CN118141354A - Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device - Google Patents
Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device Download PDFInfo
- Publication number
- CN118141354A CN118141354A CN202410572262.9A CN202410572262A CN118141354A CN 118141354 A CN118141354 A CN 118141354A CN 202410572262 A CN202410572262 A CN 202410572262A CN 118141354 A CN118141354 A CN 118141354A
- Authority
- CN
- China
- Prior art keywords
- heart rate
- scene
- determining
- frequency domain
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 99
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000005457 optimization Methods 0.000 title claims abstract description 23
- 238000001228 spectrum Methods 0.000 claims abstract description 75
- 238000012545 processing Methods 0.000 claims abstract description 51
- 238000001914 filtration Methods 0.000 claims abstract description 31
- 238000000605 extraction Methods 0.000 claims abstract description 30
- 238000013186 photoplethysmography Methods 0.000 claims description 47
- 230000003044 adaptive effect Effects 0.000 claims description 18
- 238000012549 training Methods 0.000 claims description 15
- 238000007781 pre-processing Methods 0.000 claims description 10
- 238000004364 calculation method Methods 0.000 claims description 9
- 230000003595 spectral effect Effects 0.000 claims description 8
- 230000007704 transition Effects 0.000 claims description 8
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000013527 convolutional neural network Methods 0.000 claims description 5
- 238000003066 decision tree Methods 0.000 claims description 5
- 230000002861 ventricular Effects 0.000 claims description 5
- 238000012806 monitoring device Methods 0.000 abstract 1
- 238000012805 post-processing Methods 0.000 description 8
- 238000013461 design Methods 0.000 description 6
- 238000004891 communication Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 5
- 230000036541 health Effects 0.000 description 5
- 238000009532 heart rate measurement Methods 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 5
- 230000002159 abnormal effect Effects 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000006185 dispersion Substances 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000001788 irregular Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 230000000737 periodic effect Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 230000006806 disease prevention Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 208000012661 Dyskinesia Diseases 0.000 description 1
- 208000015592 Involuntary movements Diseases 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004590 computer program Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000006073 displacement reaction Methods 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 230000014509 gene expression Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000017311 musculoskeletal movement, spinal reflex action Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000004962 physiological condition Effects 0.000 description 1
- 230000001766 physiological effect Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000241 respiratory effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009182 swimming Effects 0.000 description 1
- 230000021542 voluntary musculoskeletal movement Effects 0.000 description 1
Landscapes
- Measuring Pulse, Heart Rate, Blood Pressure Or Blood Flow (AREA)
Abstract
The application provides a heart rate monitoring value estimation optimization method and wearing equipment of a self-adaptive daily multi-scene, which relate to the technical field of heart rate monitoring and comprise the following steps: collecting accelerometer data of the wearable equipment, performing time-frequency domain processing and feature extraction, determining time domain features and frequency domain features, and determining motion state frame data by performing motion state identification; performing scene recognition processing based on the motion state frame data, and determining a current monitoring scene; matching a target filter in a pre-configured filter bank according to the current monitoring scene, and filtering interference signals of photoelectric volume pulse wave signals of the wearable equipment through the target filter; performing dual-spectrum combined spectrum peak identification based on the filtered photoplethysmogram data and the accelerometer data, and determining an estimated heart rate value; and optimizing the estimated heart rate value based on the scene type of the current monitoring scene, and determining a target heart rate monitoring value. The heart rate monitoring method and the heart rate monitoring device improve the accuracy of heart rate monitoring.
Description
Technical Field
The application relates to the technical field of heart rate monitoring, in particular to a heart rate monitoring value estimation optimization method and wearable equipment for self-adaptive daily multi-scene.
Background
With the increasing importance of people on physical health, disease prevention and sports health management, wearable devices are widely popular, and heart rate monitoring is also a type of health monitoring function with high attention. In the related art, when the wearable device monitors the heart rate, a photoplethysmography (PhotoPlethysmoGraphy, PPG) is generally adopted, and the heart rate is detected by using reflection or scattering of light and conversion of photoelectric signals; there are also methods of heart rate monitoring in combination with PPG and accelerometers (Accelerometer, ACC). However, PPG may be subject to interference from a variety of factors such as environmental interference, noise interference, and motion interference when performing heart rate monitoring; at present, in a complex daily scene, the mode of combining PPG with ACC often has the condition of inaccurate heart rate detection.
Disclosure of Invention
The application aims to provide a heart rate monitoring value estimation optimization method and wearable equipment for self-adaptive daily multi-scenario so as to alleviate at least one technical problem in the prior art.
In a first aspect, the present invention provides a heart rate monitor value estimation optimization method for adaptive daily multi-scenario, including:
Collecting accelerometer data of the wearable equipment, performing time-frequency domain processing and feature extraction on the accelerometer data, determining time domain features and frequency domain features, performing motion state identification through the time domain features and the frequency domain features, and determining motion state frame data;
performing scene recognition processing based on the motion state frame data, and determining a current monitoring scene;
Matching a target filter in a pre-configured filter bank according to the current monitoring scene, and filtering interference signals of photoelectric volume pulse wave signals of the wearable equipment through the target filter; the filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene;
performing dual-spectrum combined spectrum peak identification based on the filtered photoplethysmogram data and the accelerometer data, and determining an estimated heart rate value;
and optimizing the estimated heart rate value based on the scene type of the current monitoring scene, and determining a target heart rate monitoring value.
In an alternative embodiment, scene recognition processing is performed based on the motion state frame data, and determining the current monitored scene includes:
Determining motion state frame data of a designated frame number as a target historical motion state frame;
Carrying out one-dimensional signal convolution processing on the target historical motion state frame, carrying out scene classification prediction, determining the current monitoring scene, and outputting a corresponding first prediction probability as a target confidence coefficient;
And/or the number of the groups of groups,
And inputting the target historical motion state frame into a pre-trained decision tree model for scene classification prediction, determining the current monitoring scene, and outputting a corresponding second prediction probability as a target confidence coefficient.
In an alternative embodiment, the filter bank includes a plurality of classes of filters and further includes a time domain filter and a frequency domain filter; each target monitoring scene corresponds to a corresponding scene time domain filter and a scene frequency domain filter.
