the training data set (i.e., obtain the maximum height, maximum weight, maximum age, maximum total number of squats, maximum peak heart rate, minimum height, minimum weight, minimum age, minimum total number of squats and minimum peak heart rate in the training data set); then, based on the maximum and minimum values, standardize each sample physiological indicator data in the training data set to obtain the standardized training data set after the standardization process. In this embodiment, the age, weight, height, total number of squats and peak heart rate in each sample physiological indicator data are standardized respectively. Specifically, the age in any sample physiological indicator data is taken as an example to illustrate the specific standardization process, wherein the standardized age can be obtained by using, but not limited to, the following formula (6).
[0068]
[0069] In the above formula (6), Z represents the age in any sample physiological indicator data, X min Indicates the minimum age, X max represents the maximum age, and Z′ represents the standardized age.
[0070] In this way, by using the above formula (6), the standardization of each sample physiological indicator data in the training data set can be completed; then, the model can be initialized so that after the model is initialized, the standardized training data set can be used to perform model training; wherein, the model initialization process is shown in the following step S23.
[0071] S23. Perform initialization assignment processing on the weights and intercepts of each neuron in the BP neural network model to obtain an initial BP neural network model after the initialization assignment processing; in this embodiment, starting from the hidden layer, each neuron in the BP neural network model has a weight and an intercept; at the same time, the weight of any neuron includes the weight of the neuron connected to the any neuron in the previous layer of the network to the any neuron; for example, for a neuron in the second hidden layer (hereinafter referred to as neuron A), there are 4 neurons connected to neuron A in the first hidden layer, then the weight of neuron A is the weight of the 4 neurons connected to it in the first hidden layer to neuron A; of course, the initialization assignment can be randomly assigned.
[0072] After completing the initialization assignment of the BP neural network model, the model can be trained, that is, each sample physiological indicator in the standardized training data set is input into the initialized BP neural network model to obtain the corresponding prediction value; then, the error back propagation algorithm is used to adjust the parameters of the model (that is, the weights of each neuron) until the iteration stop condition is met, and the model training is completed to obtain the maximum oxygen uptake prediction model.
[0073] Optionally, the model training process may be but is not limited to the following step S24.
[0074] S24. The standardized sample physiological index data of each sample human body in the standardized training data set is used as input, and the maximum oxygen uptake prediction value of each sample human body is used as output to train the initial BP neural network model. During the training process, the weights of each neuron in the initial BP neural network model are reversely adjusted according to the label value and output value of the input data of the initial BP neural network model, until the iteration stop condition is met, the training of the initial BP neural network model is completed, and the trained initial BP neural network model is used as the maximum oxygen uptake prediction model.
[0075] In the specific training process, the input and output of any neuron in the initial BP neural network model are respectively:
[0076]
[0077] In the above formula (1), represents the input of any neuron, k represents the number of layers where any neuron is located, represents the weight of the pth first connected neuron in the first connected neuron set to any of the neurons, represents the output of the pth first connected neuron, represents the intercept of any one of the neurons, L is the total number of first connected neurons in the first connected neuron set, wherein each first connected neuron in the first connected neuron set is a neuron connected to any one of the neurons in the k-1th layer of the initial BP neural network model, and when k is 1, the k-1th layer is the input layer (of course, is the input of each output neuron), when k is the maximum number of layers of the initial BP neural network model, any of the neurons is an output neuron, and when any of the neurons is an output neuron, the input of any of the neurons is equal to the output.
[0078] In the above formula (2), represents the output of any neuron, and sigmoid() represents the nonlinear activation function corresponding to any neuron.
[0079] Thus, from the above formulas (1) and (2), it can be known that the input of any of the above neurons is the weight of each neuron connected to the neuron in the previous layer of the layer where the neuron is located multiplied by the corresponding output, plus the intercept; then, the sum is calculated; and the output of any neuron is obtained by substituting its input into the nonlinear activation function.
[0080] At the same time, the network error of the entire model is as follows: Where y represents the model output (i.e., the predicted value of VO2max), and d represents the label value of the model input data (i.e., the actual VO2max).
[0081] Optionally, when adjusting the model parameters, for example, it can be, but is not limited to, based on the label value and output value of the input data of the initial BP neural network model, and using the gradient descent method to reversely adjust the weights of each neuron in the initial BP neural network model; wherein, for any of the neurons, for example, it can be, but is not limited to, using the following formula (3) to calculate the weight correction amount of any of the neurons at the t+1th iteration.
[0082]
[0083] In the above formula (3), represents the weight correction amount of any neuron at the t+1th iteration (essentially, the weight correction amount from the pth neuron to the any neuron among the neurons connected to the any neuron in the k-1th layer), represents the weight correction amount of any neuron at the tth iteration, η represents the learning factor, α represents the momentum term, and δ represents the weight adjustment coefficient; in this embodiment, α is between 0.7 and 0.8.
