Respiration rate detection method and system, storage medium and wearable intelligent equipment
By obtaining the respiratory and pulse wave data of multiple users under different motion states, training the neural network model, solving the problem of high power consumption in the existing technology, achieving higher accuracy breathing rate detection and resource conservation.
Patent Information
- Application Number
- CN202510323428.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-19
- Publication Date
- 2025-07-04
AI Technical Summary
The existing respiratory rate detection methods consume high power in the process of achieving respiratory rate monitoring and insufficient hardware storage resources.
By obtaining multiple sets of breathing waves and pulse wave data of multiple users under different motion states, extracting the characteristic parameters of the pulse wave, training the neural network model, and determining the real-time breathing rate using the neural network model parameter vector and real-time pulse wave data.
It reduces power consumption during breathing rate detection, improves detection accuracy, and saves hardware storage resources.
Smart Images
Figure CN120241033A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of respiratory rate detection, and further relates to a respiratory rate detection method, system, storage medium, and wearable intelligent device. Background Art
[0002] In existing respiratory rate detection technologies, common methods include radar detection, image recognition, and analysis based on electrocardiogram (ECG) or respiratory signals, etc. Radar detection uses the principle of microwave reflection to capture the minute movements of the chest during breathing. Image recognition obtains the respiratory rate by analyzing the breathing movements of the human body in video images. Analysis based on electrocardiogram or respiratory signals uses biomedical signal processing technology to extract respiratory-related information from physiological signals. These methods have been widely used in multiple scenarios such as clinical monitoring, home health management, and exercise physiology research, providing diverse solutions for respiratory health management. However, in the process of implementing respiratory rate monitoring, these traditional methods usually have the problem of high power consumption. Summary of the Invention
[0003] To solve the above technical problems, the present application provides a respiratory rate detection method, system, storage medium, and wearable intelligent device, which reduce the power consumption during the respiratory rate detection process.
[0004] In a first aspect, the present application provides a respiratory rate detection method, including: obtaining multiple groups of training data of multiple users in different exercise states, where the training data includes training respiratory wave data and training pulse wave data within the same preset time window width; processing each group of the training pulse wave data to obtain an input vector corresponding to each group of the training pulse wave data, where the input vector includes an average peak interval, an average trough interval, an average peak amplitude, a number of peaks, and a number of troughs; processing each group of the training respiratory wave data to obtain a training respiratory rate corresponding to each group of the training respiratory wave data; training a neural network model based on multiple groups of the input vectors and multiple groups of the training respiratory rates to obtain a neural network model parameter vector; and determining the real-time respiratory rate of a user to be measured based on the real-time pulse wave data of the user to be measured within the preset time window width in the current exercise state and the neural network model parameter vector.
[0005] The above respiratory rate detection method obtains multiple sets of respiratory wave and pulse wave data of multiple users in different exercise states, and extracts characteristic parameters such as the average peak interval, average trough interval, average peak amplitude, number of peaks, and number of troughs of the pulse wave, as well as the respiratory rate corresponding to the respiratory wave, for training a neural network model. Then, based on the neural network model parameter vector and the real-time pulse wave data of the user to be measured, the real-time respiratory rate of the user to be measured is determined. This method reduces the power consumption during the respiratory rate detection process, improves the detection accuracy of the respiratory rate, and saves hardware storage resources.
[0006] In one implementation, processing each group of the to-be-trained pulse wave data to obtain an input vector corresponding to each group of the to-be-trained pulse wave data specifically includes: filtering each group of the to-be-trained pulse wave data; performing peak and trough detection on the filtered each group of the to-be-trained pulse wave data to obtain the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the to-be-trained pulse wave data; and constructing the input vector corresponding to each group of the to-be-trained pulse wave data based on the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the to-be-trained pulse wave data.