In an alternative embodiment, filtering the interference signal of the photoplethysmography signal of the wearable device by the target filter includes:
acquiring photoelectric volume pulse wave signals of the wearable equipment, and preprocessing the photoelectric volume pulse wave signals;
Respectively determining a corresponding frequency domain heart rate filter and a corresponding time domain heart rate filter according to the motion state corresponding to the current frame and the current monitoring scene;
The photoelectric volume pulse wave signals of the wearable equipment are subjected to first interference signal filtering and frequency domain conversion processing through a frequency domain heart rate filter, and filtered frequency domain signals are determined;
And performing second interference signal filtering on the photoplethysmography signal of the wearable device through a time domain heart rate filter, and determining a filtered time domain signal.
In an alternative embodiment, the dual-spectrum combined spectrum peak identification is performed based on the filtered photoplethysmography data and the accelerometer data, and the determining the estimated heart rate value includes:
determining a frequency spectrum tracking area according to the signal quality and the historical heart rate value;
Respectively extracting a first frequency spectrum of a photoelectric volume pulse wave signal and a second frequency spectrum of an accelerometer signal in a frequency spectrum tracking area;
Inputting a first spectrum of the photoelectric volume pulse wave signal and a second spectrum of the accelerometer signal in a specified time and frequency domain range into a convolutional neural network, and performing convolutional operation on the first spectrum of the photoelectric volume pulse wave signal and the second spectrum of the accelerometer signal in a specified tracking area range respectively;
Sequentially inputting the convolution result into a joint convolution layer and a full connection layer to obtain a joint main peak identification result and identification confidence;
determining a frequency domain estimated heart rate value based on the combined main peak recognition result and the recognition confidence;
The time domain estimated heart rate value is determined by performing time domain ventricular beat interval calculation on the accelerometer data and the photoplethysmography wave signals.
In an alternative embodiment, optimizing the estimated heart rate value based on the scene type of the current monitored scene, determining the target heart rate monitor value includes:
Judging the scene type of the current monitoring scene; the scene types comprise a stable scene type, a transition scene type and a switching scene type;
When the scene type is stable, determining a target heart rate monitoring value based on the frequency domain estimated heart rate value or the time domain estimated heart rate value;
When the heart rate monitoring system is in the transition scene type or the switching scene type, the joint strength of the estimated heart rate value and the time domain estimated heart rate value is determined based on a pre-trained heart rate reference decision model, and the estimated heart rate value is optimized through the joint strength, so that the target heart rate monitoring value is determined.
In an alternative embodiment, the method further comprises:
Determining a training sample based on the signal quality of the acquired time domain signal and frequency domain signal, the joint main peak recognition result, the time domain feature, the frequency domain estimated heart rate value, the time domain estimated heart rate value and the historical heart rate condition;
and training the initial heart rate reference decision model based on the training sample until convergence, and determining a trained heart rate reference decision model.
In a second aspect, the present invention provides a heart rate monitor value estimation optimization device for adaptive daily multi-scenario, including:
The data acquisition processing module is used for acquiring accelerometer data of the wearable equipment, carrying out time-frequency domain processing and feature extraction on the accelerometer data, determining time domain features and frequency domain features, carrying out motion state identification through the time domain features and the frequency domain features, and determining motion state frame data;
The scene recognition module is used for performing scene recognition processing based on the motion state frame data and determining a current monitoring scene;
The filtering module is used for matching a target filter in a pre-configured filter bank according to the current monitoring scene, and filtering interference signals of photoelectric volume pulse wave signals of the wearable equipment through the target filter; the filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene;
The combined spectrum peak identification module is used for carrying out dual-spectrum combined spectrum peak identification based on the filtered photoelectric volume pulse wave data and the accelerometer data, and determining an estimated heart rate value;
and the heart rate value optimization module is used for optimizing the estimated heart rate value based on the scene type of the current monitoring scene and determining a target heart rate monitoring value.
In a third aspect, the present invention provides a wearable device comprising a processor and a memory, the memory storing computer executable instructions executable by the processor to implement the heart rate monitor value estimation optimization method of any of the foregoing embodiments for adaptive daily multi-scenario.
In a fourth aspect, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when invoked and executed by a processor, cause the processor to implement the method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario of any of the foregoing embodiments.
The heart rate monitoring value estimation optimization method and the wearable device for the self-adaptive daily multi-scene have the following beneficial effects:
According to the embodiment of the application, the scenes are identified in advance, and corresponding heart rate monitoring calculation can be performed aiming at different scenes, so that the heart rate monitoring value is more accurate. By pre-configuring the filter bank and adopting different filters, better data preprocessing can be achieved in a specified scene, such as the problem that confusion under daily heart rate extraction can be reduced due to the fact that abundant high-frequency components tend to be reserved in a frequency band. The double spectrums in the spectrum tracking area are processed through the neural network, so that a real main peak can be found under the condition of main peak energy dispersion or other abnormal conditions, heart rate extraction errors caused by confusion of the main peak are prevented, and meanwhile, the influence of motion artifact components can be reduced more effectively. The heart rate extraction and the heart rate post-processing are determined according to scene results, complex problems can be decomposed through scene division processing, differential design is carried out on the heart rate extraction according to scenes, and a more accurate processing strategy can be adopted on appointed scenes, so that the finally output heart rate value is more accurate.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present application, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a heart rate monitor value estimation optimization method of a self-adaptive daily multi-scenario provided by an embodiment of the present application;
Fig. 2 is a flowchart of scene recognition based on ACC data according to an embodiment of the present application;
FIG. 3 is a schematic diagram of a filtering process according to an embodiment of the present application;
FIG. 4 is a flowchart of determining a frequency domain estimated heart rate based on a combined main peak recognition result according to an embodiment of the present application;
Fig. 5 is a block diagram of an adaptive daily multi-scenario heart rate monitor value estimation optimization device according to an embodiment of the present application;
Fig. 6 is a structural diagram of a wearable device according to an embodiment of the present application.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application, and it is apparent that the described embodiments are some embodiments of the present application, but not all embodiments of the present application. The components of the embodiments of the present application generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
Thus, the following detailed description of the embodiments of the application, as presented in the figures, is not intended to limit the scope of the application, as claimed, but is merely representative of selected embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
It should be noted that: like reference numerals and letters denote like items in the following figures, and thus once an item is defined in one figure, no further definition or explanation thereof is necessary in the following figures.