[0084] Furthermore, the weight adjustment coefficient has different values depending on the layer number of any neuron, specifically:
[0085] When k is the maximum number of layers of the initial BP neural network model, that is, when any neuron is an output neuron, the weight adjustment coefficient can be calculated by, for example but not limited to, using the following formula (4).
[0086]
[0087] In the above formula (4), d represents the label value of the standardized sample physiological index data input into the initial BP neural network model at the tth iteration, represents the output of any neuron, y represents the output value of the initial BP neural network model; wherein sigmoid′ represents the derivative of the function;
[0088] Similarly, when k is less than the maximum number of layers, the weight adjustment coefficient can be calculated by, for example but not limited to, using the following formula (5).
[0089]
[0090] In the above formula (5), E represents the network error of the initial BP neural network model at the tth iteration, represents the input of all neurons in the k+1th layer of the initial BP neural network model at the tth iteration, represents the weight of the pth second connected neuron in the second connected neuron set to the qth neuron in the k+1th layer of the initial BP neural network model at the tth iteration, each second connected neuron in the second connected neuron set is a neuron in the kth layer of the initial BP neural network model connected to the qth neuron in the k+1th layer, and the ranking of any neuron in the kth layer is also q; in this embodiment, assuming that any neuron is the 2nd neuron in the second hidden layer, then the qth neuron is the 2nd neuron in the third hidden layer, and each second connected neuron is a neuron in the second hidden layer connected to the 2nd neuron in the third hidden layer; at the same time, the weight of any neuron is actually the weight of the pth neuron to the any neuron among the neurons connected to the any neuron in the k-1th layer, therefore, It represents the weight from the pth neuron to the qth neuron in the kth layer, among the neurons connected to the qth neuron; of course, when q is any other value, the determination process of each second connected neuron is the same as the above example, and will not be repeated here.
[0091] Therefore, through the above formulas (3)-(5), the weights of each neuron in the model can be adjusted during the training process. In this way, the training can be terminated by continuously iterating until the iteration stop condition is met, so that the trained BP neural network model can be used as the maximum oxygen uptake prediction model; at the same time, the above iteration stop condition can be, but is not limited to, the norm of the gradient vector in the gradient descent process is less than or equal to the first threshold, or the rate of change of the mean square error of the model is less than or equal to the second threshold; in this way, the training can be terminated when any of the above conditions is met.
[0092] Therefore, through the above explanation, the training of the BP neural network model can be completed, thereby obtaining the maximum oxygen uptake prediction model; then, the physiological indicator data of the target human body is input into the prediction model to obtain the maximum oxygen uptake prediction value of the target human body, and the process is shown in the following step S3.
[0093] S3. Input the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body; in this embodiment, after the maximum oxygen uptake prediction value of the target human body is obtained by the maximum oxygen uptake prediction model, the cardiopulmonary endurance test result of the target human body can be determined based on this, and the process is shown in the following step S4.
[0094] S4. Based on the predicted value of the maximum oxygen uptake of the target human body, the cardiopulmonary endurance test result of the target human body is obtained; in specific implementation, the cardiopulmonary endurance test result is actually measured using the maximum oxygen uptake, and therefore, the predicted value of the maximum oxygen uptake of the target human body can be directly used as its corresponding cardiopulmonary endurance test result.
[0095] Thus, the cardiopulmonary endurance detection method based on neural network described in detail in the aforementioned steps S1 to S4 has the following advantages over traditional technologies: (1) Economy: It avoids dependence on expensive measuring equipment and greatly reduces the evaluation cost, making the present invention particularly suitable for resource-limited environments; (2) Convenience: The evaluation can be completed by simply performing a squat exercise with an increasing load, which is easy to popularize and implement, allowing more people to enjoy the health benefits brought by scientific exercise; (3) Efficiency: The BP neural network is used to optimize the prediction model, making the evaluation of maximum oxygen uptake faster and more accurate, and able to instantly reflect the individual's cardiopulmonary function status; (4) Personalization: It can perform personalized maximum oxygen uptake prediction based on individual differences (such as age, weight, etc.), which can better meet individual health and training needs; Therefore, by combining BP neural network technology with exercise physiology, the present invention not only improves the accuracy and practicality of maximum oxygen uptake evaluation, but also brings an innovative solution to the field of sports health, and is expected to be widely used in multiple fields such as individual health assessment and sports training plan formulation.
[0096] like Figure 3 As shown, the second aspect of this embodiment provides a hardware device for implementing the cardiopulmonary endurance detection method based on a neural network described in the first aspect of the embodiment, including:
[0097] An acquisition unit is used to acquire physiological indicator data of a target human body, wherein the physiological indicator data include the age, weight, height of the target human body, and the peak heart rate and total number of squats generated when the target human body performs several rounds of squat exercises, the number of squats in each round of the several rounds of squat exercises increases successively, the exercise time of each round of squat exercises is the same, and the difference between the number of squats in two adjacent rounds of squat exercises is a preset value.