[0007] In one implementation, processing each group of the to-be-trained respiratory wave data to obtain a to-be-trained respiratory rate corresponding to each group of the to-be-trained respiratory wave data specifically includes: filtering each group of the to-be-trained respiratory wave data; performing peak detection on the filtered each group of the to-be-trained respiratory wave data to obtain the average peak interval corresponding to each group of the respiratory wave data; and calculating the to-be-trained respiratory rate corresponding to each group of the to-be-trained respiratory wave data based on the average peak interval corresponding to each group of the to-be-trained respiratory wave data.
[0008] The above respiratory rate detection method filters, performs peak and trough detection on the to-be-trained pulse wave data in each group of the to-be-trained data, and filters and performs peak detection on the to-be-trained respiratory wave data, thereby respectively constructing an input vector corresponding to each group of the to-be-trained pulse wave data and a to-be-trained respiratory rate corresponding to each group of the to-be-trained respiratory wave data. Then, the neural network model is trained according to multiple groups of input vectors and multiple groups of to-be-trained respiratory rates to obtain a neural network parameter vector. Finally, based on the real-time pulse wave data and the neural network model parameter vector, the real-time respiratory rate of the user to be measured is determined. This method not only reduces the power consumption during the respiratory rate detection process, but also improves the detection accuracy of the respiratory rate and saves hardware storage resources.
[0009] In one implementation, determining the real-time respiratory rate of the user to be measured based on the real-time pulse wave data of the user to be measured within a preset time window width in the current motion state and the neural network model parameter vector specifically includes: obtaining the pulse wave data of the user to be measured in the current motion state, and intercepting the pulse wave data in the current motion state according to the preset time window width to obtain the real-time pulse wave data; filtering the real-time pulse wave data, and calculating the real-time respiratory rate of the user to be measured based on the filtered real-time pulse wave data and the neural network model parameter vector.
[0010] In a second aspect, the present application further provides a respiratory rate detection system, including: an acquisition module configured to acquire multiple groups of training data of multiple users in different motion states, where the training data includes training respiratory wave data and training pulse wave data within the same preset time window width; a processing module configured to process each group of the training pulse wave data to obtain an input vector corresponding to each group of the training pulse wave data, where the input vector includes an average peak interval, an average trough interval, an average peak amplitude, a number of peaks, and a number of troughs; the processing module is configured to process each group of the training respiratory wave data to obtain a training respiratory rate corresponding to each group of the training respiratory wave data; a training module configured to train a neural network model based on multiple groups of the input vectors and multiple groups of the training respiratory rates to obtain a neural network model parameter vector; a respiratory rate calculation module configured to determine the real-time respiratory rate of the user to be measured based on the real-time pulse wave data of the user to be measured within a preset time window width in the current motion state and the neural network model parameter vector.
[0011] In one implementation, the processing module includes: a filtering sub-module configured to filter each group of the training pulse wave data; a detection sub-module configured to perform peak and trough detection on each group of the filtered training pulse wave data to obtain the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the training pulse wave data; a construction sub-module configured to construct the input vector corresponding to each group of the training pulse wave data based on the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the training pulse wave data.
[0012] In one implementation, the processing module includes: a filtering sub-module configured to filter each group of the to-be-trained respiratory wave data; a detection sub-module configured to perform peak detection on each group of the filtered to-be-trained respiratory wave data to obtain the average peak interval corresponding to each group of the respiratory wave data; and a calculation sub-module configured to calculate the to-be-trained respiratory rate corresponding to each group of the to-be-trained respiratory wave data based on the average peak interval corresponding to each group of the to-be-trained respiratory wave data.
[0013] In one implementation, the respiratory rate calculation module is configured to obtain the pulse wave data of the user to be measured in the current exercise state, and intercept the pulse wave data in the current exercise state according to the preset time window width to obtain the real-time pulse wave data; the respiratory rate calculation module is configured to filter the real-time pulse wave data and calculate the real-time respiratory rate of the user to be measured based on the filtered real-time pulse wave data and the neural network model parameter vector.
[0014] In a third aspect, the present application further provides a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the respiratory rate detection method described in any of the above implementations are implemented.