With the increasing importance of people on physical health, disease prevention and sports health management, wearable devices capable of monitoring physiological conditions at any time have become the primary choice for a vast population. The heart rate is an important sign signal, and the physical condition of the wearer can be fed back in real time. People wear equipment for a long time in daily complex scenes, and how to improve the heart rate precision of the wearable equipment is always attracting attention.
In wearable devices, assisting heart rate detection with photoplethysmography (PPG) has become a common solution. The PPG works on the principle that the skin is illuminated by a light emitting diode LED as a light source using reflection or scattering of light, and the photo sensor PD detects the remaining transmitted or reflected light, after which the optical signal is converted into an electrical signal.
In the PPG signal collected by the wearable device, there may be influence caused by environmental interference, noise and motion, thereby causing signal distortion, resulting in low heart rate detection accuracy. Optical effects are typically caused by ambient light, such as indoor lighting typically includes flicker, and the shift in the light signal is periodically detected; physiological effects typically result from voluntary or involuntary movements of the wearer's body causing mechanical displacement of the sensor relative to the tissue, even small movements can affect the PPG signal, such as respiratory movements can be coupled into the PPG waveform, walking and running, etc. introducing motion artifacts, etc. Meanwhile, the performances of the light emitting diode and the photosensitive element, the design of heart rate structures, the design of a software-level dimming mechanism and the like can have complex influence on signals.
In the prior art, data acquired by combining a PPG and an Accelerometer (ACC) are generally considered, and the conventional heart rate estimation method comprises four links of preprocessing, artifact removal, heart rate extraction and heart rate value post-processing.
Specifically, the preprocessing module generally performs baseline drift removal, filtering, noise reduction and other processes on the original signal.
The artifact removal module is commonly used in two modes, namely, firstly, carrying out Fourier transform or wavelet transform and the like on PPG data and ACC data respectively to obtain a frequency domain signal, and then removing spectral peak influence of the ACC data in frequency domain spectral peaks of the PPG data, such as spectral subtraction; and secondly, adopting an adaptive filtering method, directly filtering out motion artifact interference in the PPG signal according to the correlation between the motion artifact and the ACC signal, and then carrying out modes such as Fourier transform or wavelet transform to obtain frequency domain information.
The heart rate extraction module generally refers to performing heart rate value calculation on the time domain or frequency domain information obtained through the processing through confirming the time domain RR interval or the frequency domain spectrum peak, so as to obtain a preliminary heart rate value.
The heart rate value post-processing module is very important, and the heart rate value obtained by the heart rate extraction module can fall in an error interval in consideration of all links from acquisition and processing, so that the heart rate value needs to be corrected according to the conditions of historical heart rate, historical spectrum peaks and the like.
This flow is a more general solution, and people in prior art promote heart rate precision through optimizing four above modules or adding extra sensor, but unfortunately, under complicated daily scene, still often can appear heart rate detection inaccurate condition.
In the prior art, a part of the technologies focus on breaking through the artifact removal method:
① ACC spectral peak analysis plus PPG frequency domain spectral peak processing sometimes cannot remove all important motion artifact components because spectral peaks may occur in adjacent intervals.
② The adaptive filtering method eliminates noise and is sensitive to a predefined reference signal, and unsuitable reference signal selection can influence the performance of motion artifact elimination;
③ The more complex signal decomposition and reconstruction method involves a large amount of matrix operations, and the algorithm has higher complexity and cannot run on the wearable device.
A part of technologies select and optimize heart rate extraction and heart rate value post-processing methods:
① The heart rate in the time domain and the heart rate in the frequency domain are weighted and summed, and the self-adaptive weight is preset or learned by using the historical data according to different scenes, but even in the same scene, the difference of the signals can cause poor weighting and summation effect.
② The heart rate range is preset according to the historical heart rate, such as +/-6, searching and tracking are carried out in a preset range, but follow-up cannot jump out due to error accumulation and signal conditions in the processing process.
③ According to the historical wave crest, identifying potential curve clusters, completing the extraction of the current heart rate value through a jump mechanism of a spectrum peak curve on a time-frequency chart, and maintaining the complex curve clusters and continuously carrying out the processes of tracking, deleting and the like.
With the development of the neural network method, some technologies choose to directly input signals or input time-frequency domain features by means of the neural network method and output corresponding heart rates, the former does not need a great deal of expert experience, the latter still needs an expert to complete feature design, but all depend on a great deal of data, and the data sets are easy to be over-fitted, so that generalization is poor.
Another part of the technical options is to resort to additional sensors, including inertial sensors that are sensitive to motion but insensitive to changes in the optical environment, and third wavelength light that is sensitive to changes in the optical but insensitive to motion, all requiring an additional large amount of computation and additional power consumption. In addition, techniques to choose to use GPS, barometer, etc. to aid in deciding on the current motion state, or to invoke ECG-assisted heart rate detection, involve extensive computation and additional power consumption, and ECG assistance also requires the user to use a specified measurement regime.
Based on the above, in consideration of various states of the wearable device wearer, environmental interference, noise, performance of the exercise hardware, a dimming mechanism of a software layer and the like in a daily complex scene, various abnormal conditions can be encountered under a general heart rate solution to cause heart rate value errors. The embodiment of the application starts from two major directions of scene deconstruction and time-frequency domain signal processing optimization, provides a heart rate monitoring value estimation optimization method and wearing equipment for self-adaptive daily multi-scene, simplifies the calculation mode of heart rate monitoring values, and improves the accuracy degree of heart rate value estimation.
The embodiment of the application provides a heart rate monitoring value estimation optimization method of a self-adaptive daily multi-scene, which is shown in fig. 1 and mainly comprises the following steps:
Step S110, accelerometer data of the wearable device are collected, time-frequency domain processing and feature extraction are carried out on the accelerometer data, time domain features and frequency domain features are determined, motion state identification is carried out through the time domain features and the frequency domain features, and motion state frame data are determined.
In one embodiment, the acquisition of the accelerometer of the wearable device may include data acquisition in a variety of situations, such as in an all-weather worn device, heart rate measurement may be artificially initiated actively, while the wearer is often required to remain stationary; the background opening can also be automatically monitored at an irregular time, and the wearer is not necessarily in what state. Meanwhile, the device generally supports real-time heart rate measurement in an on-exercise mode, wherein the exercise mode selection is associated with the state of the wearer, and the exercise mode comprises periodic exercise and aperiodic exercise.
