Deep learning-based respiratory training method
Through deep learning-based breathing training methods, the running stage is predicted and the recommended breathing frequency distribution is generated, which solves the problem of poor breathing during running, improves running efficiency and safety, and is especially suitable for novices in running.
Patent Information
- Application Number
- CN202411477443.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2025-02-14
AI Technical Summary
During running, it is difficult for many people, especially novices, to maintain a scientific and reasonable breathing rhythm, resulting in shortness of breath and decreased physical strength, affecting running efficiency and may lead to respiratory irritation and damage.
Using deep learning-based breathing training method, by obtaining heart rate, blood oxygen content and breathing frequency data during running, predict the running stage and generate the recommended breathing frequency distribution, and guide users to adjust the breathing frequency in real time.
Effectively guide users to maintain a reasonable breathing pattern during running, reduce early fatigue caused by poor breathing or rhythm disorder, improve running efficiency, and reduce sports injuries, especially suitable for novices in running.
Smart Images
Figure CN119425024A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of deep learning, and in particular relates to a breathing training method based on deep learning. Background Art
[0002] In sports, running, as a widely participated aerobic exercise, can not only enhance cardiopulmonary function, but also improve physical endurance and overall health level.
[0003] Running is not just a simple exercise of the leg muscles, it also requires the coordination of breathing rhythm. The correct breathing rhythm can not only improve running efficiency, but also effectively reduce sports injuries. Therefore, it is necessary to maintain a reasonable breathing method during running. However, in actual situations, some runners, especially some novice runners, often cannot maintain a scientific and reasonable breathing rhythm during running, resulting in shortness of breath and rapid decline in physical strength in a short period of time, affecting running efficiency, and may even cause irritation and damage to the respiratory tract due to hyperventilation.
[0004] Therefore, how to provide an effective solution to improve running efficiency and reduce sports injuries has become a difficult problem to be solved in the prior art. Summary of the invention
[0005] The purpose of the present invention is to provide a breathing training method based on deep learning to solve the above-mentioned problems existing in the prior art.
[0006] In order to achieve the above object, the present invention adopts the following technical solutions:
[0007] In a first aspect, the present invention provides a deep learning-based breathing training method, which is applied to a smart wearable device, comprising:
[0008] Obtain the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running;
[0009] The heart rate time series data, the blood oxygen content time series data, and the respiratory frequency time series data are used as inputs of a pre-trained running phase prediction model based on deep learning to perform operations to obtain a time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase, and a fatigue running phase;
[0010] Based on the time series distribution of the multiple running stages and the recommended breathing strategies for each running stage, generating a recommended breathing frequency distribution for the user, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running;
[0011] The breathing recommendation strategy indicates that breathing should be performed at different breathing frequencies during different running phases.
[0012] Based on the above disclosed content, the present invention obtains the heart rate timing data, blood oxygen content timing data and respiratory rate timing data within a specified time after the user starts running, and uses the heart rate timing data, blood oxygen content timing data and respiratory rate timing data as inputs of a pre-trained and deep learning-based running phase prediction model to perform calculations, thereby obtaining the timing distribution of multiple running phases of the user, the multiple running phases including a running start phase, a stable running phase and a fatigue running phase, and then based on the timing distribution of multiple running phases and the recommended breathing strategies for each running phase, generates a recommended breathing frequency distribution of the user to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running. In this way, during the running process, according to the previous changes in the individual's heart rate, blood oxygen content and breathing frequency, the time series distribution of multiple running stages can be predicted, and according to the recommended breathing strategy for each running stage, a recommended breathing frequency distribution can be generated to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during the running process, thereby guiding the user to maintain a reasonable breathing method during the running process, so that the breathing during the exercise is coordinated with the muscle movement, and the premature fatigue caused by poor breathing or rhythm disorder is reduced, the running efficiency is improved, and at the same time, excessive ventilation can be avoided to reduce the stimulation and damage to the respiratory tract and reduce sports injuries. It is particularly suitable for the breathing guidance of novice runners during the running process.