[0015] In a fourth aspect, the present application further provides a wearable intelligent device, including a memory and a processor, and a computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the respiratory rate detection method described in any of the above implementations are implemented.
[0016] Compared with the prior art, the present invention has at least one of the following beneficial effects:
[0017] 1. By obtaining multiple groups of respiratory wave and pulse wave data of multiple users in different exercise states, and extracting characteristic parameters such as the average peak interval, average trough interval, average peak amplitude, number of peaks, and number of troughs of the pulse wave, as well as the respiratory rate corresponding to the respiratory wave, for training a neural network model. Then, the real-time respiratory rate of the user to be measured is determined through the neural network model parameter vector and the real-time pulse wave data of the user to be measured. This method reduces the power consumption during the respiratory rate detection process, improves the detection accuracy of the respiratory rate, and saves hardware storage resources.
[0018] 2. By filtering the to-be-trained pulse wave data in each group of to-be-trained data, detecting peaks and troughs, and filtering the to-be-trained respiratory wave data and detecting peaks, input vectors corresponding to each group of to-be-trained pulse wave data and to-be-trained respiratory rates corresponding to each group of to-be-trained respiratory wave data are respectively constructed. Then, the neural network model is trained according to multiple groups of input vectors and multiple groups of to-be-trained respiratory rates to obtain a neural network parameter vector. Finally, according to the real-time pulse wave data and the neural network model parameter vector, the real-time respiratory rate of the user to be measured is determined. This method not only reduces the power consumption in the process of detecting the respiratory rate, but also improves the detection accuracy of the respiratory rate and saves hardware storage resources. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] The above characteristics, technical features, advantages and their implementation manners of the present invention will be further described below in a clear and understandable manner in combination with the drawings in the preferred embodiments.
[0020] Figure 1 FIG. shows a flowchart of a respiratory rate detection method provided by an embodiment of the present application;
[0021] Figure 2 FIG. shows a flowchart of determining an input vector provided by an embodiment of the present application;
[0022] Figure 3 FIG. shows a flowchart of determining a to-be-trained respiratory rate provided by an embodiment of the present application;
[0023] Figure 4 FIG. shows a schematic diagram of peak-valley detection of to-be-trained pulse wave data provided by an embodiment of the present application;
[0024] Figure 5 FIG. shows a schematic diagram of peak detection of to-be-trained respiratory wave data provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the specific embodiments of the present invention will be described below with reference to the drawings. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts, and other embodiments can also be obtained.
[0026] To simplify the drawings, only the parts related to the invention are schematically shown in each figure, and they do not represent the actual structure of the product. In addition, to simplify the drawings and facilitate understanding, in some figures, for components with the same structure or function, only one of them is schematically illustrated, or only one of them is labeled. In this text, "one" not only means "only this one", but also can mean "more than one" situation.
[0027] It should be further understood that the term "and / or" used in the description of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.
[0028] In this text, it should be noted that unless otherwise clearly defined and limited, the terms "install", "connect", and "link" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two components. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.
[0029] In addition, in the description of this application, the terms "first", "second", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.
[0030] It should be noted that the above embodiments can be freely combined as needed. The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.
[0031] Respiratory rate refers to the number of breaths of the human body per unit time, usually expressed in "times / minute". It is an important physiological parameter for measuring the respiratory function of the human body and reflects the working frequency and intensity of the respiratory system. Under normal circumstances, the respiratory rate of adults is between 12 and 20 times per minute, but during exercise, emotional excitement or in a diseased state, the respiratory rate may increase significantly. The change of the respiratory rate can provide important references for clinical diagnosis, health monitoring and sports physiology research, and is one of the key indicators for evaluating the health status and physiological state of the human body.
[0032] Respiratory waves refer to physiological fluctuations caused by the ups and downs of the chest and abdomen and the exchange of gases in the lungs during the breathing process. It reflects the activity state of the respiratory system and can usually be detected by surface sensors or physiological monitoring equipment. For example, respiratory waves can be collected by placing electrodes on the human body. The frequency and amplitude of respiratory waves are closely related to the respiratory frequency, depth and breathing pattern. They are important physiological signals for evaluating respiratory function, monitoring sleep quality and diagnosing respiratory diseases. Under normal circumstances, respiratory waves show regular periodic changes, but in pathological conditions, their frequency, rhythm or amplitude may be abnormal.