The time-frequency domain processing of the accelerometer data refers to processing of the time domain or the frequency domain respectively, so as to obtain a time domain signal and a frequency domain signal. Further, the time domain signals are respectively extracted with features, the time domain features comprise but are not limited to amplitude mean value, wave peak value, wave trough value, wave peak and wave trough difference value and the like of one frame of data, the frequency domain features comprise but are not limited to frequency spectrum peak value information, frequency spectrum dispersion and the like, the motion state of the current frame is identified according to the extracted time domain features and frequency domain features and stored in the historical motion state, and the motion state frame data are determined.
Step S120, scene recognition processing is carried out based on the motion state frame data, and the current monitoring scene is determined.
Numerous scenarios of heart rate monitoring are contemplated in which the wearable device is used, such as including, but not limited to, stationary, walking, running, riding, batting, swimming, other periodic movements, and other non-periodic movements. The heart rates in different scenes have large differences, so that scene recognition processing can be performed in advance, and the current monitoring scene corresponding to the current heart rate monitoring can be determined.
In an optional implementation manner, the scene recognition processing is performed based on the motion state frame data, a current monitoring scene may be determined first, and the motion state frame data of a specified frame number may be determined as a target historical motion state frame; carrying out one-dimensional signal convolution processing on the target historical motion state frame, carrying out scene classification prediction, determining the current monitoring scene, and outputting a corresponding first prediction probability as a target confidence coefficient;
In another optional implementation manner, the scene recognition processing is performed based on the motion state frame data, and the target historical motion state frame may be input to a pre-trained decision tree model to perform scene classification prediction, determine the current monitored scene where the target historical motion state frame is located, and output the corresponding second prediction probability as the target confidence.
In one embodiment, fig. 2 shows a process of scene recognition based on ACC data, where ACC data is collected first, ACC belongs to time domain signal preprocessing, then frequency domain signals are obtained through fourier transform, time-frequency domain feature extraction is performed, the motion state of the current frame is determined, and the motion state of the historical N frames is used to realize scene recognition and input confidence.
In practical application, the historical state of the latest N frames can be directly subjected to one-dimensional signal convolution to realize classification prediction, and probability conditions are output as confidence references; the scene classification can also be completed directly by using a trained decision tree model, and the probability condition is output as a confidence coefficient reference.
Step S130, matching a target filter in a pre-configured filter bank according to the current monitoring scene, and filtering interference signals of the photoplethysmography wave signals of the wearable device through the target filter.
The filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene. Optionally, to improve coverage of the filter bank, the filter bank includes multiple types of filters and further includes a time domain filter and a frequency domain filter; each target monitoring scene corresponds to a corresponding scene time domain filter and a scene frequency domain filter.
In some embodiments, filtering the interference signal of the photoplethysmography signal of the wearable device by the target filter may include the following steps 3-1 to 3-4:
Step 3-1, acquiring a photoelectric volume pulse wave signal of the wearable device, and preprocessing the photoelectric volume pulse wave signal;
step 3-2, respectively determining a corresponding frequency domain heart rate filter and a corresponding time domain heart rate filter according to the motion state corresponding to the current frame and the current monitoring scene;
Step 3-3, performing first interference signal filtering and frequency domain conversion processing on the photoplethysmography wave signals of the wearable device through a frequency domain heart rate filter, and determining the filtered frequency domain signals;
and 3-4, performing second interference signal filtering on the photoplethysmography signals of the wearable equipment through a time domain heart rate filter, and determining the filtered time domain signals.
In some embodiments, referring to fig. 3, current frame PPG data is input, and the PPG data is subjected to time domain raw signal preprocessing. Filtering the frequency domain and the time domain respectively, namely selecting a frequency domain heart rate filter according to the motion state of the current frame and the scene recognition result, and performing Fourier transform on the PPG signal after filtering to obtain a frequency domain signal; and selecting a time domain heart rate filter according to the current frame motion state and the scene recognition result, filtering the PPG signal to obtain a time domain signal, and finally obtaining a filtered time domain signal and a filtered frequency domain signal so as to carry out the next processing.
When the actual measurement of the signal is considered, various factors influence the signal, and meanwhile, as the emphasis point of the time-frequency domain heart rate extraction is different, different filters can be selected through the filter bank according to the real-time scene recognition result, and different filters are adopted for the subsequent time domain and frequency domain processing respectively. The preset filter bank adopted by the application can realize more targeted pretreatment on complex daily scenes, and meanwhile, the computational complexity caused by self-adaptive filtering is avoided.
Step S140, carrying out dual-spectrum combined spectrum peak identification based on the filtered photoplethysmography data and the accelerometer data, and determining an estimated heart rate value.
In an actual heart rate monitoring scene, spectrum holes or spectrum dissipation of ACC and PPG signals can be caused by hand inching or irregular actions and the like, the identified main peak can shift, and the real main peak energy is dispersed to an adjacent area. Erroneous ACC peak identification may result in incorrect artifact cancellation during PPG peak processing, and erroneous PPG peak identification may result in a predicted heart rate value deviating from the true heart rate. Therefore, the embodiment of the application carries out the combined spectrum peak identification through the double spectrum combined spectrum peak identification, namely the PPG spectrum and the ACC spectrum, so that the main peak obtained by extraction, namely the heart rate value, is realized.
In some embodiments, the performing dual-spectrum joint spectrum peak identification based on the filtered photoplethysmography data and the accelerometer data to determine the estimated heart rate value may include the following steps 4-1 to 4-6 when implemented:
Step 4-1, determining a frequency spectrum tracking area according to the signal quality and the historical heart rate value; generally, the spectrum tracking area can refer to the historical heart rate value and in a preset range near the spectrum, and under the condition that the signal quality is extremely poor, a plurality of tracking areas are simultaneously set, so that the situation that the heart rate is jumped out of the conventional preset range and the heart rate value output cannot be tracked correctly is avoided. Wherein the multi-tracking area design references scene recognition results.