[0013] In one possible design, the method further includes:
[0014] Based on the recommended respiratory frequency distribution, determining a recommended respiratory frequency at a current time point;
[0015] Breathing guidance prompt information is generated based on the recommended breathing frequency at the current time point to guide the user to breathe according to the breathing frequency indicated by the breathing guidance prompt information.
[0016] In a possible design, the breathing guidance prompt information is voice information or vibration information.
[0017] In one possible design, the method further includes:
[0018] Establishing a training sample data set, wherein the sample data in the training sample data set includes heart rate time series data, blood oxygen content time series data, and breathing rate time series data within a specified time after the sample user starts running, as well as time series distribution of multiple running stages of the sample user;
[0019] The heart rate time series data, blood oxygen content time series data and respiratory rate time series data in the sample data are used as sample inputs of the running phase prediction model, and the time series distribution of multiple running phases of the sample user in the sample data is used as sample outputs of the running phase prediction model for training to obtain a trained running phase prediction model;
[0020] The running stage prediction model includes an input layer, multiple hidden layers and an output layer, and the node parameters of each hidden layer are H l =tanh(W l H l-1 +B l ), where H l represents the node parameters of the lth hidden layer, W l represents the weight matrix of the lth hidden layer, H l-1 represents the node parameters of the l-1th hidden layer, B l represents the bias vector of the lth hidden layer. When l = 1, H l-1 Represents the sample input during model training, and tanh represents the hyperbolic tangent function.
[0021] In one possible design, before the heart rate time series data, the blood oxygen content time series data, and the respiratory rate time series data are used as inputs of a pre-trained deep learning-based running phase prediction model for operation, the method further includes:
[0022] Standardizing the heart rate time series data, the blood oxygen content time series data, and the respiratory rate time series data;
[0023] The step of using the heart rate time series data, the blood oxygen content time series data and the respiratory frequency time series data as inputs of a pre-trained deep learning-based running phase prediction model for operation includes:
[0024] The standardized heart rate time series data, the standardized blood oxygen content time series data and the standardized respiratory frequency time series data are used as inputs of a pre-trained deep learning-based running phase prediction model for calculation.
[0025] In one possible design, the normalization formula is in Represents the standardized data of the d-th dimension of the m time series data of the i-th data, represents the d-th dimension of m time series data of the i-th data, E(F i d ) represents the mean of the d-dimensional data of the i-th data, var(F i d) represents the standard deviation of the d-dimensional data of the i-th data.
[0026] In a possible design, generating the recommended breathing frequency distribution of the user based on the time series distribution of the multiple running stages and the breathing recommendation strategy for each running stage includes:
[0027] Based on the time series distribution of the multiple running stages and the recommended breathing frequencies indicated by the recommended breathing strategies for each running stage, a recommended breathing frequency curve of the user is fitted through linear fitting.
[0028] In a second aspect, the present invention provides a deep learning-based breathing training device, which is applied to a smart wearable device, comprising:
[0029] An acquisition unit, used to acquire the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running;
[0030] a computing unit, configured to use the heart rate time series data, the blood oxygen content time series data, and the respiratory frequency time series data as inputs of a pre-trained running phase prediction model based on deep learning to perform computing to obtain a time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase, and a fatigue running phase;
[0031] a generating unit, configured to generate a recommended breathing frequency distribution of the user based on the time series distribution of the plurality of running stages and the breathing recommendation strategy of each running stage, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running;
[0032] The breathing recommendation strategy indicates that breathing should be performed at different breathing frequencies during different running phases.
[0033] In a third aspect, the present invention provides an intelligent wearable device, comprising a memory, a processor and a transceiver that are communicatively connected in sequence, 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 deep learning-based breathing training method as described in the first aspect or any possible design of the first aspect.
[0034] In a fourth aspect, the present invention provides a computer-readable storage medium having instructions stored thereon, which, when executed on a computer, executes the deep learning-based breathing training method described in the first aspect or any possible design of the first aspect.
[0035] In a fifth aspect, the present invention provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the deep learning-based breathing training method as described in the first aspect or any possible design of the first aspect.