[0033] The pulse wave refers to the pressure fluctuation caused by the blood flowing in the blood vessels as the heart beats. It is emitted from the heart, propagates along the arterial system, and reflects the functional state of the cardiovascular system. The propagation speed, shape and intensity of the pulse wave are closely related to factors such as the contractility of the heart, the elasticity of the blood vessels, and the viscosity of the blood. The embodiment of the present application pre-trains the neural network model based on multiple segments of respiratory wave data and pulse wave data. After the training is completed, the real-time respiratory rate of the user to be tested is determined based on the real-time pulse wave data of the user to be tested and the parameter vector of the neural network model, which can achieve at least one of the following beneficial effects: reducing the power consumption in the process of detecting the respiratory rate; or, improving the detection accuracy of the respiratory rate and saving hardware storage resources.
[0034] The following is explained with reference to the accompanying drawings:
[0035] Reference Figure 1 , which shows a flow chart of a respiratory rate detection method provided by an embodiment of the present application. Figure 1 As shown, including:
[0036] S100, obtaining multiple groups of training data of multiple users in different motion states, wherein the training data include breathing wave data and pulse wave data within the same preset time window.
[0037] S110, processing each group of pulse wave data to be trained to obtain an input vector corresponding to each group of pulse wave data to be trained, the input vector including an average peak interval, an average trough interval, an average peak amplitude, a number of peaks, and a number of troughs.
[0038] S120, processing each group of respiratory wave data to be trained to obtain a respiratory rate to be trained corresponding to each group of respiratory wave data to be trained.
[0039] S130, training the neural network model based on the multiple groups of input vectors and the multiple groups of breathing rates to be trained to obtain a neural network model parameter vector.
[0040] S140. Determine the real-time respiratory rate of the user to be measured based on the real-time pulse wave data and the neural network model parameter vector of the user to be measured within a preset time window width in the current exercise state.
[0041] Synchronously collect the respiratory wave data and pulse wave data of multiple users in different exercise states. For example, the respiratory wave data and pulse wave data of multiple users in the static state can be collected, and the respiratory wave data and pulse wave data of multiple users in the running state can be collected, and mark the actual number of breaths of the respiratory wave data per minute.
[0042] Segment the collected respiratory wave data and pulse wave data according to a preset time window width (the preset time window width can be 1 minute) to obtain multiple segments of respiratory wave data and pulse wave data of multiple users. Take the respiratory wave data and pulse wave data within the same preset time window width of each user as the respiratory wave data to be trained and the pulse wave data to be trained, and then construct the data to be trained. That is, the respiratory wave data to be trained and the pulse wave data to be trained in each group of data to be trained are the respiratory wave data and pulse wave data of the same user in the same exercise state and within the same preset time window width.
[0043] Perform band-pass filtering, peak and trough detection on the pulse wave data to be trained in each group of data to be trained to obtain the average peak interval, average trough interval, average peak amplitude, number of peaks and number of troughs of each group of pulse wave data to be trained, and then construct the input vector of each group of pulse wave data to be trained. Perform band-pass filtering and peak detection on the respiratory wave data to be trained in each group of data to be trained to obtain the average peak interval of each group of respiratory wave data to be trained, and the respiratory rate to be trained corresponding to each group of respiratory wave data to be trained can be obtained according to the average peak interval.
[0044] Furthermore, train the neural network model (the neural network model can be a Bayesian neural network model) according to multiple groups of input vectors and multiple groups of respiratory rates to be trained to obtain the neural network model parameter vector. Then, determine the real-time respiratory rate of the user to be measured based on the real-time pulse wave data of the user to be measured within the preset time window width in the current exercise state and the neural network model parameter vector.