Step 4-2, respectively extracting a first frequency spectrum of the photoelectric volume pulse wave signal and a second frequency spectrum of the accelerometer signal in the frequency spectrum tracking area;
step 4-3, inputting a first spectrum of the photoelectric volume pulse wave signal and a second spectrum of the accelerometer signal in a specified time and frequency domain range into a convolutional neural network, and performing convolutional operation on the first spectrum of the photoelectric volume pulse wave signal and the second spectrum of the accelerometer signal in a specified tracking area range respectively; the convolutional neural network needs to supervise and learn with a large amount of historical data, and the problem of overfitting can be avoided because the task to be learned is clear and the input frequency spectrum is normalized.
And 4-4, inputting the convolution result into the joint convolution layer and the full connection layer in sequence to obtain a joint main peak identification result and identification confidence.
And 4-5, determining a frequency domain estimated heart rate value based on the combined main peak recognition result and the recognition confidence. Fig. 4 shows a flow of determining a frequency domain predicted heart rate in combination with a primary peak identification result.
And 4-6, determining a time domain estimated heart rate value by performing time domain ventricular beat interval calculation on the accelerometer data and the photoelectric volume pulse wave signals. The temporal ventricular beat intervals, i.e., temporal RR intervals, are not described in detail herein.
Alternatively, only the PPG may be identified by an independent spectral peak in a partial scene, i.e. only the PPG needs to be convolved, and then the main peak and the confidence level are directly output by using the full-connection layer.
In addition, the original time domain signal or the extracted time domain features can be added into the network to participate in training together.
According to the method, the double frequency spectrums are processed in the frequency spectrum tracking area, so that the main peak can be effectively identified, the influence of motion artifact components is reduced, the heart rate extraction error is reduced, and the heart rate accuracy is improved.
Step S150, optimizing the estimated heart rate value based on the scene type of the current monitoring scene, and determining a target heart rate monitoring value.
Considering that the heart rate monitoring modes under different scenes are various, such as active heart rate measurement, background irregular automatic monitoring, real-time heart rate measurement and the like, the heart rate monitoring modes are various, and scene changes caused by front-back dynamic changes are faced, such as running exercise started by a user, but the user can walk or rest, or just do some warm-up actions, and flexible scene recognition and switching can effectively assist heart rate extraction and post-processing. Therefore, in some embodiments, the optimizing the estimated heart rate value based on the scene type of the current monitored scene to determine the target heart rate monitoring value may include the following steps 5-1 to 5-3 when implemented:
Step 5-1, judging the scene type of the current monitored scene; the scene types comprise a stable scene type, a transition scene type and a switching scene type;
Step 5-2, determining a target heart rate monitoring value based on the frequency domain estimated heart rate value or the time domain estimated heart rate value when the scene is in the stable scene type;
And 5-3, determining the joint strength of the estimated heart rate value and the estimated heart rate value in the time domain based on a pre-trained heart rate reference decision model when the model is in the transition scene type or the switching scene type, and optimizing the estimated heart rate value through the joint strength to determine a target heart rate monitoring value.
In an example, a simple scenario is illustrated to describe a heart rate extraction and post-processing strategy, in an active heart rate measurement mode, after a user is in a static scenario with high probability, respectively completing time domain heart rate and frequency domain heart rate extraction, the processing strategies of the two heart rates can be dynamically adjusted through time-frequency domain signal quality and extracted time-frequency domain characteristic information, for example, under the conditions of frequency domain peak energy dispersion or harmonic energy highlighting, the time domain heart rate is more believed, and when the time domain signal quality is abnormal, the frequency domain heart rate is more believed.
In one embodiment, the reference selections of the time domain heart rate and the frequency domain heart rate may be selected by a heart rate reference decision model, for example, by which it may be determined what scenario the frequency domain heart rate is selected as the reference, what scenario the time domain heart rate is selected as the reference, and what scenario the frequency domain heart rate and the time domain heart rate are combined to determine the final target heart rate.
According to the embodiment of the application, the neural network model is trained in advance to perform reference selection of the time domain frequency domain heart rate, and the training steps of the heart rate reference decision model comprise: determining a training sample based on the signal quality of the acquired time domain signal and frequency domain signal, the joint main peak recognition result, the time domain feature, the frequency domain estimated heart rate value, the time domain estimated heart rate value and the historical heart rate condition; and training the initial heart rate reference decision model based on the training sample until convergence, and determining a trained heart rate reference decision model.
Through the heart rate reference decision model, comprehensive judgment can be performed in various heart rate monitoring modes, scene switching is performed in a period of monitoring duration, and time-frequency domain signal quality, combined main peak recognition conditions, characteristic conditions of time-frequency domain extraction, time-frequency domain heart rate extraction results, historical heart rate conditions and the like are jointly considered. The complex problems can be decomposed by scene processing, the heart rate extraction is differentially designed according to scenes, and a more accurate post-processing strategy can be adopted for the appointed scenes, so that the finally output heart rate value is more accurate.
According to the embodiment of the application, the scenes are identified in advance, and corresponding heart rate monitoring calculation can be performed aiming at different scenes, so that the heart rate monitoring value is more accurate. By pre-configuring the filter bank and adopting different filters, better data preprocessing can be achieved in a specified scene, such as the problem that confusion under daily heart rate extraction can be reduced due to the fact that abundant high-frequency components tend to be reserved in a frequency band. The double spectrums in the spectrum tracking area are processed through the neural network, so that a real main peak can be found under the condition of main peak energy dispersion or other abnormal conditions, heart rate extraction errors caused by confusion of the main peak are prevented, and meanwhile, the influence of motion artifact components can be reduced more effectively. The heart rate extraction and the heart rate post-processing are determined according to scene results, complex problems can be decomposed through scene division processing, differential design is carried out on the heart rate extraction according to scenes, and a more accurate processing strategy can be adopted on appointed scenes, so that the finally output heart rate value is more accurate.
In summary, the heart rate monitoring value estimation optimization method provided by the embodiment of the application can be self-adaptive to daily multiple scenes, and only ACC and PPG are needed according to scene switching processing strategies of a wearer, and excessively complex calculation is not needed; the method has the advantages that the factors influencing the heart rate value error under the daily complex scene are comprehensively considered, the universal heart rate scheme is optimized from two directions of scene deconstructing and time-frequency domain signal processing optimizing, the heart rate prediction accuracy performance can be effectively improved by the optimizing method, and the heart rate condition can be intelligently identified even under very poor signals.