[0036] Beneficial effects:
[0037] The deep learning-based breathing training method provided by the present invention can predict the time series distribution of multiple running stages according to the changes in the individual's heart rate, blood oxygen content and breathing frequency in the early stage during the running process, and generate a recommended breathing frequency distribution according to the breathing recommendation strategy of each running stage to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during the running process, thereby guiding the user to maintain a reasonable breathing method during the running process, so that the breathing during the exercise is coordinated with the muscle movement, and the premature fatigue caused by poor breathing or rhythm disorder is reduced, the running efficiency is improved, and at the same time, excessive ventilation can be avoided to reduce the stimulation and damage to the respiratory tract and reduce sports injuries. It is particularly suitable for the breathing guidance of novice runners during the running process, and is convenient for practical application and promotion. BRIEF DESCRIPTION OF THE DRAWINGS
[0038] Figure 1 A flowchart of a deep learning-based breathing training method provided in an embodiment of the present application;
[0039] Figure 2 A schematic diagram of the structure of the running phase prediction model provided in the embodiment of the present application;
[0040] Figure 3 A schematic block diagram of a deep learning-based breathing training device provided in an embodiment of the present application;
[0041] Figure 4 A schematic block diagram of a smart wearable device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0042] 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.
[0043] 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.
[0044] 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.
[0045] In order to improve running efficiency and reduce sports injuries, an embodiment of the present application provides a breathing training method based on deep learning. The breathing training method based on deep learning can guide users to maintain a reasonable breathing method during running, improve running efficiency and reduce sports injuries.
[0046] The deep learning-based breathing training method provided in the embodiment of the present application can be applied to smart wearable devices, such as smart bracelets, smart watches, etc. It can be understood that the execution subject does not constitute a limitation on the embodiment of the present application.
[0047] The deep learning-based breathing training method provided in the embodiments of the present application will be described in detail below.
[0048] like Figure 1 As shown, it is a flowchart of the deep learning-based breathing training method provided in the first aspect of the embodiment of the present application. The deep learning-based breathing training method may include but is not limited to the following steps S101-S105.
[0049] Step S101. Obtain the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running.
[0050] In the embodiment of the present application, the user can wear a smart wearable device, and can directly detect the user's heart rate data and respiratory rate data in real time through the smart wearable device (such as a smart bracelet, smart watch), or detect the user's heart rate data and respiratory rate data in real time through some wearable devices specifically used to detect respiratory rate and heart rate. At the same time, the user's blood oxygen content data can be detected through devices such as blood oxygen clips, photoelectric sensor cuffs (devices that measure blood oxygen saturation by detecting changes in light after passing through fingers through photoelectric sensors).
[0051] When the user is running, the smart wearable device can obtain the user's heart rate data and blood oxygen content data in real time, as well as the blood oxygen content detected by the blood oxygen content detection device, and obtain the heart rate time series data, blood oxygen content time series data and respiratory rate time series data within a specified time after the user starts running. The specified time can be set according to actual conditions, for example, it can be 1 minute, 2 minutes or 3 minutes.
[0052] Step S102: The heart rate time series data, the blood oxygen content time series data and the respiratory rate time series data are used as inputs of a pre-trained running phase prediction model based on deep learning to perform operations to obtain the time series distribution of multiple running phases of the user.
[0053] The multiple running stages include a running start stage, a stable running stage and a fatigue running stage.
[0054] In the embodiment of the present application, a running phase prediction model for running phase prediction can be pre-trained. When training the model, a training sample data set can be first established, and the training sample data set includes a large amount of sample data. The sample data in the training sample data set includes heart rate time series data, blood oxygen content time series data, and respiratory frequency time series data within a specified time after the sample user starts running, as well as the time series distribution of multiple running phases of the sample user. Then, the heart rate time series data, blood oxygen content time series data, and respiratory frequency time series data in the sample data are used as sample inputs of the running phase prediction model, and the time series distribution of multiple running phases of the sample user in the sample data is used as the sample output of the running phase prediction model for training, so as to obtain a trained running phase prediction model.