[0098] The maximum oxygen uptake prediction unit is used to obtain a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the maximum oxygen uptake prediction model adopts a neural network model.
[0099] The maximum oxygen uptake prediction unit is further used to input the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body.
[0100] The detection unit is used to obtain a cardiopulmonary endurance test result of the target human body based on the predicted maximum oxygen uptake value of the target human body.
[0101] The working process, working details and technical effects of the device provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.
[0102] like Figure 4 As shown, the third aspect of this embodiment provides another cardiopulmonary endurance detection device based on a neural network. Taking the device as an electronic device as an example, it includes: a memory, a processor and a transceiver that are communicatively connected in sequence, wherein the memory is used to store computer programs, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the cardiopulmonary endurance detection method based on a neural network as described in the first aspect of the embodiment.
[0103] For example, the memory may include, but is not limited to, random access memory (RAM), read only memory (ROM), flash memory, first input first output (FIFO) and / or first in last out (FILO), etc. Specifically, the processor may include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor may be implemented in at least one hardware form of DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). At the same time, the processor may also include a main processor and a coprocessor. The main processor is a processor for processing data in the awake state, also known as a CPU (Central Processing Unit); the coprocessor is a low-power processor for processing data in the standby state.
[0104] In some embodiments, the processor may be integrated with a GPU (Graphics Processing Unit, image processor), and the GPU is responsible for rendering and drawing the content to be displayed on the display screen. For example, the processor may not be limited to a microprocessor of the STM32F105 series, a reduced instruction set computer (RISC) microprocessor, an X86 or other architecture processor, or a processor with an integrated embedded neural network processor (NPU); the transceiver may be, but is not limited to, a wireless fidelity (WIFI) wireless transceiver, a Bluetooth wireless transceiver, a general packet radio service technology (Genera l Packet Rad ioService, GPRS) wireless transceiver, a ZigBee protocol (a low-power LAN protocol based on the IEEE 802.15.4 standard, ZigBee) wireless transceiver, a 3G transceiver, a 4G transceiver and / or a 5G transceiver, etc. In addition, the device may also include, but is not limited to, a power module, a display screen and other necessary components.
[0105] The working process, working details and technical effects of the electronic device provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.
[0106] The fourth aspect of this embodiment provides a storage medium storing instructions for the cardiopulmonary endurance detection method based on a neural network as described in the first aspect of the embodiment, that is, the storage medium stores instructions, and when the instructions are executed on a computer, the cardiopulmonary endurance detection method based on a neural network as described in the first aspect of the embodiment is executed.
[0107] The storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, a CD, a hard disk, a flash memory, a USB flash drive and / or a memory stick, etc. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.
[0108] The working process, working details and technical effects of the storage medium provided in this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.
[0109] The fifth aspect of this embodiment provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the neural network-based cardiopulmonary endurance testing method as described in the first aspect of the embodiment, wherein the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.
[0110] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.
Claims
1. A cardiopulmonary endurance detection method based on neural network, characterized in that: include: Acquiring physiological index data of a target human body, wherein the physiological index data includes the age, weight, height of the target human body, and a peak heart rate and a total number of squats generated when the target human body performs several rounds of squats, the number of squats in each round of the several rounds of squats increases sequentially, the exercise time of each round of squats is the same, and the difference between the number of squats in two adjacent rounds of squats is a preset value; Obtaining a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output, and the maximum oxygen uptake prediction model adopts a neural network model; Inputting the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body; Based on the predicted value of the maximum oxygen uptake of the target human body, a cardiopulmonary endurance test result of the target human body is obtained.
2. The method according to claim 1, characterized in that: The maximum oxygen uptake prediction model adopts a BP neural network model, wherein the maximum oxygen uptake prediction model is trained by the following method; Acquire a training data set, wherein the training data set includes sample physiological indicator data of a number of sample human bodies and label data of each sample human body, wherein the label data of any sample human body is the actual maximum oxygen uptake of any sample human body; Performing standardization processing on the training data set to obtain a standardized training data set after the standardization processing; Performing initialization assignment processing on the weights and intercepts of each neuron in the BP neural network model, so as to obtain an initial BP neural network model after the initialization assignment processing; The standardized sample physiological index data of each sample human body in the standardized training data set is used as input, and the maximum oxygen uptake prediction value of each sample human body is used as output to train the initial BP neural network model. During the training process, the weights of each neuron in the initial BP neural network model are reversely adjusted according to the label value and output value of the input data of the initial BP neural network model. When the iteration stop condition is met, the training of the initial BP neural network model is completed, so that the trained initial BP neural network model is used as the maximum oxygen uptake prediction model.