Based on the above method embodiment, the embodiment of the present application further provides a device for optimizing heart rate monitor value estimation of a self-adaptive daily multi-scenario, as shown in fig. 5, the device mainly includes the following parts:
The data acquisition processing module 510 is configured to acquire accelerometer data of the wearable device, perform time-frequency domain processing and feature extraction on the accelerometer data, determine time domain features and frequency domain features, perform motion state identification through the time domain features and the frequency domain features, and determine motion state frame data;
The scene recognition module 520 is configured to perform scene recognition processing based on the motion state frame data, and determine a current monitored scene;
The filtering module 530 is configured to match a target filter in a pre-configured filter bank according to a current monitoring scene, and perform interference signal filtering on the photoplethysmography signal of the wearable device through the target filter; the filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene;
the combined spectrum peak recognition module 540 is configured to perform dual-spectrum combined spectrum peak recognition based on the filtered photoplethysmography data and the accelerometer data, and determine an estimated heart rate value;
the heart rate value optimizing module 550 is configured to optimize the estimated heart rate value based on the scene type of the current monitored scene, and determine the target heart rate monitoring value.
In a possible embodiment, the scene recognition module 520 is further configured to:
Determining motion state frame data of a designated frame number as a target historical motion state frame;
Carrying out one-dimensional signal convolution processing on the target historical motion state frame, carrying out scene classification prediction, determining the current monitoring scene, and outputting a corresponding first prediction probability as a target confidence coefficient;
And/or the number of the groups of groups,
And inputting the target historical motion state frame into a pre-trained decision tree model for scene classification prediction, determining the current monitoring scene, and outputting a corresponding second prediction probability as a target confidence coefficient.
In a possible implementation, the filter bank includes multiple classes of filters and further includes a time domain filter and a frequency domain filter; each target monitoring scene corresponds to a corresponding scene time domain filter and a scene frequency domain filter.
In a possible embodiment, the filtering module 530 is further configured to:
acquiring photoelectric volume pulse wave signals of the wearable equipment, and preprocessing the photoelectric volume pulse wave signals;
Respectively determining a corresponding frequency domain heart rate filter and a corresponding time domain heart rate filter according to the motion state corresponding to the current frame and the current monitoring scene;
The photoelectric volume pulse wave signals of the wearable equipment are subjected to first interference signal filtering and frequency domain conversion processing through a frequency domain heart rate filter, and filtered frequency domain signals are determined;
And performing second interference signal filtering on the photoplethysmography signal of the wearable device through a time domain heart rate filter, and determining a filtered time domain signal.
In a possible embodiment, the above-mentioned joint spectrum peak identifying module 540 is further configured to:
determining a frequency spectrum tracking area according to the signal quality and the historical heart rate value;
Respectively extracting a first frequency spectrum of a photoelectric volume pulse wave signal and a second frequency spectrum of an accelerometer signal in a frequency spectrum tracking area;
Inputting a first spectrum of the photoelectric volume pulse wave signal and a second spectrum of the accelerometer signal in a specified time and frequency domain range into a convolutional neural network, and performing convolutional operation on the first spectrum of the photoelectric volume pulse wave signal and the second spectrum of the accelerometer signal in a specified tracking area range respectively;
Sequentially inputting the convolution result into a joint convolution layer and a full connection layer to obtain a joint main peak identification result and identification confidence;
determining a frequency domain estimated heart rate value based on the combined main peak recognition result and the recognition confidence;
The time domain estimated heart rate value is determined by performing time domain ventricular beat interval calculation on the accelerometer data and the photoplethysmography wave signals.
In a possible embodiment, the heart rate value optimization module 550 is further configured to:
Judging the scene type of the current monitoring scene; the scene types comprise a stable scene type, a transition scene type and a switching scene type;
When the scene type is stable, determining a target heart rate monitoring value based on the frequency domain estimated heart rate value or the time domain estimated heart rate value;
When the heart rate monitoring system is in the transition scene type or the switching scene type, the joint strength of the estimated heart rate value and the time domain estimated heart rate value is determined based on a pre-trained heart rate reference decision model, and the estimated heart rate value is optimized through the joint strength, so that the target heart rate monitoring value is determined.
In a possible embodiment, the apparatus further includes a model training module configured to:
Determining a training sample based on the signal quality of the acquired time domain signal and frequency domain signal, the joint main peak recognition result, the time domain feature, the frequency domain estimated heart rate value, the time domain estimated heart rate value and the historical heart rate condition;
and training the initial heart rate reference decision model based on the training sample until convergence, and determining a trained heart rate reference decision model.
The implementation principle and the generated technical effects of the heart rate monitoring value estimation optimization device of the self-adaptive daily multi-scene provided by the embodiment of the application are the same as those of the embodiment of the method, and for the sake of brief description, the corresponding content in the embodiment of the heart rate monitoring value estimation optimization method of the self-adaptive daily multi-scene can be referred to.
The embodiment of the application also provides a wearable device, as shown in fig. 6, which is a schematic structural diagram of the wearable device, and the wearable device can be a smart watch for example. The wearable device 100 includes a processor 61 and a memory 60, where the memory 60 stores computer executable instructions executable by the processor 61, and the processor 61 executes the computer executable instructions to implement any of the above-mentioned methods for optimizing heart rate monitor value estimation in adaptive daily multi-scenario.
In the embodiment shown in fig. 6, the wearable device further comprises a bus 62 and a communication interface 63, wherein the processor 61, the communication interface 63 and the memory 60 are connected by the bus 62.
The memory 60 may include a high-speed random access memory (RAM, random Access Memory), and may further include a non-volatile memory (non-volatile memory), such as at least one disk memory. The communication connection between the system network element and at least one other network element is achieved via at least one communication interface 63 (which may be wired or wireless), and may use the internet, a wide area network, a local network, a metropolitan area network, etc. Bus 62 may be an ISA (Industry Standard Architecture ) bus, a PCI (PERIPHERAL COMPONENT INTERCONNECT, peripheral component interconnect standard) bus, or EISA (Extended Industry Standard Architecture ) bus, among others. The bus 62 may be classified as an address bus, a data bus, a control bus, or the like. For ease of illustration, only one bi-directional arrow is shown in FIG. 6, but not only one bus or type of bus.