[0045] In the embodiment of the present application, by obtaining multiple groups of respiratory wave and pulse wave data of multiple users in different exercise states, and extracting characteristic parameters such as the average peak interval, average trough interval, average peak amplitude, number of peaks and number of troughs of the pulse wave, and the respiratory rate corresponding to the respiratory wave, for training the neural network model. Then, through the neural network model parameter vector and the real-time pulse wave data of the user to be measured, the real-time respiratory rate of the user to be measured is determined. This method reduces the power consumption in the process of respiratory rate detection, improves the detection accuracy of the respiratory rate, and saves hardware storage resources at the same time.
[0046] Reference appendix Figure 2 , which shows a flowchart for determining an input vector provided by an embodiment of the present application.
[0047] As Figure 2 shown, it includes:
[0048] S200, filtering each group of pulse wave data to be trained.
[0049] S210, detecting peaks and troughs of each group of filtered pulse wave data to be trained, and obtaining the average peak interval, average trough interval, average peak amplitude, number of peaks, and number of troughs corresponding to each group of pulse wave data.
[0050] S220, constructing an input vector corresponding to each group of pulse wave data to be trained based on the average peak interval, average trough interval, average peak amplitude, number of peaks, and number of troughs corresponding to each group of pulse wave data.
[0051] Among them, the process of determining the respiration rate to be trained is similar to the process of determining the input vector. For example, reference appendix Figure 3 , which shows a flowchart for determining the respiration rate to be trained provided by an embodiment of the present application. As Figure 3 shown, it includes:
[0052] S300, filtering each group of respiration wave data to be trained.
[0053] S310, detecting peaks of each group of filtered respiration wave data to be trained, and obtaining the average peak interval corresponding to each group of respiration wave data.
[0054] S320, calculating the respiration rate to be trained corresponding to each group of respiration wave data to be trained based on the average peak interval corresponding to each group of respiration wave data to be trained.
[0055] The pulse wave data to be trained and the respiration wave data to be trained in each group of data to be trained are essentially vectors composed of 100 data points. That is, each group of pulse wave data to be trained and each group of respiration wave data to be trained can be respectively expressed as PPG part =[ppg1, ppg2... ppg100], RSP part =[rsp1, rsp2... rsp100]. Among them, the data points ppg1, ppg2... ppg100 are signal sampling points of the pulse wave data to be trained (or called pulse wave signal sampling points), and rsp1, rsp2... rsp100 are signal sampling points of the respiration wave data to be trained (or called true respiration wave signal sampling points).
[0056] For each group of pulse wave data to be trained PPG partPerform band - pass filtering (the pass - band range of the band - pass filter is [0.7Hz, 3.5Hz], and the stop - band cut - off frequency is [0.5Hz, 5Hz]) to obtain the filtered PPG data of each group of training pulse waves filter , the filtered PPG data of each group of training pulse waves filter can be expressed as PPG filter = [ppg filter1 , ppg filter2 ... ppg filter100 . Perform peak and valley detection on the filtered PPG data of each group of training pulse waves filter (refer to Appendix Figure 4 ), and obtain the average peak - to - peak interval (or average peak - peak interval) t filter , average valley - to - valley interval (or average valley - valley interval) t peak , average peak amplitude amp low , number of peaks num mean and number of valleys num peak corresponding to each group of training pulse wave data. Construct the average peak - to - peak interval t low , average valley - to - valley interval t filter , average peak amplitude amp peak , number of peaks num low , and number of valleys num mean corresponding to each group of training pulse wave data into the input vector X (or the input vector X of the neural network) for each group of training pulse wave data. The input vector X = [t peak , t low , amp peak , num low , num mean , num peak , num low .