[0055] See also Figure 2 The running stage prediction model includes an input layer, multiple hidden layers and an output layer. The node parameters of each hidden layer can be expressed as H l =tanh(W l H l-1 +B l ), where H l represents the node parameters of the lth hidden layer, W l represents the weight matrix of the lth hidden layer, H l-1 represents the node parameters of the l-1th hidden layer, B l represents the bias vector of the lth hidden layer. When l = 1, H l-1 Represents the sample input during model training, and tanh represents the hyperbolic tangent function.
[0056] When training the prediction model for the running phase, the heart rate time series data, blood oxygen content time series data, and respiratory rate time series data in the sample data can be standardized first so that the data value range is [-1,1].
[0057] The standardization formula can be expressed as in Represents the standardized data of the d-th dimension of the m time series data of the i-th data, represents the d-th dimension of m time series data of the i-th data, E(F i d ) represents the mean of the d-dimensional data of the i-th data, var(F i d ) represents the standard deviation of the d-dimensional data of the i-th data.
[0058] It can be understood that if the heart rate timing data, blood oxygen content timing data and respiratory rate timing data in the sample data are standardized when training the running phase prediction model, then the heart rate timing data, blood oxygen content timing data and respiratory rate timing data within a specified time after the user starts running are also required to be standardized before being used as the input of the pre-trained and deep learning-based running phase prediction model for calculation, and then the standardized heart rate timing data, the standardized blood oxygen content timing data and the standardized respiratory rate timing data are used as the input of the pre-trained and deep learning-based running phase prediction model for calculation to obtain the time series distribution of multiple running phases of the user, that is, the time distribution of the running starting phase, the time distribution of the stable running phase and the time distribution of the fatigue running phase.
[0059] Step S103: Generate a recommended breathing frequency distribution of the user based on the time series distribution of multiple running stages and the breathing recommendation strategy for each running stage.
[0060] The breathing recommendation strategy indicates that the user should breathe at different breathing frequencies in different running stages, so as to guide the user to breathe at the breathing frequency indicated by the recommended breathing frequency distribution during running.
[0061] In an embodiment of the present application, when generating the recommended breathing frequency distribution of the user, the recommended breathing frequency curve of the user can be fitted by linear fitting based on the time series distribution of multiple running stages and the recommended breathing frequency indicated by the breathing recommendation strategy for each running stage. The recommended breathing frequency curve indicates that the user breathes at different breathing frequencies at different time periods.
[0062] Step S104: Determine the recommended breathing frequency at the current time point based on the recommended breathing frequency distribution.
[0063] Step S105: Generate breathing guidance prompt information based on the recommended breathing frequency at the current time point to guide the user to breathe according to the breathing frequency indicated by the breathing guidance prompt information.
[0064] The attraction guidance prompt information is voice information or vibration information. For example, the user can be prompted to breathe according to the breathing frequency indicated by the voice content through a prompt voice, or the smart wearable device can be controlled to vibrate to prompt the user to breathe according to the vibration frequency of the smart wearable device.
[0065] In summary, the breathing training method based on deep learning provided by the present invention obtains the heart rate time series data, blood oxygen content time series data and breathing frequency time series data within a specified time after the user starts running, and uses the heart rate time series data, blood oxygen content time series data and breathing frequency time series data as inputs of a pre-trained and deep learning-based running phase prediction model to perform calculations, thereby obtaining the time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase and a fatigue running phase, and then based on the time series distribution of multiple running phases and the breathing recommendation strategies for each running phase, generates a recommended breathing frequency distribution of the user, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running. In this way, during the running process, according to the previous changes in the individual's heart rate, blood oxygen content and breathing frequency, the time series distribution of multiple running stages can be predicted, and according to the recommended breathing strategy for each running stage, a recommended breathing frequency distribution can be generated to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during the running process, thereby guiding the user to maintain a reasonable breathing method during the running process, so that the breathing during the exercise is coordinated with the muscle movement, and the premature fatigue caused by poor breathing or rhythm disorder is reduced, the running efficiency is improved, and at the same time, excessive ventilation can be avoided to reduce the stimulation and damage to the respiratory tract and reduce sports injuries. It is particularly suitable for the breathing guidance of novice runners during the running process, which is convenient for practical application and promotion.
[0066] See also Figure 3 In a second aspect, an embodiment of the present application provides a deep learning-based breathing training device, which is applied to a smart wearable device. The deep learning-based breathing training device includes:
[0067] An acquisition unit, used to acquire the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running;
[0068] a computing unit, configured to use the heart rate time series data, the blood oxygen content time series data, and the respiratory frequency time series data as inputs of a pre-trained running phase prediction model based on deep learning to perform computing to obtain a time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase, and a fatigue running phase;
[0069] a generating unit, configured to generate a recommended breathing frequency distribution of the user based on the time series distribution of the plurality of running stages and the breathing recommendation strategy of each running stage, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running;
[0070] The breathing recommendation strategy indicates that breathing should be performed at different breathing frequencies during different running phases.
[0071] The working process, working details and technical effects of the deep learning-based breathing training device provided in the second aspect of this embodiment can be found in the first aspect of the embodiment and will not be described in detail here.
[0072] like Figure 4 As shown, the third aspect of an embodiment of the present application provides a smart wearable device, comprising a memory, a processor and a transceiver that are communicatively connected in sequence, 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 deep learning-based breathing training method as described in the first aspect of the embodiment.
[0073] For specific examples, the memory may include but is not limited to random access memory (RAM), read-only memory (ROM), flash memory, first-in-first-out memory (FIFO) and / or first-in-last-out memory (FILO), etc.; the processor may be but is not limited to a microprocessor of the STM32F105 series, an ARM (Advanced RISC-Machines), an X86 or other architecture processor, or a processor with an integrated NPU (neural-network processing units); the transceiver may be but is not limited to a WiFi (Wireless Fidelity) wireless transceiver, a Bluetooth wireless transceiver, a General Packet Radio Service (GPRS) wireless transceiver, a ZigBee protocol (a low-power local area network protocol based on the IEEE802.15.4 standard, ZigBee) wireless transceiver, a 3G transceiver, a 4G transceiver and / or a 5G transceiver, etc.
[0074] In a fourth aspect of this embodiment, a computer-readable storage medium is provided that stores instructions for the deep learning-based breathing training method described in the first aspect of the embodiment, that is, the computer-readable storage medium stores instructions, and when the instructions are executed on a computer, the deep learning-based breathing training method described in the first aspect is executed. The computer-readable storage medium refers to a carrier for storing data, which may include but is not limited to a floppy disk, an optical disk, a hard disk, a flash memory, a USB flash drive, and / or a memory stick, and the computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices.
[0075] A fifth aspect of this embodiment provides a computer program product comprising instructions, which, when executed on a computer, causes the computer to execute the deep learning-based breathing training 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 device.
[0076] It should be understood that certain details are provided in the following description to facilitate a complete understanding of the example embodiments. However, it should be understood by those of ordinary skill in the art that the example embodiments can be implemented without these certain details. For example, the system can be shown in a block diagram to avoid obscuring the example with unnecessary details. In other examples, well-known processes, structures, and techniques may not be shown in unnecessary detail to avoid obscuring the example embodiments.
[0077] 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 breathing training method based on deep learning, applied to smart wearable devices, characterized in that: include: Obtain the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running; The heart rate time series data, the blood oxygen content time series data, and the respiratory frequency time series data are used as inputs of a pre-trained running phase prediction model based on deep learning to perform operations to obtain a time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase, and a fatigue running phase; Based on the time series distribution of the multiple running stages and the recommended breathing strategies for each running stage, generating a recommended breathing frequency distribution for the user, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running; The breathing recommendation strategy indicates that breathing should be performed at different breathing frequencies during different running phases.
2. The deep learning-based breathing training method according to claim 1, characterized in that: The method further comprises: Based on the recommended respiratory frequency distribution, determining a recommended respiratory frequency at a current time point; Breathing guidance prompt information is generated based on the recommended breathing frequency at the current time point to guide the user to breathe according to the breathing frequency indicated by the breathing guidance prompt information.
3. The deep learning-based breathing training method according to claim 2, characterized in that: The breathing guidance prompt information is voice information or vibration information.
4. The deep learning-based breathing training method according to claim 1, characterized in that: The method further comprises: Establishing a training sample data set, wherein the sample data in the training sample data set includes heart rate time series data, blood oxygen content time series data, and breathing rate time series data within a specified time after the sample user starts running, as well as time series distribution of multiple running stages of the sample user; The heart rate time series data, blood oxygen content time series data and respiratory rate time series data in the sample data are used as sample inputs of the running phase prediction model, and the time series distribution of multiple running phases of the sample user in the sample data is used as sample outputs of the running phase prediction model for training to obtain a trained running phase prediction model; The running stage prediction model includes an input layer, multiple hidden layers and an output layer, and the node parameters of each hidden layer are H l =tanh(W l H l-1 +B l ), where H l represents the node parameters of the lth hidden layer, W l represents the weight matrix of the lth hidden layer, H l-1 represents the node parameters of the l-1th hidden layer, B l represents the bias vector of the lth hidden layer. When l = 1, H l-1 Represents the sample input during model training, and tanh represents the hyperbolic tangent function.
5. The deep learning-based breathing training method according to claim 1, characterized in that: Before computing the heart rate time series data, the blood oxygen content time series data, and the respiratory rate time series data as inputs of a pre-trained deep learning-based running phase prediction model, the method further includes: Standardizing the heart rate time series data, the blood oxygen content time series data, and the respiratory rate time series data; The step of using the heart rate time series data, the blood oxygen content time series data and the respiratory frequency time series data as inputs of a pre-trained deep learning-based running phase prediction model for operation includes: The standardized heart rate time series data, the standardized blood oxygen content time series data and the standardized respiratory frequency time series data are used as inputs of a pre-trained deep learning-based running phase prediction model for calculation.
6. The deep learning-based breathing training method according to claim 5, characterized in that: The standardization formula is: in Represents the standardized data of the d-th dimension of the m time series data of the i-th data, represents the d-th dimension of m time series data of the i-th data, E(F i d ) represents the mean of the d-dimensional data of the i-th data, var(F i d ) represents the standard deviation of the d-dimensional data of the i-th data.
7. The deep learning-based breathing training method according to claim 1, characterized in that: The generating the recommended breathing frequency distribution of the user based on the time series distribution of the multiple running stages and the breathing recommendation strategy of each running stage includes: Based on the time series distribution of the multiple running stages and the recommended breathing frequencies indicated by the recommended breathing strategies for each running stage, a recommended breathing frequency curve of the user is fitted through linear fitting.
8. A deep learning-based breathing training device, applied to smart wearable devices, characterized in that: include: An acquisition unit, used to acquire the heart rate time series data, blood oxygen content time series data and breathing rate time series data within a specified time after the user starts running; a computing unit, configured to use the heart rate time series data, the blood oxygen content time series data, and the respiratory frequency time series data as inputs of a pre-trained running phase prediction model based on deep learning to perform computing to obtain a time series distribution of multiple running phases of the user, wherein the multiple running phases include a running start phase, a stable running phase, and a fatigue running phase; a generating unit, configured to generate a recommended breathing frequency distribution of the user based on the time series distribution of the plurality of running stages and the breathing recommendation strategy of each running stage, so as to guide the user to breathe according to the breathing frequency indicated by the recommended breathing frequency distribution during running; The breathing recommendation strategy indicates that breathing should be performed at different breathing frequencies during different running phases.
9. A smart wearable device, characterized in that: It comprises a memory, a processor and a transceiver which are communicatively connected in sequence, 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 deep learning-based breathing training method as described in any one of claims 1 to 7.
10. A computer program product comprising a computer program or instructions, characterized in that When the computer program or the instruction is executed by a computer, the deep learning-based breathing training method according to any one of claims 1 to 7 is implemented.