The processor 61 may be an integrated circuit chip with signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuits of hardware in the processor 61 or by instructions in the form of software. The processor 61 may be a general-purpose processor, including a central processing unit (Central Processing Unit, CPU for short), a network processor (Network Processor, NP for short), and the like; but may also be a digital signal Processor (DIGITAL SIGNAL Processor, DSP), application Specific Integrated Circuit (ASIC), field-Programmable gate array (FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present application may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in the memory, and the processor 61 reads the information in the memory, and combines the hardware to complete the steps of the heart rate monitor value estimation optimization method of the adaptive daily multi-scenario in the foregoing embodiment.
The embodiment of the application also provides a computer readable storage medium, which stores computer executable instructions that, when being called and executed by a processor, cause the processor to implement the above-mentioned heart rate monitoring value estimation optimization method of adaptive daily multi-scenario, and the specific implementation can be found in the foregoing method embodiment, which is not described herein.
The method for optimizing heart rate monitoring value estimation of self-adaptive daily multi-scenario and the computer program product of the wearable device provided by the embodiment of the application comprise a computer readable storage medium storing program codes, wherein the instructions included in the program codes can be used for executing the method described in the method embodiment, and specific implementation can be seen in the method embodiment and will not be repeated here.
The relative steps, numerical expressions and numerical values of the components and steps set forth in these embodiments do not limit the scope of the present application unless it is specifically stated otherwise.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a non-volatile computer readable storage medium executable by a processor. Based on this understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present application. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In the description of the present application, it should be noted that the terms "first," "second," and the like are used merely to distinguish between descriptions and should not be construed as indicating or implying relative importance.
In the description of the present application, it should also be noted that, unless explicitly specified and limited otherwise, the terms "disposed," "mounted," "connected," and "connected" are to be construed broadly, and may be, for example, fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; can be directly connected or indirectly connected through an intermediate medium, and can be communication between two elements. The specific meaning of the above terms in the present application will be understood in specific cases by those of ordinary skill in the art.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and not for limiting the same; although the application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the application.
Claims (10)
1. The heart rate monitoring value estimation optimization method of the self-adaptive daily multi-scene is characterized by comprising the following steps of:
Collecting accelerometer data of a wearable device, performing time-frequency domain processing and feature extraction on the accelerometer data, determining time domain features and frequency domain features, performing motion state identification through the time domain features and the frequency domain features, and determining motion state frame data;
performing scene recognition processing based on the motion state frame data to determine a current monitoring scene;
Matching a target filter in a pre-configured filter bank according to the current monitoring scene, and filtering interference signals of photoelectric volume pulse wave signals of the wearable equipment through the target filter; the filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene;
performing dual-spectrum combined spectrum peak identification based on the filtered photoplethysmogram data and the accelerometer data, and determining an estimated heart rate value;
and optimizing the estimated heart rate value based on the scene type of the current monitoring scene, and determining a target heart rate monitoring value.
2. The method for optimizing heart rate monitor value estimation for adaptive daily multi-scene as recited in claim 1, wherein performing scene recognition processing based on the motion state frame data, determining a current monitored scene comprises:
Determining the motion state frame data of a designated frame number as a target historical motion state frame;
Carrying out one-dimensional signal convolution processing on the target historical motion state frame, carrying out scene classification prediction, determining the current monitoring scene, and outputting a corresponding first prediction probability as a target confidence coefficient;
And/or the number of the groups of groups,
And inputting the target historical motion state frame into a pre-trained decision tree model for scene classification prediction, determining the current monitoring scene, and outputting a corresponding second prediction probability as a target confidence coefficient.
3. The method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario according to claim 1, wherein the filter bank comprises a plurality of classes of filters further comprising a time domain filter and a frequency domain filter; each target monitoring scene corresponds to a corresponding scene time domain filter and a scene frequency domain filter.
4. The method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario according to claim 1, wherein performing interference signal filtering on the photoplethysmography signal of the wearable device by the target filter comprises:
acquiring a photoplethysmogram signal of the wearable device, and preprocessing the photoplethysmogram signal;
Respectively determining a corresponding frequency domain heart rate filter and a corresponding time domain heart rate filter according to the motion state corresponding to the current frame and the current monitoring scene;
The photoelectric volume pulse wave signals of the wearable equipment are subjected to first interference signal filtering and frequency domain conversion processing through the frequency domain heart rate filter, and filtered frequency domain signals are determined;
And performing second interference signal filtering on the photoplethysmography signal of the wearable device through the time domain heart rate filter to determine a filtered time domain signal.
5. The method of claim 1, wherein determining the estimated heart rate value based on the filtered photoplethysmography pulse wave data and the accelerometer data by performing bispectral joint spectral peak identification comprises:
determining a frequency spectrum tracking area according to the signal quality and the historical heart rate value;
Respectively extracting a first frequency spectrum of a photoelectric volume pulse wave signal and a second frequency spectrum of an accelerometer signal in the frequency spectrum tracking area;
Inputting the first spectrum of the photoplethysmogram signal and the second spectrum of the accelerometer signal in a specified time and frequency domain range into a convolutional neural network, and respectively carrying out convolutional operation on the first spectrum of the photoplethysmogram signal and the second spectrum of the accelerometer signal in a specified tracking area range;
Sequentially inputting the convolution result into a joint convolution layer and a full connection layer to obtain a joint main peak identification result and identification confidence;
Determining a frequency domain estimated heart rate value based on the combined main peak recognition result and the recognition confidence;
And determining a time domain estimated heart rate value by performing time domain ventricular beat interval calculation on the accelerometer data and the photoplethysmography wave signals.
6. The method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario of claim 5, wherein optimizing the estimated heart rate value based on the scenario type of the current monitored scenario, determining a target heart rate monitor value, comprises:
Judging the scene type of the current monitoring scene; the scene types comprise a stable scene type, a transition scene type and a switching scene type;
Determining a target heart rate monitoring value based on the frequency domain estimated heart rate value or the time domain estimated heart rate value when the stable scene type exists;
When the transition scene type or the switching scene type is adopted, the joint strength of the estimated heart rate value and the time domain estimated heart rate value is determined based on a pre-trained heart rate reference decision model, and the estimated heart rate value is optimized through the joint strength, so that a target heart rate monitoring value is determined.