[0057] Similarly, perform band - pass filtering (the pass - band range of the band - pass filter is [0.7Hz, 3.5Hz], and the stop - band cut - off frequency is [0.5Hz, 5Hz]) on each group of training respiratory wave data RSP part to obtain the filtered RSP data of each group of training respiratory waves part , the filtered RSP data of each group of training respiratory waves part can be expressed as RSP filter = [rsp filter1 , rsp filter2 ... rsp filter100 . Perform peak detection on the filtered RSP data of each group of training respiratory waves filter (refer to Appendix Figure 5 ), and obtain the RSP data of each group of training respiratory wavesfilter The corresponding average peak interval (or average respiratory peak-to-peak interval) rsp peak According to the average peak interval corresponding to each group of respiratory wave data to be trained, calculate the corresponding respiratory rate to be trained (or the true respiratory rate of the person being measured within 1 minute) RSP real , and the corresponding calculation formula is RSP real = 1 / rsp peak Thus, after processing each group of data to be trained, the corresponding input vector X and the respiratory rate to be trained RSP will be obtained real , and the input vector X and the respiratory rate to be trained RSP of each group of data to be trained real correspond to each other
[0058] Furthermore, after obtaining multiple groups of respiratory rates to be trained RSP real and multiple groups of input vectors X, the multiple groups of respiratory rates to be trained RSP real and the multiple groups of input vectors X can be written in matrix form respectively. For example where X is the neural network input matrix constructed by multiple groups of input vectors X (the multiple groups of input vectors X are X1, X2, X3...X in the aforementioned matrix respectively n ), Y is the neural network output matrix constructed by multiple groups of respiratory rates to be trained RSP real Using the neural network input matrix X as the input of the neural network model and the neural network output matrix Y as the output of the neural network model to train the neural network model, and finally obtaining the neural network model parameter vector W. The neural network model parameter vector W can be expressed as W = [w1, w2...wi]. Among them, w1, w2...wi are the trained neural network parameters, and i is the number of neural network parameters. Then, according to the real-time pulse wave data of the user to be measured within the preset time window width in the current exercise state and the neural network model parameter vector W, determine the real-time respiratory rate of the user to be measured
[0059] In the embodiment of the present application, by filtering, peak and valley detection for the pulse wave data to be trained in each group of data to be trained, and filtering and peak detection for the respiratory wave data to be trained, the input vector corresponding to each group of pulse wave data to be trained and the respiratory rate to be trained corresponding to each group of respiratory wave data to be trained are respectively constructed. Then, the neural network model is trained according to multiple groups of input vectors and multiple groups of respiratory rates to be trained to obtain the neural network parameter vector. Finally, according to the real-time pulse wave data and the neural network model parameter vector, the real-time respiratory rate of the user to be measured is determined. This method not only reduces the power consumption in the process of detecting the respiratory rate, but also improves the detection accuracy of the respiratory rate and saves hardware storage resources
[0060] In one embodiment of the present application, based on the real-time pulse wave data and the neural network model parameter vector of the user to be measured within a preset time window width in the current exercise state, the real-time respiratory rate of the user to be measured is determined, which specifically includes: obtaining the pulse wave data of the user to be measured in the current exercise state, and intercepting the pulse wave data in the current exercise state according to the preset time window width to obtain the real-time pulse wave data; filtering the real-time pulse wave data, and calculating the real-time respiratory rate of the user to be measured based on the filtered real-time pulse wave data and the neural network model parameter vector.
[0061] Real-time collect the pulse wave data of the user to be measured in different exercise states (the sampling frequency can be 100Hz), and intercept the pulse wave data of the user to be measured during the entire exercise process according to the preset time window width to obtain the real-time pulse wave data PPG online . Among them, the real-time pulse wave data PPG online 's vector representation form is PPG online = [ppg online1 , ppg online2 ... ppg online100 . Perform band-pass filtering on the real-time pulse wave data PPG online (the passband range of the band-pass filter is [0.7Hz, 3.5Hz], and the stopband cut-off frequency is [0.5Hz, 5Hz]) to obtain the filtered real-time pulse wave data PPG online-filter . Then, substitute the filtered real-time pulse wave data PPG online-filter and the neural network model parameter vector W into the formula Y real＝ W * PPG online-filter to obtain the real-time respiratory rate Y real of the user to be measured.
[0062] The embodiment of the present application also provides a respiratory rate detection system, including: an acquisition module configured to acquire multiple groups of training data of multiple users in different exercise states, where the training data includes training respiratory wave data and training pulse wave data within the same preset time window width; a processing module configured to process each group of training pulse wave data to obtain an input vector corresponding to each group of training pulse wave data, where the input vector includes the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs; a processing module configured to process each group of training respiratory wave data to obtain the training respiratory rate corresponding to each group of training respiratory wave data; a training module configured to train the neural network model based on multiple groups of input vectors and multiple groups of training respiratory rates to obtain the neural network model parameter vector; a respiratory rate calculation module configured to determine the real-time respiratory rate of the user to be measured based on the real-time pulse wave data and the neural network model parameter vector of the user to be measured within the preset time window width in the current exercise state.
[0063] The detailed content of the embodiments of the present application has been described in the foregoing embodiments, and will not be elaborated herein.
[0064] In an embodiment of the present application, the processing module includes: a filtering sub-module configured to filter each group of pulse wave data to be trained; a detection sub-module configured to perform peak and trough detection on each group of filtered pulse wave data to be trained, and obtain the average peak interval, average trough interval, average peak amplitude, number of wave peaks, and number of wave troughs corresponding to each group of pulse wave data to be trained; a construction sub-module configured to construct an input vector corresponding to each group of pulse wave data to be trained based on the average peak interval, average trough interval, average peak amplitude, number of wave peaks, and number of wave troughs corresponding to each group of pulse wave data to be trained.
[0065] In an embodiment of the present application, the processing module includes: a filtering sub-module configured to filter each group of respiratory wave data to be trained; a detection sub-module configured to perform peak detection on each group of filtered respiratory wave data to be trained, and obtain the average peak interval corresponding to each group of respiratory wave data; a calculation sub-module configured to calculate the respiratory rate to be trained corresponding to each group of respiratory wave data to be trained based on the average peak interval corresponding to each group of respiratory wave data to be trained.
[0066] In an embodiment of the present application, the respiratory rate calculation module is configured to obtain the pulse wave data of the user to be measured in the current exercise state, and intercept the pulse wave data in the current exercise state according to a preset time window width to obtain real-time pulse wave data; the respiratory rate calculation module is configured to filter the real-time pulse wave data, and calculate the real-time respiratory rate of the user to be measured based on the filtered real-time pulse wave data and the neural network model parameter vector.
[0067] The embodiments of the present application also provide a storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the respiratory rate detection method in any of the above embodiments are implemented.
[0068] The embodiments of the present application also provide a wearable intelligent device, including a memory and a processor, and a computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the respiratory rate detection method in any of the above embodiments are implemented.
[0069] It should be noted that the above embodiments can be freely combined as needed. The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.
Claims
1. A method for detecting respiratory rate, characterized in that, Including: Obtain multiple sets of training data to be trained for multiple users in different exercise states, where the training data to be trained includes training respiratory wave data and training pulse wave data within the same preset time window width; Process each set of the training pulse wave data to obtain an input vector corresponding to each set of the training pulse wave data, where the input vector includes an average peak interval, an average trough interval, an average peak amplitude, the number of peaks, and the number of troughs; Process each set of the training respiratory wave data to obtain a training respiratory rate corresponding to each set of the training respiratory wave data; Train a neural network model based on multiple sets of the input vectors and multiple sets of the training respiratory rates to obtain a neural network model parameter vector; Determine the real-time respiratory rate of a user to be measured based on the real-time pulse wave data of the user to be measured within the preset time window width in the current exercise state and the neural network model parameter vector.
2. The respiration rate detection method according to claim 1, wherein The processing of each set of the training pulse wave data to obtain an input vector corresponding to each set of the training pulse wave data specifically includes: Filter each set of the training pulse wave data; Perform peak and trough detection on each set of the filtered training pulse wave data to obtain the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each set of the training pulse wave data; Construct the input vector corresponding to each set of the training pulse wave data based on the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each set of the training pulse wave data.
3. The respiratory rate detection method according to claim 1, wherein The processing of each set of the training respiratory wave data to obtain a training respiratory rate corresponding to each set of the training respiratory wave data specifically includes: Filter each set of the training respiratory wave data; Perform peak detection on each set of the filtered training respiratory wave data to obtain the average peak interval corresponding to each set of the respiratory wave data; Calculate the training respiratory rate corresponding to each set of the training respiratory wave data based on the average peak interval corresponding to each set of the training respiratory wave data.
4. The respiratory rate detection method according to any one of claims 1-3, characterized in that, The determining of the real-time respiratory rate of a user to be measured based on the real-time pulse wave data of the user to be measured within the preset time window width in the current exercise state and the neural network model parameter vector specifically includes: Obtain the pulse wave data of the user to be measured in the current exercise state, and intercept the pulse wave data in the current exercise state according to the preset time window width to obtain the real-time pulse wave data; Filter the real-time pulse wave data, and calculate the real-time respiratory rate of the user to be measured based on the filtered real-time pulse wave data and the neural network model parameter vector.
5. A respiratory rate detection system, characterized in that, Including: An obtaining module configured to obtain multiple sets of training data to be trained for multiple users in different exercise states, where the training data to be trained includes training respiratory wave data and training pulse wave data within the same preset time window width; A processing module, configured to process each group of the to-be-trained pulse wave data to obtain an input vector corresponding to each group of the to-be-trained pulse wave data, where the input vector includes an average peak interval, an average trough interval, an average peak amplitude, a number of peaks, and a number of troughs; The processing module is configured to process each group of the to-be-trained respiration wave data to obtain a to-be-trained respiration rate corresponding to each group of the to-be-trained respiration wave data; A training module, configured to train a neural network model based on multiple groups of the input vectors and multiple groups of the to-be-trained respiration rates to obtain a neural network model parameter vector; A respiration rate calculation module, configured to determine a real-time respiration rate of a to-be-tested user based on real-time pulse wave data of the to-be-tested user within a preset time window width in a current exercise state and the neural network model parameter vector.
6. The respiration rate detection system according to claim 5, characterized in that, The processing module includes: A filtering sub-module, configured to filter each group of the to-be-trained pulse wave data; A detection sub-module, configured to perform peak and trough detection on each group of the filtered to-be-trained pulse wave data to obtain the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the to-be-trained pulse wave data; A construction sub-module, configured to construct the input vector corresponding to each group of the to-be-trained pulse wave data based on the average peak interval, the average trough interval, the average peak amplitude, the number of peaks, and the number of troughs corresponding to each group of the to-be-trained pulse wave data.
7. The respiration rate detection system according to claim 5, characterized in that, The processing module includes: A filtering sub-module, configured to filter each group of the to-be-trained respiration wave data; A detection sub-module, configured to perform peak detection on each group of the filtered to-be-trained respiration wave data to obtain the average peak interval corresponding to each group of the respiration wave data; A calculation sub-module, configured to calculate the to-be-trained respiration rate corresponding to each group of the to-be-trained respiration wave data based on the average peak interval corresponding to each group of the to-be-trained respiration wave data.
8. The respiratory rate detection system according to any one of claims 5-7, characterized in that, The respiration rate calculation module is configured to obtain the pulse wave data of the to-be-tested user in the current exercise state and intercept the pulse wave data in the current exercise state according to the preset time window width to obtain the real-time pulse wave data; The respiration rate calculation module is configured to filter the real-time pulse wave data and calculate the real-time respiration rate of the to-be-tested user based on the filtered real-time pulse wave data and the neural network model parameter vector.
9. A storage medium, characterized in that, A computer program is stored thereon, and when the computer program is executed by a processor, the steps of the respiration rate detection method according to any one of claims 1-4 are implemented.
10. A wearable intelligent device, characterized in that, It includes a memory and a processor, and a computer program is stored on the memory, and when the computer program is executed by the processor, the steps of the respiration rate detection method according to any one of claims 1-4 are implemented.