3. The method according to claim 2, characterized in that The BP neural network model includes an input layer, an output layer and three hidden layers, wherein the input layer includes 5 input neurons, the output layer includes one output neuron, and each neuron in the BP neural network model corresponds to a nonlinear activation function.
4. The method according to claim 2, characterized in that: During the training process, the input and output of any neuron in the initial BP neural network model are: In formula (1), represents the input of any neuron, k represents the number of layers where any neuron is located, represents the weight of the pth first connected neuron in the first connected neuron set to any of the neurons, represents the output of the pth first connected neuron, represents the intercept of any one of the neurons, L is the total number of first connected neurons in the first connected neuron set, wherein each first connected neuron in the first connected neuron set is a neuron connected to any one of the neurons in the k-1th layer of the initial BP neural network model, when k is 1, the k-1th layer is the input layer, when k is the maximum number of layers of the initial BP neural network model, any one of the neurons is an output neuron, and when any one of the neurons is an output neuron, the input of any one of the neurons is equal to the output; In formula (2), represents the output of any neuron, and sigmoid() represents the nonlinear activation function corresponding to any neuron.
5. The method according to claim 4, characterized in that According to the label value and output value of the input data of the initial BP neural network model, the weights of each neuron in the initial BP neural network model are reversely adjusted, including: According to the label value and output value of the input data of the initial BP neural network model, and using the gradient descent method, the weight of each neuron in the initial BP neural network model is reversely adjusted; Wherein, for any of the neurons, the following formula (3) is used to obtain the weight correction amount of any of the neurons at the t+1th iteration; In the above formula (3), represents the weight correction of any neuron at the t+1th iteration, represents the weight correction of any neuron at the tth iteration, η represents the learning factor, α represents the momentum term, and δ represents the weight adjustment coefficient.
6. The method according to claim 5, characterized in that When k is the maximum number of layers of the initial BP neural network model; In formula (4), d represents the label value of the standardized sample physiological index data input into the initial BP neural network model at the tth iteration, represents the output of any neuron, and y represents the output value of the initial BP neural network model; Wherein, when k is less than the maximum number of layers; In formula (5), E represents the network error of the initial BP neural network model at the tth iteration, represents the input of all neurons in the k+1th layer of the initial BP neural network model at the tth iteration, It represents the weight of the pth second connected neuron in the second connected neuron set to the qth neuron in the k+1th layer of the initial BP neural network model at the tth iteration, each second connected neuron in the second connected neuron set is a neuron in the kth layer of the initial BP neural network model connected to the qth neuron in the k+1th layer, and the ranking of any neuron in the kth layer is also q.
7. The method according to claim 2, characterized in that The training data set is subjected to standardization processing to obtain a standardized training data set after the standardization processing, including: Obtaining the maximum and minimum values of all sample physiological indicator data in the training data set; According to the maximum value and the minimum value, each sample physiological indicator data in the training data set is standardized to obtain the standardized training data set after the standardization process.
8. A cardiopulmonary endurance detection device based on a neural network, characterized in that: include: an acquisition unit, configured to acquire physiological index data of a target human body, wherein the physiological index data includes the age, weight, height of the target human body, and a peak heart rate and a total number of squats generated when the target human body performs several rounds of squats, the number of squats in each round of the several rounds of squats increases sequentially, the exercise time of each round of squats is the same, and the difference between the number of squats in two adjacent rounds of squats is a preset value; A maximum oxygen uptake prediction unit, used to obtain a maximum oxygen uptake prediction model, wherein the maximum oxygen uptake prediction model is trained by taking sample physiological index data of a number of sample human bodies as input and the maximum oxygen uptake prediction value of each sample human body as output; The maximum oxygen uptake prediction unit is further used to input the physiological index data of the target human body into the maximum oxygen uptake prediction model to obtain the maximum oxygen uptake prediction value of the target human body; The detection unit is used to obtain a cardiopulmonary endurance test result of the target human body based on the predicted maximum oxygen uptake value of the target human body.
9. A cardiopulmonary endurance detection device based on a neural network, characterized in that: include: A memory, a processor and a transceiver which are sequentially connected in communication, wherein the memory is used to store a computer program, the transceiver is used to send and receive messages, and the processor is used to read the computer program to execute the cardiopulmonary endurance detection method based on a neural network as described in any one of claims 1 to 7.
10. A computer program product comprising instructions, characterized in that When the instructions are executed on a computer, the computer is caused to execute the cardiopulmonary endurance detection method based on a neural network as described in any one of claims 1 to 7.
Citation Information
Cited By
Method for realizing exercise health management based on individualized cardiopulmonary function digital twinborn model
CN120164618A