7. The method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario of claim 6, further comprising:
Determining a training sample based on the signal quality of the acquired time domain signal and frequency domain signal, the joint main peak recognition result, the time domain feature, the frequency domain estimated heart rate value, the time domain estimated heart rate value and the historical heart rate condition;
and training the initial heart rate reference decision model based on the training sample until convergence, and determining a trained heart rate reference decision model.
8. An adaptive daily multi-scene heart rate monitor value estimation optimizing device, which is characterized by comprising:
the data acquisition processing module is used for acquiring accelerometer data of the wearable equipment, carrying out time-frequency domain processing and feature extraction on the accelerometer data, determining time domain features and frequency domain features, and carrying out motion state identification through the time domain features and the frequency domain features to determine motion state frame data;
The scene recognition module is used for performing scene recognition processing based on the motion state frame data and determining a current monitoring scene;
The filtering module is used for matching a target filter in a pre-configured filter bank according to the current monitoring scene, and carrying out interference signal filtering on the photoplethysmography wave signals of the wearable equipment through the target filter; the filter bank comprises a plurality of types of filters, and each type of filter corresponds to a corresponding target monitoring scene;
the combined spectrum peak identification module is used for carrying out dual-spectrum combined spectrum peak identification based on the filtered photoplethysmogram data and the accelerometer data, and determining an estimated heart rate value;
and the heart rate value optimization module is used for optimizing the estimated heart rate value based on the scene type of the current monitoring scene and determining a target heart rate monitoring value.
9. A wearable device comprising a processor and a memory, the memory storing computer-executable instructions executable by the processor to implement the method of optimizing heart rate monitor value estimation for adaptive daily multi-scenario of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which, when invoked and executed by a processor, cause the processor to implement the method for optimizing heart rate monitor value estimation for adaptive daily multi-scenario according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410572262.9A CN118141354B (en) | 2024-05-10 | 2024-05-10 | Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410572262.9A CN118141354B (en) | 2024-05-10 | 2024-05-10 | Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN118141354A true CN118141354A (en) | 2024-06-07 |
| CN118141354B CN118141354B (en) | 2024-07-02 |
Family
ID=91289010
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410572262.9A Active CN118141354B (en) | 2024-05-10 | 2024-05-10 | Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118141354B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107708538A (en) * | 2015-06-16 | 2018-02-16 | 高通股份有限公司 | Sane heart rate estimation |
| US20190192082A1 (en) * | 2016-09-06 | 2019-06-27 | Soongsil University Research Consortium Techno-Park | Apparatus for eliminating motion artifacts by using ppg signal and method thereof |
| CN115886744A (en) * | 2022-11-09 | 2023-04-04 | 广东粤港澳大湾区黄埔材料研究院 | Motion pulse wave denoising method and device |
| CN117179751A (en) * | 2023-08-30 | 2023-12-08 | 芯海科技（深圳）股份有限公司 | Signal processing method, device, electronic equipment and storage medium |
-
2024
- 2024-05-10 CN CN202410572262.9A patent/CN118141354B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107708538A (en) * | 2015-06-16 | 2018-02-16 | 高通股份有限公司 | Sane heart rate estimation |
| US20190192082A1 (en) * | 2016-09-06 | 2019-06-27 | Soongsil University Research Consortium Techno-Park | Apparatus for eliminating motion artifacts by using ppg signal and method thereof |
| CN115886744A (en) * | 2022-11-09 | 2023-04-04 | 广东粤港澳大湾区黄埔材料研究院 | Motion pulse wave denoising method and device |
| CN117179751A (en) * | 2023-08-30 | 2023-12-08 | 芯海科技（深圳）股份有限公司 | Signal processing method, device, electronic equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN118141354B (en) | 2024-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP3558101B1 (en) | Methods and systems for determining abnormal cardiac activity | |
| Sundararaj | An efficient threshold prediction scheme for wavelet based ECG signal noise reduction using variable step size firefly algorithm | |
| EP3626167B1 (en) | A method of generating a model for heart rate estimation from a photoplethysmography signal and a method and a device for heart rate estimation | |
| US20160223514A1 (en) | Method for denoising and data fusion of biophysiological rate features into a single rate estimate | |
| US20190192079A1 (en) | System and a method for motion artifact reduction in a ppg signal | |
| CN111588367A (en) | Heart rate detection method and device and computer readable storage medium | |
| Sai | A review on arrhythmia classification using ECG signals | |
| WO2023015932A1 (en) | Deep learning-based heart rate measurement method and wearable device | |
| CN111832537B (en) | Abnormal electrocardiosignal identification method and abnormal electrocardiosignal identification device | |
| Kumar et al. | STSR: spectro-temporal super-resolution analysis of a reference signal less photoplethysmogram for heart rate estimation during physical activity | |
| CN118141354B (en) | Self-adaptive daily multi-scene heart rate monitoring value estimation optimization method and wearable device | |
| Karri et al. | A real-time cardiac arrhythmia classification using hybrid combination of delta modulation, 1D-CNN and blended LSTM | |
| Zheng et al. | Tiny-PPG: A lightweight deep neural network for real-time detection of motion artifacts in photoplethysmogram signals on edge devices | |
| Hosni et al. | Remote real-time heart rate monitoring with recursive motion artifact removal using PPG signals from a smartphone camera | |
| CA3139034A1 (en) | System and method for filtering time-varying data for physiological signal prediction | |
| CN111461206A (en) | Electroencephalogram-based fatigue detection method with steering wheel embedded with electroencephalogram sensor | |
| Choi et al. | Online SVM-based personalizing method for the drowsiness detection of drivers | |
| CN115105084A (en) | Heart beat positioning and classifying method, system, medium and equipment based on deep learning | |
| CN113499064A (en) | Wi-Fi perception human body tumbling detection method and system in bathroom scene | |
| Lutin et al. | Learning based quality indicator aiding heart rate estimation in wrist-worn PPG | |
| CN116035548B (en) | Method and device for detecting heart rhythm state, intelligent wearable device and storage medium | |
| CN111128387A (en) | Method and equipment for establishing epileptic seizure detection model | |
| CN116841401B (en) | Health monitoring method based on mouse, monitoring unit, mouse and storage medium | |
| Rylo et al. | Gesture recognition of wrist motion based on wearables sensors | |
| KR102593035B1 (en) | Method and apparatus for diagnosing epilepsy based on deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |