Pressure monitoring method and device based on wearable equipment and electronic equipment

Abstract

The invention discloses a pressure monitoring method and device based on wearable equipment and electronic equipment. The method comprises the following steps: acquiring current photoelectric plethysmography (PPG) signals corresponding to a plurality of channels collected by wearable equipment; based on the current PPG signals corresponding to the multiple channels, candidate pulse wave sequences corresponding to the multiple channels are determined, and the candidate pulse wave sequences comprise multiple pulse waves of the corresponding channels; determining a target pulse wave sequence from the candidate pulse wave sequences corresponding to the multiple channels based on the pulse wave continuity evaluation results and the pulse wave quality evaluation results corresponding to the multiple channels; performing feature extraction on the target pulse wave sequence to obtain current heart rate variability features; and determining a current pressure monitoring result according to the current heart rate variability characteristic. According to the invention, the technical problem of low pressure monitoring accuracy and reliability of wearable equipment in a dynamic environment in the prior art is solved.

Description

Technical Field

[0001] This invention relates to the field of wearable technology, and more specifically, to a pressure monitoring method, apparatus, and electronic device based on wearable devices. Background Technology

[0002] In the current wearable technology field, all-weather stress monitoring has become an important part of user health management. Wearable devices in this field generally use photoplethysmography (PPG) signals to assess heart rate variability (HRV) and then estimate stress levels. However, these technologies face significant challenges in terms of accuracy and reliability under dynamic environments, such as when users are exercising or under varying lighting conditions. For example, because PPG signal quality is affected by various factors, such as exercise, changes in ambient light, and wearing tightness, the assessment of signal quality in these technologies is often limited to the global signal-to-noise ratio (SNR) of the entire data stream, lacking a detailed consideration of the quality of each individual pulse wave. This makes HRV feature extraction more difficult when signal quality is unstable or abnormal, thus affecting the accuracy of stress monitoring. The determination of motion status largely relies on the overall acceleration (ACC) level, failing to effectively distinguish between short-term, severe shaking and small-amplitude movements. This results in a large amount of data being excluded due to motion interference, reducing the effective data volume for stress monitoring and affecting the continuity and reliability of stress values. On-device storage and computing resources are limited, while HRV feature extraction and stress calculation models in related technologies have high resource requirements, making it difficult to run them in real-time and efficiently on wearable devices. This limits the real-time performance and individual user adaptability of stress monitoring. In summary, related technologies suffer from low accuracy and reliability in stress monitoring by wearable devices under dynamic environments.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a pressure monitoring method, device, and electronic device based on wearable devices, to at least solve the technical problem of low accuracy and reliability of pressure monitoring by wearable devices in dynamic environments in related technologies.

[0005] According to one aspect of the present invention, a stress monitoring method based on a wearable device is provided, comprising: acquiring current photoplethysmography (PPG) signals corresponding to multiple channels collected by the wearable device, wherein the current PPG signals corresponding to the multiple channels are collected for a target object during the current time period; determining candidate pulse wave sequences corresponding to multiple channels based on the current PPG signals corresponding to the multiple channels, wherein the candidate pulse wave sequences include multiple pulse waves of the corresponding channels; determining a target pulse wave sequence from the candidate pulse wave sequences corresponding to the multiple channels based on pulse wave continuity evaluation results and pulse wave quality evaluation results, wherein the pulse wave continuity evaluation results are used to indicate the continuity of pulse waves in the corresponding candidate pulse wave sequences, and the pulse wave quality evaluation results are used to indicate the template matching degree of the multiple pulse waves included in the corresponding candidate pulse wave sequences, and the template matching degree is used to indicate the matching degree between the corresponding pulse wave and the template waveform; performing feature extraction on the target pulse wave sequence to obtain current heart rate variability features; and determining the current stress monitoring result based on the current heart rate variability features.

[0006] According to another aspect of the present invention, a pressure monitoring device based on a wearable device is also provided, comprising: a PPG signal acquisition module, configured to acquire current photoplethysmography (PPG) signals corresponding to multiple channels collected by the wearable device, wherein the current PPG signals corresponding to the multiple channels are acquired for a target object during the current time period; a candidate pulse wave sequence determination module, configured to determine candidate pulse wave sequences corresponding to multiple channels based on the current PPG signals corresponding to the multiple channels, wherein the candidate pulse wave sequences include multiple pulse waves of the corresponding channels; and a target pulse wave sequence determination module, configured to determine the target pulse wave sequence based on the current PPG signals corresponding to the multiple channels. The system uses pulse wave continuity assessment results and pulse wave quality assessment results to determine the target pulse wave sequence from the candidate pulse wave sequences corresponding to multiple channels. The pulse wave continuity assessment result indicates the continuity of the pulse waves in the corresponding candidate pulse wave sequence, while the pulse wave quality assessment result indicates the template matching degree of each pulse wave included in the corresponding candidate pulse wave sequence. The template matching degree indicates the matching degree between the corresponding pulse wave and the template waveform. A feature extraction module is used to extract features from the target pulse wave sequence to obtain the current heart rate variability features. A pressure monitoring module is used to determine the current pressure monitoring result based on the current heart rate variability features.

[0007] According to another aspect of the present invention, a non-volatile storage medium is also provided, which stores a plurality of instructions adapted for loading and executing any one of them by a processor in a pressure monitoring method based on a wearable device.

[0008] According to another aspect of the present invention, an electronic device is also provided, including one or more processors and a memory, the memory being used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement any one of the stress monitoring methods based on wearable devices.

[0009] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps of any one of the stress monitoring methods based on a wearable device.

[0010] In this embodiment of the invention, the current photoplethysmography (PPG) signals corresponding to multiple channels are acquired by a wearable device. These current PPG signals are acquired from the target object during the current time period. Based on the current PPG signals, candidate pulse wave sequences are determined for each channel, each candidate pulse wave sequence including multiple pulse waves from the corresponding channel. Based on the pulse wave continuity evaluation results and pulse wave quality evaluation results, a target pulse wave sequence is determined from the candidate pulse wave sequences. The pulse wave continuity evaluation results indicate the continuity of the pulse waves in the corresponding candidate pulse wave sequence, and the pulse wave quality evaluation results indicate... The template matching degree corresponding to each of the multiple pulse waves included in the candidate pulse wave sequence is used to indicate the matching degree between the corresponding pulse wave and the template waveform. Feature extraction is performed on the target pulse wave sequence to obtain the current heart rate variability features. Based on the current heart rate variability features, the current pressure monitoring result is determined. This achieves the goal of analyzing the continuity and quality of multi-channel pulse wave signals and selecting the best signal for HRV feature extraction. It enables accurate monitoring and feedback of the pressure level of the monitored object even under motion or signal interference conditions. This improves the technical effect of enhancing the accuracy and reliability of pressure monitoring of wearable devices in dynamic environments, and solves the technical problem of low accuracy and reliability of pressure monitoring of wearable devices in dynamic environments in related technologies. Attached Figure Description

[0011] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0012] Figure 1 This is a flowchart of a pressure monitoring method based on a wearable device according to an embodiment of the present invention;

[0013] Figure 2 This is a schematic diagram of an optional pulse wave sequence fusion according to an embodiment of the present invention;

[0014] Figure 3 This is a flowchart of an optional pressure monitoring method based on a wearable device according to an embodiment of the present invention;

[0015] Figure 4 This is a schematic diagram of a pressure monitoring device based on a wearable device according to an embodiment of the present invention. Detailed Implementation

[0016] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0017] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0018] First, to facilitate understanding of the embodiments of the present invention, some terms or nouns involved in the present invention will be explained below:

[0019] Numerical Python (Numpy) is an open-source library for numerical computation in Python. Numpy provides high-performance multidimensional array objects and tools for manipulating arrays, making it a fundamental library for scientific computing, data processing, and analysis.

[0020] Scientific Python (SciPy), built on NumPy, provides a suite of advanced functions for scientific computing, including optimization, linear algebra, integration, interpolation, Fourier transform, and statistics. SciPy is an important tool for solving complex mathematical calculations in scientific and engineering problems.

[0021] Neurokit2 is a Python library specifically designed for physiological signal analysis. It provides a suite of tools and functions for processing physiological signals such as electrocardiograms (ECGs), electroencephalograms (EEGs), and eye movements, particularly for extracting and analyzing features like heart rate variability. Neurokit2 aims to simplify the physiological signal analysis process, allowing researchers and developers to focus more on data interpretation and application rather than spending significant time on signal preprocessing and feature extraction.

[0022] According to an embodiment of the present invention, a method embodiment for pressure monitoring based on a wearable device is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0023] Figure 1 This is a flowchart of a pressure monitoring method based on a wearable device according to an embodiment of the present invention, such as... Figure 1 As shown, the method includes the following steps:

[0024] Step S102: Obtain the current photoplethysmography (PPG) signals corresponding to each of the multiple channels collected by the wearable device, wherein the current PPG signals corresponding to each of the multiple channels are collected for the target object in the current time period.

[0025] The entity executing steps S102 to S110 can be a wearable device. In this step, the wearable device can simultaneously collect data from multiple physiological signal acquisition channels through its built-in PPG sensor. Each channel collects PPG signals from the target object (i.e., the wearer) during the current monitoring period. These multiple channels can be different sensor locations on the wearable device, or different operating modes of the same sensor location (such as different wavelengths of light), to obtain more comprehensive PPG signals.

[0026] Step S104: Based on the current PPG signals corresponding to each of the multiple channels, determine the candidate pulse wave sequences corresponding to each of the multiple channels, wherein the candidate pulse wave sequences include multiple pulse waves of the corresponding channels.

[0027] This step, based on PPG signals acquired from multiple channels, uses signal processing algorithms (such as filtering and peak detection) to determine the candidate pulse wave sequence for each channel. A pulse wave sequence refers to a series of pulse wave signals captured by a PPG sensor during a continuous heartbeat cycle. Multiple pulse waves correspond one-to-one with multiple sampling times; that is, multiple pulse waves correspond to different sampling times. The validity of each pulse wave included in the candidate pulse wave sequence for each channel is unknown, and it may include some invalid pulse waves. This step provides a basis for subsequent quality assessment and selection of the target sequence pulse wave.

[0028] In one optional embodiment, based on the current PPG signals corresponding to each of the multiple channels, candidate pulse wave sequences corresponding to each of the multiple channels are determined, including: performing pulse wave decomposition on the current PPG signals corresponding to each of the multiple channels to obtain initial pulse wave sequences corresponding to each of the multiple channels; identifying abnormal fluctuation segments in the initial pulse wave sequences corresponding to each of the multiple channels based on the current acceleration signal collected by the wearable device in the current time period, wherein the abnormal fluctuation segment is a segment between peaks where the relative fluctuation intensity of the acceleration signal is greater than a preset intensity threshold, and the segment between peaks is a pulse wave segment between the peaks corresponding to two pulse waves; removing the pulse wave corresponding to the end peak of the abnormal fluctuation segment in the initial pulse wave sequences corresponding to each of the multiple channels to obtain candidate pulse wave sequences corresponding to each of the multiple channels, wherein the end peak is the peak corresponding to the end position of the abnormal fluctuation segment.

[0029] Optionally, firstly, the PPG signals acquired in real time from each channel are preprocessed and subjected to waveform analysis to identify and segment a series of pulse waves. These pulse waves constitute the initial pulse wave sequence for each channel. Pulse wave identification and segmentation can be based on peak-valley detection, where the peak (wave crest) represents the point of maximum blood volume change caused by each heartbeat. Each pulse wave is a signal segment within a predetermined neighborhood of the corresponding peak; for example, 0.4N can be taken before the pulse wave peak and 0.6N after it to obtain each independent pulse wave. Simultaneously, acceleration signals (ACC) from the wearable device are acquired to assess the wearer's (i.e., the target object's) motion state during the signal acquisition period. By comparing the relative fluctuation intensity of the acceleration signal (such as the amplitude of acceleration value change) with a preset intensity threshold, it is possible to identify whether there are abnormal signal fluctuations in the segment between two pulse wave peaks due to strenuous exercise or other dynamic factors. These abnormal segments may affect the quality of the PPG signal and the accuracy of the pulse waves. After identifying abnormal fluctuation segments, pulse waves located at the end of these abnormal segments are removed from the initial pulse wave sequence of each channel. This is because the ending peak of abnormal segments may be affected by motion or other fluctuations, causing them to inaccurately reflect the heart's pulsating characteristics. Removing these pulse waves results in a more accurate sequence, better suited for subsequent HRV feature extraction. After these two steps, the influence of abnormal fluctuation segments is removed from the initial pulse wave sequence of each channel, resulting in candidate pulse wave sequences. These sequences will be used for further evaluation to determine the final target pulse wave sequences for calculating HRV features and the stress index.

[0030] This embodiment of the method, when determining candidate pulse wave sequences, not only separates the pulse wave from the PPG signal, but also combines acceleration signals to identify and eliminate pulse waves interfered with by vigorous exercise or other dynamic factors. This generates a purer and more reliable pulse wave sequence for subsequent HRV feature extraction and stress assessment. This method improves data quality in dynamic environments, ensuring that the extracted HRV features and calculated stress index remain highly accurate even when the wearer is exercising or active.

[0031] In one optional embodiment, pulse wave decomposition is performed on the current PPG signals corresponding to each of the multiple channels to obtain initial pulse wave sequences corresponding to each of the multiple channels. This includes: identifying amplitude aberration segments in the current PPG signals corresponding to each of the multiple channels, wherein the amplitude aberration segments include aberration segments in the current PPG signals whose amplitude is greater than a first preset amplitude, and aberration segments in the current PPG signals whose consecutive predetermined number of difference values ​​are all less than a second preset amplitude, the difference values ​​being obtained by performing a differential operation on the corresponding current PPG signals; replacing the amplitude aberration segments in the current PPG signals corresponding to each of the multiple channels with predetermined values ​​to obtain preprocessed PPG signals corresponding to each of the multiple channels; and performing pulse wave decomposition on the preprocessed PPG signals corresponding to each of the multiple channels based on the peaks of the preprocessed PPG signals corresponding to each of the multiple channels to obtain candidate pulse wave sequences corresponding to each of the multiple channels, wherein each pulse wave is a signal segment within a predetermined neighborhood range of the corresponding peak.

[0032] Optionally, within the current PPG signal of each channel, segments with amplitudes exceeding the normal range are first identified. These abnormal segments may be caused by poor sensor contact, external light interference, or significant body movement. Identification includes two main categories of anomalies: when the amplitude exceeds a first preset amplitude (e.g., th1), it is marked as an amplitude overshoot anomaly. The first preset amplitude can be set based on the statistical characteristics of the current PPG signal (e.g., the mean plus a few (e.g., 3) standard deviations). When the signal changes very slowly over a continuous period, i.e., a predetermined number of consecutive difference values ​​(signal rate of change) are all less than a second preset amplitude, this segment is marked as an amplitude undershoot anomaly. This anomaly may stem from a weak signal or sensor malfunction. Once segments with abnormal amplitudes are identified, their values ​​are replaced with a predetermined value, such as 0 or other padding values, to eliminate the impact of the abnormal signal on subsequent processing and ensure signal continuity and consistency. For example, for the current PPG signal acquired in the current time period, amplitude anomalies are judged point by point. If the amplitude exceeds the threshold th1, it is considered that there is an amplitude anomaly at the current point. Here, th1 can be obtained by adding 3 times the standard deviation to the mean amplitude of the current PPG signal. A differential operation is performed on the current PPG signal. If n1 consecutive differential values ​​are less than th2, it is considered that there is an amplitude anomaly. Here, th2 is obtained by adding 0.01 times the mean differential value of the current PPG signal. The positions of the two amplitude anomalies are recorded, and all amplitude anomaly segments are set to 0, resulting in the preprocessed PPG signal PPG1. The preprocessed PPG signal (i.e., the preprocessed PPG signal) will be used for pulse wave decomposition. That is, based on the peak position of the preprocessed PPG signal, the signal segment near each peak is defined as a pulse wave. The above process is applied to all acquisition channels, extracting a series of pulse waves from the preprocessed PPG signal of each channel and combining them into the initial pulse wave sequence of that channel. This is the basis for subsequent fine signal quality assessment based on template waveform matching and motion intensity.

[0033] This embodiment focuses on key steps in signal preprocessing and initial pulse wave screening. It aims to obtain a cleaner pulse wave sequence by eliminating the influence of amplitude aberration segments and using precise peak-based segmentation, for subsequent quality assessment and feature extraction. This method effectively improves the reliability of PPG signals collected by wearable devices and the accuracy of processing results. Especially in complex and dynamic wearing environments, it better ensures the quality of the pulse wave sequence, providing a solid data foundation for accurate stress assessment.

[0034] Step S106: Based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, a target pulse wave sequence is determined from the candidate pulse wave sequences corresponding to each of the multiple channels. The pulse wave continuity evaluation results are used to indicate the continuity of the pulse waves in the corresponding candidate pulse wave sequence, and the pulse wave quality evaluation results are used to indicate the template matching degree of each of the multiple pulse waves included in the corresponding candidate pulse wave sequence. The template matching degree is used to indicate the matching degree between the corresponding pulse wave and the template waveform.

[0035] Optionally, the continuity and quality of candidate pulse wave sequences for each channel are first evaluated. Continuity evaluation ensures that the pulse waves in the sequence are continuous and conform to physiological characteristics, while quality evaluation uses template matching to determine the similarity between each pulse wave and a predefined template waveform. Based on the continuity and quality evaluation results, the optimal pulse wave sequence is selected from all candidate sequences as the target pulse wave sequence for subsequent heart rate variability (HRV) feature extraction. Template matching assesses pulse wave quality by calculating the similarity between the pulse wave and the template waveform; this can be achieved using, but is not limited to, correlation coefficients or distance metrics. A high matching degree indicates a good pulse wave morphology and minimal interference.

[0036] In an optional embodiment, before determining the target pulse wave sequence from the candidate pulse wave sequences corresponding to the multiple channels based on the pulse wave continuity assessment results and pulse wave quality assessment results corresponding to the multiple channels, the method further includes: marking the pulse wave position corresponding to the amplitude abnormal segment in the candidate pulse wave sequence corresponding to the multiple channels as a discontinuous position, and marking the pulse wave position corresponding to the end peak of the fluctuation abnormal segment as a discontinuous position, thereby obtaining the pulse wave continuity assessment results corresponding to the multiple channels.

[0037] Optionally, before generating candidate pulse wave sequences, the continuity of pulse wave positions related to amplitude and fluctuation abnormal segments in the candidate pulse wave sequences corresponding to multiple channels is evaluated. Specifically, pulse waves corresponding to amplitude abnormal segments and pulse waves corresponding to the end peaks of fluctuation abnormal segments are marked as discontinuous positions, indicating that the quality of these pulse waves may be disturbed and unsuitable as the basis for subsequent HRV feature extraction and stress assessment. After the above marking, the continuity evaluation results of pulse waves corresponding to multiple channels are obtained. These results include which pulse waves are continuous and reliable, and which are marked as abnormal or discontinuous. This information is crucial for the next step of screening target pulse wave sequences, helping to identify and select high-quality pulse wave sequences, thereby improving the accuracy and stability of HRV features.

[0038] This embodiment further refines the process of evaluating the continuity of pulse wave sequences. By marking pulse wave locations affected by anomalies as discontinuous, the quality of the pulse wave sequence in each channel can be evaluated more accurately, providing a more reliable data foundation for subsequent determination of the target pulse wave sequence. This preprocessing and labeling process helps improve the accuracy of signal quality assessment, ensuring the accuracy and effectiveness of HRV feature calculation and stress assessment results even under the complex and dynamic wearing environments that wearable devices may encounter.

[0039] In an optional embodiment, before determining the target pulse wave sequence from the candidate pulse wave sequences corresponding to multiple channels based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, the method further includes: obtaining the pulse wave quality evaluation result of any channel among the multiple channels by performing a point-by-point averaging operation on multiple pulse waves in the candidate pulse wave sequence of any channel to obtain the template waveform corresponding to any channel, wherein the point-by-point averaging operation is used to calculate the average value of the values ​​corresponding to the same point in the multiple pulse waves; calculating the template matching degree between the multiple pulse waves in the candidate pulse wave sequence of any channel and the template waveform respectively to obtain the pulse wave quality evaluation result of any channel; and obtaining the pulse wave quality evaluation results corresponding to each of the multiple channels by using the method of obtaining the pulse wave quality evaluation result of any channel.

[0040] Optionally, for any one of the multiple channels, multiple pulse waves are first extracted from the candidate pulse wave sequence corresponding to that channel. Then, a point-by-point averaging operation is performed on these pulse waves, that is, the average value of all pulse waves at the same time point is calculated, finally obtaining a template waveform representing the average pulse wave morphology of that channel. The generation of the template waveform is based on the statistical analysis of multiple pulse waves, which can reflect the morphological characteristics of typical pulse waves and provide an objective reference standard for subsequent quality assessment. Subsequently, the template matching degree of each pulse wave in the candidate pulse wave sequence of any of the above channels is calculated. The template matching degree calculation can be based on similarity measures, such as correlation coefficient, dynamic time warping (DTW), or Euclidean distance, reflecting the degree of morphological similarity between each pulse wave and the template waveform. The pulse wave quality assessment result is the set of these template matching degree values, which are directly related to the quality of the pulse wave. A pulse wave with a high template matching degree represents a good signal quality, while a low template matching degree may indicate that the signal is interfered with or the data is abnormal. For example, a template waveform is generated by extracting each pulse wave based on its peak position. The similarity between the pulse wave and the template waveform is calculated. If the template matching degree is greater than th3 (the corresponding value range can be set between 0.7 and 1), it is considered a high-quality pulse wave. The template matching degree of all pulse waves is retained. The template waveform is continuously retained and gradually updated with newly acquired high-quality pulse waves. The template waveform is obtained as follows: After extracting the peak position of the pulse wave, the average heart rate is calculated based on the data length and the number of pulse waves. The average number of points N corresponding to each pulse wave can be calculated based on the sampling rate. Taking 0.4N forward from the peak position and 0.6N backward, each independent pulse wave is obtained. Then, the average value of each pulse wave is calculated point by point to obtain the template waveform. It should be noted that for each pressure measurement, the data is streamed, and only n seconds of data are input at a time. n seconds of data can generate a template waveform tempo_1. After another n seconds of data, the previous template waveform will be updated with a new pulse wave. Using the aforementioned logic for calculating the matching degree between pulse waves and template waveforms, the quality of candidate pulse wave sequences for all channels is evaluated. This means that a pulse wave quality evaluation result is generated for each channel, containing quality information for each pulse wave in that channel, expressed in the form of template matching degree.

[0041] Through the above steps, the quality of the pulse wave in each channel's pulse wave sequence can be objectively evaluated, thus providing data support for subsequently determining the optimal target pulse wave sequence from candidate sequences across multiple channels. Template matching degree is not only a quantitative indicator of pulse wave quality but also the foundation of multi-channel fusion strategies. It helps select the optimal signal source when signal quality is inconsistent, ensuring the accuracy of HRV feature extraction and stress assessment.

[0042] In one optional embodiment, based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, a target pulse wave sequence is determined from the candidate pulse wave sequences corresponding to each of the multiple channels. This includes: determining effective pulse wave segments in the candidate pulse wave sequences corresponding to each of the multiple channels based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, and marking other pulse waves besides the effective pulse wave segments as discontinuous positions to obtain optimized pulse wave sequences corresponding to each of the multiple channels. The effective pulse wave segments include high-quality pulse waves with a template matching degree greater than a matching degree threshold, and pulse wave segments composed of a specified number or more undetermined pulse waves. The undetermined pulse waves are pulse waves with a template matching degree within a preset threshold range, where the upper limit of the preset threshold range is the matching degree threshold. Based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels, a target pulse wave sequence is determined from the optimized pulse wave sequences corresponding to each of the multiple channels. The average matching degree is the average value of the template matching degrees of the pulse waves included in the corresponding optimized pulse wave sequence.

[0043] Optionally, pulse waves are first screened based on their matching degree with the template waveform. Pulse waves with a template matching degree exceeding a preset matching degree threshold are considered high-quality pulse waves and will be retained in the sequence. Additionally, pending pulse waves whose matching degree falls within a preset threshold range are examined; these pending pulse waves have a template matching degree lower than that of high-quality pulse waves but higher than a lower limit. If these pending pulse waves appear consecutively in the sequence, reaching a certain number (e.g., 30% of the total number of pulse waves in the sequence), they are also considered partially valid and included in the valid pulse wave segments. The purpose of this is to maximize the preservation of the continuity of the pulse wave sequence while ensuring the quality of most pulse waves. Based on the above screening, all invalid pulse wave segments in the candidate pulse wave sequences of each channel are marked as discontinuous positions and removed or ignored from the sequence. Thus, each channel will retain an optimized pulse wave sequence consisting of valid and continuous pulse wave segments. Next, the final target pulse wave sequence is selected based on the number of pulse waves in the optimized pulse wave sequence of each channel and the average template matching degree. Specifically, the target sequence can be the channel sequence that contains the most valid pulse waves and has the highest average template matching degree for these pulse waves. This step ensures that the target sequence not only has enough data points but also the highest overall signal quality, making it the most suitable basis for HRV feature calculation.

[0044] For example, all pulse waves in each channel can be filtered. If the template matching degree of non-high-quality pulse waves among n² consecutive high-quality pulse waves (e.g., 30% of the current PPG signal) is greater than th5 (between 0.5 and 0.7), and the number of non-high-quality pulse waves is less than n³ (which can be set to 10% of n²), then these non-high-quality pulse waves are retained. Otherwise, only high-quality pulse waves are used, and the pulse wave continuity marking is updated synchronously. The optimal channel is selected based on the number of pulse waves and the template matching degree in the pulse wave sequence. The corresponding pulse wave sequence is the optimal channel pulse wave sequence. The corresponding selection rule can be: first, check the number of available pulse waves, select the one with the most available pulse waves; if the numbers are equal, calculate the average template matching degree of available pulse waves for each channel, and take the one with the highest average as the optimal pulse wave sequence. The updating of pulse wave continuity marking takes into account that non-high-quality pulse waves (i.e., poor-quality pulse waves) in the pulse wave sequence are not used (exception: when there are a small number of non-high-quality pulse waves in a continuous n² sequence, these non-high-quality pulse waves are also used). Based on where the non-high-quality pulse waves are, the discontinuous ones in the usable pulse waves are marked. For example, if there are 15 pulse waves in total, and the 4th pulse wave is a non-high-quality pulse wave, then the 3rd and 5th pulse waves should be marked as discontinuous.

[0045] This embodiment of the method effectively selects the optimal pulse wave sequence from multiple channels for subsequent processing. This process considers not only the quality of the pulse wave (measured by template matching) but also the continuity of the sequence, avoiding the impact of discontinuous pulse waves on HRV feature calculation. The final target pulse wave sequence, due to its high quality and continuity, will provide solid data support for more accurate stress assessment. This method is particularly crucial in resource-constrained wearable devices because it ensures the accuracy and reliability of stress monitoring while reducing computational load.

[0046] In one optional embodiment, determining a target pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels includes: determining an optimal pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels; detecting whether there are inferior pulse waves in the optimal pulse wave sequence, wherein inferior pulse waves are pulse waves with a template matching degree less than or equal to a matching degree threshold; and detecting whether there are inferior pulse waves in the optimal pulse wave sequence. In the case of poor pulse waves, a specified pulse wave sequence is determined from other pulse wave sequences based on the specified location of the poor pulse wave. The other pulse wave sequences are pulse wave sequences other than the target pulse wave sequence from the optimized pulse wave sequences corresponding to multiple channels. The pulse wave at the specified location in the specified pulse wave sequence is a high-quality pulse wave. The optimal pulse wave sequence and the specified pulse wave sequence are fused according to the specified location to obtain the target pulse wave sequence. Alternatively, if there are no poor pulse waves in the optimal pulse wave sequence, the optimal pulse wave sequence is used as the target pulse wave sequence.

[0047] Optionally, an optimal pulse wave sequence is first determined based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to multiple channels. Then, the optimal pulse wave sequence is checked for low-quality pulse waves—that is, pulse waves with a template matching degree less than or equal to the matching degree threshold. This detection mechanism ensures that all pulse waves in the final sequence are of high quality, thereby guaranteeing the accuracy of HRV features and the reliability of stress assessment. If a low-quality pulse wave is indeed detected in the optimal sequence, a high-quality pulse wave at the corresponding position is searched from the optimized pulse wave sequences of other channels, based on the position of the low-quality pulse wave in the sequence (i.e., a specified position). This means searching for pulse waves of higher quality but similar in position to the low-quality pulse wave in the optimized sequences of each channel for replacement. The found high-quality pulse wave is then fused with the optimal pulse wave sequence at the specified position. By replacing the low-quality pulse wave, the sequence is further optimized, thereby minimizing the impact of signal quality issues on subsequent stress assessment. If no poor-quality pulse waves are found in the optimal pulse wave sequence, then the sequence itself is the target pulse wave sequence. This means that after initial screening and optimization, the sequence has reached a sufficiently high quality standard and can be directly used for subsequent HRV feature extraction and stress assessment.

[0048] For example, if a non-optimal pulse wave (i.e., a poor pulse wave) exists in the optimal channel pulse wave sequence, other channels are searched for high-quality pulse waves within ±5 of the pulse wave peak position. The position of the optimal pulse wave in these channels is selected to replace the peak position of the non-optimal pulse wave in the optimal channel. If no high-quality pulse wave is found, the position of the pulse wave with the highest template matching degree is selected for replacement. Specifically, when a pulse wave P_i in the optimal channel is a non-optimal pulse wave, other channels are searched for high-quality pulse waves within the range [P_i-5, P_i+5]. If there are multiple high-quality pulse waves, such as P_i_c0, P_i_c1, and P_i_c2, then these three pulse waves are examined to see if any are high-quality. If so, the high-quality pulse wave P_i with the highest template matching degree is used to replace the pulse wave P_i. If not, check which of P_i_c0, P_i_c1, P_i_c2, and P_i has the highest template matching degree. Assuming P_i_c1 has the highest, use P_i_c1 as the replacement. If P_i has the highest, leave it as is. If there are discontinuous pulse waves in the optimal channel pulse wave sequence, search for the pulse wave with the highest template matching degree in other channels between adjacent pulse waves, and replace the discontinuous positions in the optimal channel pulse wave sequence to obtain the final pulse wave sequence, i.e., the target pulse wave sequence. Figure 2 This is a schematic diagram of an optional pulse wave sequence fusion according to an embodiment of the present invention, such as... Figure 2 As shown, assuming the optimal channel (i.e., the optimal pulse wave sequence) is PPG_0, which contains one non-high-quality pulse wave (i.e., a poor-quality pulse wave), the position within the black box is determined as the location of the non-high-quality pulse wave. Within the black box, pulse waves from other channels are searched, and pulse wave sequences PPG_2 and PPG_3 are found to contain high-quality pulse waves within the black box. PPG_3 is the best among these, and the corresponding non-high-quality pulse wave in PPG_0 is replaced with the high-quality pulse wave from PPG_3 within the black box.

[0049] This embodiment not only identifies an optimal sequence containing the most effective pulse waves and the highest signal quality, but also employs flexible cross-channel fusion techniques for optimization when a few non-ideal pulse waves exist in the sequence. This mechanism significantly improves the utilization efficiency and reliability of pulse wave data, ensuring stable output of high-quality, continuous pulse wave sequences for subsequent processing, even in complex and variable daily wearing environments. Furthermore, this method highlights the advantages of multi-sensor data fusion, fully utilizing multiple PPG channels on wearable devices and enhancing the overall data quality through complementarity—a feat unmatched by traditional single-channel signal processing.

[0050] In one optional embodiment, a target pulse wave sequence is determined from the optimized pulse wave sequences corresponding to multiple channels based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each channel. This includes: determining a first weight corresponding to the number of pulse waves and a second weight corresponding to the average matching degree; performing a weighted summation operation on the number of pulse waves and the average matching degree of the optimized pulse wave sequences of any channel based on the first and second weights to obtain a comprehensive pulse wave quality score for any channel; obtaining a comprehensive pulse wave quality score for each of the multiple channels based on the first and second weights, using the same method as obtaining the comprehensive pulse wave quality score for any channel; and determining the target pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels based on the comprehensive pulse wave quality scores for each of the multiple channels.

[0051] Optionally, a first weight corresponding to the number of pulse waves and a second weight corresponding to the average pulse wave matching degree are first determined. These weights reflect the different importance of quantity and quality when selecting target pulse wave sequences. The weight settings can be based on experimental results, user needs, or scenario characteristics to ensure that the comprehensive scoring mechanism can balance data volume and data quality, optimizing the selection of the final sequence. For the optimized pulse wave sequence of each channel, a weighted sum of the number of pulse waves and the average matching degree is used to calculate the pulse wave comprehensive quality score. In this way, even if the number of pulse waves in a certain channel is slightly less, if the pulse wave quality is high, its comprehensive quality score may still be high, and vice versa. Using the above method for calculating the pulse wave comprehensive quality score, the optimized pulse wave sequence of each channel is scored, resulting in a set of pulse wave comprehensive quality scores, corresponding to the sequence of each channel. Based on the pulse wave comprehensive quality scores of all channels, the optimized pulse wave sequence with the highest score is selected as the target pulse wave sequence for subsequent HRV feature extraction and stress assessment.

[0052] In this embodiment, a more comprehensive and objective sequence quality assessment method is provided by introducing a weighted summation mechanism of pulse wave quantity and mean matching degree. This mechanism allows the weight of each indicator to be adjusted according to the actual situation. For example, in a sports scenario, the weight of pulse wave quality may be reduced while the weight of quantity may be reduced; conversely, in a resting scenario, the weight of quantity may be higher. By dynamically adjusting the weights, it is possible to respond more flexibly to the wearing conditions in different scenarios, ensuring that the selected target pulse wave sequence contains both a sufficient number of effective data points and high signal quality, thereby improving the accuracy and stability of HRV feature calculation and pressure monitoring.

[0053] Step S108: Extract features from the target pulse wave sequence to obtain the current heart rate variability features.

[0054] Optionally, once the target pulse wave sequence is determined, current heart rate variability (HRV) features are extracted based on this sequence. HRV features include multiple indicators, which may include, but are not limited to, the root mean square of successive differences (RMSSD) based on the time domain, the standard deviation of normal-to-normal intervals (SDNN), and the low-frequency / high-frequency component (LF / HF) ratio based on the frequency domain. In this embodiment, the RR interval refers to the time interval between the current pulse wave (R wave) and the next pulse wave (R wave) in the target pulse wave sequence. These features can reflect the complexity of cardiac autonomic nervous system activity and are related to an individual's psychological stress level.

[0055] In one optional embodiment, feature extraction is performed on the target pulse wave sequence to obtain the current heart rate variability features, including: detecting whether there are discontinuous pulse waves in the target pulse wave sequence; if there are discontinuous pulse waves in the target pulse wave sequence, removing the discontinuous pulse waves in the target pulse wave sequence to obtain a processed pulse wave sequence; and performing feature extraction on the processed pulse wave sequence to obtain the current heart rate variability features.

[0056] Optionally, before extracting HRV features from the target pulse wave sequence, the presence of discontinuous pulse waves in the sequence is first detected. Discontinuous pulse waves may occur due to reasons such as removal during signal quality assessment, replacement during pulse wave fusion, or signal loss caused by motion or environmental factors. These discontinuities can interfere with the calculation of HRV features, leading to biased results. If discontinuous pulse waves are detected in the sequence, they are removed to obtain a more continuous and stable pulse wave sequence. The purpose of removing discontinuous pulse waves is to eliminate the impact of signal interruption on subsequent HRV feature calculation, ensuring the physiological consistency and stability of the features. After sequence preprocessing is completed and continuity and stability are ensured, feature extraction is performed on the processed pulse wave sequence to obtain the current heart rate variability features. HRV features are calculated based on pulse wave intervals (i.e., the time difference between adjacent pulse wave peaks), including but not limited to RMSSD (root mean square of the difference between adjacent RR intervals) and SDNN (root mean square of all RR intervals). These features can reflect the fluctuations in autonomic nervous system activity and are important indicators for assessing psychological stress levels. By implementing continuity checks and pulse wave removal before feature extraction, the calculation of HRV features is ensured to be based on continuous and stable data, thereby improving the accuracy and reliability of the features. This step is particularly critical for stress monitoring, as even small changes in HRV features can significantly impact stress assessment results. By ensuring that processed pulse wave sequences are used for HRV feature extraction, more stable and accurate stress assessment results can be obtained, even in complex environments with variations in signal quality or user activity levels.

[0057] Step S110: Determine the current stress monitoring result based on the current heart rate variability characteristics.

[0058] Optionally, based on the extracted current HRV features, a pre-trained model can be used to calculate the current stress monitoring result. The model can be a statistical model, a machine learning model, or a rule-based model, which can output a quantified stress index based on the relationship between HRV feature values ​​and stress levels. This step transforms HRV features into an interpretable stress state, providing users with real-time stress feedback.

[0059] In one optional embodiment, determining the current stress monitoring result based on the current heart rate variability characteristics includes: using a stress monitoring model based on the current heart rate variability characteristics and the current object information of the target object to obtain the current stress monitoring result, wherein the stress monitoring model is obtained through machine learning based on multiple sets of historical data, and each set of historical data includes historical heart rate variability characteristics, as well as corresponding historical object information and historical stress monitoring results.

[0060] Optionally, the stress monitoring model is trained on a large historical dataset. This historical dataset contains multiple sets of data, each including historical heart rate variability (HRV) characteristics, corresponding historical object information, and historical stress monitoring results. This historical data covers stress levels of different individuals under various environmental conditions, such as work stress, exercise stress, and rest stress, providing comprehensive learning samples for the model. The goal of model training is to learn the correlation between HRV characteristics and object information, and how these factors jointly determine the stress monitoring results. Once the current HRV characteristics and object information of the target object are obtained, this data is input into the pre-trained stress monitoring model for processing. Based on the learned relationships, the stress monitoring model comprehensively analyzes the input HRV characteristics and object information, outputting the corresponding current stress monitoring results. This result reflects the target object's psychological stress state at the current moment, providing timely feedback to the user to help them understand their stress level and take timely coping measures.

[0061] This embodiment proposes a stress monitoring model that combines HRV characteristics and individual attribute information to provide more personalized and contextualized assessment results. Related stress assessments often rely solely on physiological signals, neglecting individual differences and environmental factors, which may lead to poor generalization or inaccuracy of the assessment results. This embodiment's method, by incorporating subject information, enables the model to consider specific individual attributes and their current activity state, such as whether they are in motion, thereby improving the accuracy and applicability of stress assessment.

[0062] Optionally, after acquiring and calculating the current HRV features, a lightweight stress estimation model (i.e., stress monitoring model) is further introduced, such as a linear regression model, a lightweight multi-layer perceptron (MLP), or a simplified logic model based on threshold segmentation, to calculate the user's stress on the wearable device. This model takes multidimensional HRV features and individual user information (i.e., the current object information of the target object) as input and outputs the corresponding stress index based on pre-trained parameters. Taking an explicit regression model as an example, the linear regression model can adopt the following exemplary structure: an input layer, used to input several HRV features (such as RMSSD, SDNN, heart rate, etc.), with feature dimensions ranging from 3 to 20; weight coefficients and bias terms: the model is a single-layer linear mapping, and stress estimation is completed by weighted summation combined with bias; the output layer is used to output a continuous value as the stress index. Taking a lightweight multilayer perceptron (MLP) as an example, the lightweight MLP model structure includes: an input layer for inputting HRV and user information feature vectors; hidden layers set to 1 to 3 layers, each containing approximately 8 to 64 neurons, and lightweight activation functions such as rectified linear units (ReLU), hyperbolic tangent (Tanh), or logistic sigmoid (Sigmoid) can be used; the output layer includes 1 neuron, used to output the stress index as the current stress monitoring result.

[0063] As an optional implementation, taking a simplified logic model based on threshold segmentation as an example, when there are multiple current heart rate variability features (such as RMSSD, SDNN, etc.), it can be set that the older the age, the smaller the typical heart rate variability value, and the high stress judgment can be appropriately relaxed. An age-based adjustment factor delta_age is determined based on age. Based on the stress intervals corresponding to each of the multiple current heart rate variability features, the weight values ​​corresponding to each of the multiple current heart rate variability features, and the age-based adjustment factor, the current stress monitoring result is obtained. For example, stress is divided into high, medium, and low intervals according to SDNN. If the SDNN falls within the 0-40 percentile of the population's nighttime SDNN mean, it is a low stress interval (W_SDNN=0.1); if it is in the 40-60 percentile, it is a medium stress interval (W_SDNN=0.3); and if it is in the 60-100 percentile, it is a high stress interval (W_SDNN=0.6). RMSSD is processed in the same way; finally, the current stress monitoring result is obtained as stress=w1. W_SDNN+w2 W_RMSSD+delta_age, where w1 and w2 represent the weight values ​​of the feature RMSSD and SDNN respectively, and W_SDNN and W_RMSSD represent the stress intervals of the feature RMSSD and SDNN respectively.

[0064] In one optional embodiment, based on the current heart rate variability characteristics and the current object information of the target object, a stress monitoring model is used to obtain the current stress monitoring result, including: based on the current heart rate variability characteristics and the current object information, a stress monitoring model is used to obtain a predicted stress monitoring result; the historical stress range of the target object in a preset reference time period is obtained; the offset between the historical stress range and the reference stress range is determined; if the offset is greater than a preset offset threshold, the predicted stress monitoring result is corrected based on the offset to obtain the current stress monitoring result.

[0065] Optionally, the system first receives the target subject's current heart rate variability (HRV) characteristics and current subject information (such as age and gender), and inputs this data into a pre-trained stress monitoring model. The model calculates preliminary predicted stress monitoring results based on this information. This stress monitoring model is trained using machine learning on multiple sets of historical data. Each set of historical data includes historical HRV characteristics, corresponding historical subject information, and historical stress monitoring results. The model learns the statistical association between heart rate variability and stress levels. The system queries the target subject's historical stress monitoring results during a preset reference period (such as nighttime sleep, as this stage is relatively stable and can be considered a baseline for stress levels) to determine the stress level range within that period. Next, the system compares the target subject's historical stress range during the reference period with a pre-set reference stress range (a standard stress range defined based on statistical data from a broad population) to determine the offset between the two. The existence of an offset may indicate that the model prediction needs individualized adjustment to reflect the target subject's true stress state. If the calculated offset exceeds a preset offset threshold (indicating a significant difference between the model's prediction and the user's historical baseline), the initial predicted stress monitoring results are corrected to narrow the gap between the prediction and the user's historical data. Correction methods may include, but are not limited to, linear adjustment, scaling, or function mapping, depending on the magnitude and direction of the offset. This correction step aims to improve the individualization and accuracy of stress monitoring results, ensuring that the results not only reflect general human physiological patterns but also take into account the unique physiological characteristics of the target individual and long-term trends in stress levels.

[0066] Optionally, to improve the individual adaptability of stress assessment, this embodiment introduces a baseline correction mechanism based on long-term user data. During continuous device use, the user's HRV characteristics in low-activity or stable states are automatically collected, and an individualized physiological baseline is constructed. This individualized physiological baseline reflects the user's own resting state autonomic nervous system regulation level. Subsequently, when calculating the stress index, this embodiment compares the current HRV characteristics with the user's baseline characteristics. Through offset correction, normalization mapping, or difference gain, the stress results output by the model are dynamically adjusted to make the stress value more consistent with the user's individual physiological changes. The specific implementation process is as follows: Baseline acquisition: The stress index is collected during the user's nighttime sleep, and the average stress value during this period is calculated as the user's baseline. Nighttime sleep is typically a low-stress state, so this average can be used as a low-stress reference; Deviation detection: The nighttime baseline average is compared with the expected low-stress level (s0). If the nighttime average is significantly higher than the reference low pressure level, adjustments are made using the following methods: Offset correction: Add / subtract an offset to the overall model output to adjust the nighttime average to the reference low pressure level; Normalization mapping: Perform linear or nonlinear mapping on the pressure output based on the nighttime baseline to ensure the output falls within the range of s0 to 100, while maintaining the characteristics of daytime fluctuations; Difference gain: Adjust the sensitivity of daytime pressure fluctuations according to the degree of baseline deviation, amplifying or compressing small deviations.

[0067] In the correction process described in this embodiment, individualized adjustments to stress monitoring results are achieved by combining current heart rate variability characteristics, subject information, and individual historical data. This technological advantage lies in its ability to overcome the limitations of general models when dealing with individual differences, ensuring that stress assessments more closely reflect the user's actual physiological state and long-term stress trends, thereby improving the reliability and applicability of the monitoring results.

[0068] Through the above steps S102 to S110, the continuity and quality of multi-channel pulse wave signals can be analyzed, and HRV features can be extracted from the optimized signals. This enables accurate monitoring and feedback of the pressure level of the monitored object even under motion or signal interference conditions, thereby improving the technical effect of enhancing the accuracy and reliability of pressure monitoring of wearable devices in dynamic environments. This solves the technical problem of low accuracy and reliability of pressure monitoring of wearable devices in dynamic environments in related technologies.

[0069] 24 / 7 stress monitoring helps users understand their physical and mental state. Current wearable devices generally estimate heart rate variability (HRV) using PPG signals and further extrapolate stress levels. However, wearable-based stress monitoring methods suffer from the following problems: Coarse-grained signal quality assessment: In practical use, wrist PPG signals are easily affected by factors such as movement, wearing tightness, ambient light, and skin condition, resulting in unstable signal quality. The main technology uses a global SNR index, which cannot provide fine-grained quality scoring for each pulse wave, leading to difficulty in extracting HRV features or insufficient accuracy. Coarse-grained motion state judgment: Methods in related technologies only use overall ACC to determine movement or stillness, lacking refined processing for short-term violent shaking or small-amplitude movements, resulting in a large amount of unusable data and inability to extract HRV and stress values. Weak multi-channel fusion strategy: Although multi-channel signals exist, methods in related technologies typically fuse the original PPG signals without dynamic selection of pulse waves, making it difficult to obtain high-quality pulse wave sequences. Limited on-device storage / computing: Wearable devices (such as watches or wristbands) have limited storage and computing power, making it impossible to frequently save long-term data or perform computationally intensive DTW or run large-scale deep learning models.

[0070] To address the aforementioned problems, and in conjunction with the above embodiments and optional embodiments, the present invention proposes an optional implementation method. Figure 3 This is a flowchart of an optional pressure monitoring method based on a wearable device according to an embodiment of the present invention. In this embodiment, multi-channel pulse wave (PPG) signals are acquired by the wearable device, along with acceleration (ACC) signals used to assess motion status. A hierarchical coarse-grained and fine-grained PPG signal quality assessment is performed on a small segment of the real-time acquired PPG and ACC signals to screen out high-quality PPG segments and superior pulse waves. The coarse-grained assessment evaluates the signal quality of the PPG segments, while the fine-grained assessment refines each pulse wave based on template matching and motion intensity. After performing the above assessment on all pulse waves in several small PPG signal segments acquired in a single measurement, the optimal channel is determined based on template matching and motion status. Using the optimal channel as the primary channel and the remaining channels as auxiliary channels, multi-channel pulse wave fusion is performed to obtain the final pulse wave sequence. HRV features are calculated based on the final pulse wave sequence. During the calculation process, the continuity of the sequence is checked and constrained, and discontinuous intervals are eliminated to ensure the physiological consistency of the HRV features and the reliability of the calculation results. In terms of environment configuration, set up an algorithm development environment and ensure that necessary libraries are installed. Install libraries such as NumPy and SciPy for signal processing and quality evaluation. Install libraries such as Neurokit2 for pulse wave peak extraction. Figure 3 As shown, the method specifically includes:

[0071] S1, coarse-grained signal quality assessment: Given the duration-dependent nature of HRV feature calculation and the limited computing and storage resources of terminal devices, this embodiment introduces a segment-by-segment signal quality assessment strategy. After acquiring n seconds of PPG and ACC signals, a quality assessment is performed, storing the high-quality pulse wave peak positions, continuity, and templates. Once a pressure monitoring data acquisition is complete, subsequent processing is performed based on the positions of all pulse wave peaks. Specifically, this includes:

[0072] S11, Coarse-grained evaluation: For the current PPG signal acquired in the current time period, perform amplitude anomaly judgment point by point. If the amplitude exceeds the threshold th1, it is considered that there is an amplitude too large anomaly at the current point. Here, th1 is obtained by adding 3 times the standard deviation to the amplitude mean of the current PPG signal. Perform a differential operation on the current PPG signal. If n1 consecutive differential values ​​are less than th2, it is considered that there is an amplitude too small anomaly. Here, th2 can be 0.01 times the differential mean of the current PPG signal. Record the positions of the two amplitude anomalies, set all amplitude anomaly segments to 0, and obtain the preprocessed PPG signal PPG1.

[0073] S12, Fine-grained template matching: Peak extraction is performed on PPG1. The continuity of the pulse wave is marked based on the position of the amplitude anomaly, since the PPG of the amplitude anomaly segment is unusable and all values ​​are 0. The last pulse wave before the amplitude anomaly and the first pulse wave after the amplitude anomaly are discontinuous. Pulse quality is determined using template matching: a template is generated for each pulse wave based on the peak position of the pulse wave. The similarity between the pulse wave and the template waveform is calculated. If the template matching degree is greater than th3 (the corresponding value range can be set between 0.7 and 1), it is considered a high-quality pulse wave. The template matching degree of all pulse waves is retained. The template waveform is continuously retained and gradually updated with newly acquired high-quality pulse waves. The specific method of obtaining the template waveform is the same as in the previous embodiment and will not be repeated here.

[0074] S13, Fine-grained - Motion Intensity: Analyze the ACC signal between the peak positions of every two pulse waves. If the relative fluctuation intensity is greater than th4 (e.g., 0.01), it is considered that there is motion between the two pulse waves. The latter pulse wave is discarded and the position is marked as discontinuous.

[0075] S2, Optimal Channel Selection and Channel Fusion, specifically includes:

[0076] S21, Optimal Channel Selection: After a measurement is completed, all pulse waves in each channel are filtered. If the template matching degree of non-high-quality pulse waves among consecutive n2 high-quality pulse waves (e.g., 30% of the number of pulse waves in the current PPG signal) is greater than th5 (between 0.5 and 0.7), and the number of non-high-quality pulse waves is less than n3 (which can be set to 10% of n2), then these non-high-quality pulse waves are retained; otherwise, only high-quality pulse waves are used. The pulse wave continuity marker is updated. The optimal channel is selected based on the number of pulse waves and the template matching degree in the pulse wave sequence. The corresponding pulse wave sequence is the optimal channel pulse wave sequence. The corresponding selection rules are the same as in the aforementioned embodiment and will not be repeated here.

[0077] S2, Multi-channel fusion: If the optimal channel pulse wave sequence contains non-high-quality pulse waves (i.e., inferior pulse waves), search for high-quality pulse waves within ±5 of the peak position in other channels. Replace the peak position of the non-high-quality pulse wave in the optimal channel with the position of the optimal pulse wave. If no high-quality pulse wave is found, replace it with the pulse wave position that has the highest template matching degree. If the optimal channel pulse wave sequence contains discontinuous pulse waves, search for the pulse wave with the highest template matching degree in other channels between adjacent pulse waves. Replace the discontinuous positions in the optimal channel pulse wave sequence to obtain the final pulse wave sequence.

[0078] S3, HRV and pressure calculations, specifically including:

[0079] S31, HRV Calculation: HRV features are calculated based on the optimal pulse wave sequence. For HRV features related to continuity, such as RMSSD and SDNN, discontinuous pulse waves are removed based on the pulse wave continuity obtained during coarse-grained signal quality assessment and channel fusion to ensure the physiological consistency of HRV features and the reliability of calculation results.

[0080] S32, Stress Calculation: After acquiring and calculating the current HRV features, a lightweight stress estimation model (i.e., stress monitoring model) is further introduced, such as a linear regression model, a lightweight multilayer perceptron (MLP), or a simplified logic model based on threshold segmentation, to calculate the user's stress at the wearable device. This model takes multidimensional HRV features and individual user information (i.e., the current object information of the target object) as input and outputs the corresponding stress index based on pre-trained parameters. The specific implementation of each model is the same as in the aforementioned embodiments and will not be repeated here. To improve the individual adaptability of stress assessment, this embodiment introduces a baseline correction mechanism based on long-term user data. During continuous device wear, the user's HRV features in low-activity or stable states are automatically collected, and an individualized physiological baseline is constructed. The individualized physiological baseline reflects the user's own resting state autonomic nervous system regulation level. Subsequently, when calculating the stress index, the current HRV features are compared with the user's baseline features. Through offset correction, normalization mapping, or difference gain, the stress results output by the model are dynamically adjusted so that the stress value better matches the user's individual physiological characteristics. The specific implementation method is the same as the aforementioned embodiments, and will not be repeated here.

[0081] It should be noted that related technologies typically rely on signal-to-noise ratio (SNR) and other methods for overall quality assessment, resulting in low data utilization efficiency and low pressure output rates. Furthermore, considering only a single channel or performing channel fusion at the PPG data level fails to accurately identify the optimal pulse wave. Additionally, these technologies do not consider the continuity of the pulse wave in HRV calculations, limiting the accuracy of HRV features and impacting pressure accuracy. Moreover, most related technical solutions do not incorporate individual adaptation models, relying solely on uniform, fixed parameters for pressure calculation, thus failing to adaptively compensate for differences in the physiological baselines of different users. Unlike related technologies, this embodiment employs a dual-granularity evaluation approach of "whole-segment quality + individual pulse wave quality" in the coarse-grained and fine-grained pulse wave quality assessment stage, significantly improving the accuracy of pulse wave screening and thus significantly increasing the pressure output rate. Through a robust multi-channel fusion strategy, multiple channels are independently evaluated for quality, and the optimal channel is selected based on template matching. Other channels are then used for supplementation and correction, resulting in a more stable pulse wave sequence that is more robust to individual differences. By using continuous HRV feature extraction, the continuity of pulse wave sequences, which is commonly ignored in related technologies and prone to HRV feature deviation due to missing peaks, is effectively addressed. This embodiment integrates pulse wave continuity into the HRV calculation process, effectively avoiding errors caused by discontinuous pulse waves and improving the accuracy of pressure calculation. Through a device-constrained PPG signal segmentation processing mechanism, a small-segment segmentation evaluation and real-time template update strategy is adopted, retaining only necessary pulse wave information each time. This significantly reduces computational and memory overhead while maintaining accuracy, achieving near-offline real-time performance. By using an individual baseline-adaptive stress calculation mechanism, the long-term baseline characteristics of users are dynamically accumulated during continuous wear. By establishing an individualized baseline, the stress results are continuously corrected, enabling the model to adapt to the physiological differences and long-term trends of different users, thereby significantly improving the adaptability and stability of stress assessment to individuals.

[0082] It should be noted that, in addition to the methods described in the above embodiments, the following alternative embodiments can also be used: 1) Multimodal data fusion: Introducing more physiological signals and motion-related signals (such as skin conductance EDA, gyroscope signals, etc.) to assist in stress assessment, further improving the stability and accuracy of stress results. 2) Adopting a more transparent stress rating system: Constructing a more transparent and easily understandable stress level mapping mechanism, converting continuous stress values ​​into graded stress states, improving users' understanding and interpretability of their own stress levels. 3) Performing personalized model continuous learning: Based on the baseline correction of the above embodiments, introducing a long-term personalized learning mechanism to dynamically model changes in users' physiological rhythms (such as seasonal changes, changes in daily routines, long-term stress trends, etc.), enabling the stress algorithm to maintain stable accuracy over a longer time scale. 4) Adaptive scene recognition: By adding a scene recognition module (such as resting, walking, exercising, sleeping, etc.), the stress algorithm can automatically adjust feature processing strategies and model weights according to different scenarios, thereby improving reliability in high-noise, strenuous exercise scenarios.

[0083] This embodiment achieves at least one of the following effects: 1) Higher accuracy: Through multi-granularity signal quality analysis, multi-channel fusion, and HRV feature extraction, the method in this embodiment can extract more accurate HRV features and stress assessment results. 2) Better flexibility: This embodiment evaluates the pulse wave signal quality based on the individual user's situation using template waveform matching. 3) Stronger robustness: In everyday exercise scenarios, the combination of coarse and fine signal quality analysis with multi-channel fusion can still calculate relatively accurate stress values. 4) Enhanced individualized reliability: During inference, the model dynamically adapts to the user's individual baseline, making the stress estimation more consistent with the individual's physiological baseline and long-term trends, thus maintaining stable and reliable stress results under different environments and conditions. In summary, this embodiment not only improves the accuracy of stress values ​​but also increases the stress value output rate by improving data utilization efficiency.

[0084] This embodiment also provides a pressure monitoring device based on a wearable device, which is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the terms "module" and "device" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0085] According to embodiments of the present invention, an apparatus embodiment for implementing the above-described pressure monitoring method based on wearable devices is also provided. Figure 4 This is a schematic diagram of a pressure monitoring device based on a wearable device according to an embodiment of the present invention, as shown below. Figure 4 As shown, the aforementioned pressure monitoring device based on a wearable device includes: a PPG signal acquisition module 400, a candidate pulse wave sequence determination module 402, a target pulse wave sequence determination module 404, a feature extraction module 406, and a pressure monitoring module 408, wherein:

[0086] PPG signal acquisition module 400 is used to acquire the current photoplethysmography (PPG) signals corresponding to each of the multiple channels collected by the wearable device. The current PPG signals corresponding to each of the multiple channels are acquired for the target object in the current time period.

[0087] The candidate pulse wave sequence determination module 402 is connected to the PPG signal acquisition module 400 and is used to determine the candidate pulse wave sequence corresponding to each of the multiple channels based on the current PPG signal corresponding to each of the multiple channels. The candidate pulse wave sequence includes multiple pulse waves of the corresponding channel.

[0088] The target pulse wave sequence determination module 404, connected to the candidate pulse wave sequence determination module 402, is used to determine the target pulse wave sequence from the candidate pulse wave sequences corresponding to multiple channels based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to multiple channels. The pulse wave continuity evaluation results are used to indicate the continuity of the pulse waves in the corresponding candidate pulse wave sequence, and the pulse wave quality evaluation results are used to indicate the template matching degree of each of the multiple pulse waves included in the corresponding candidate pulse wave sequence. The template matching degree is used to indicate the matching degree between the corresponding pulse wave and the template waveform.

[0089] The feature extraction module 406 is connected to the target pulse wave sequence determination module 404 and is used to extract features from the target pulse wave sequence to obtain the current heart rate variability features.

[0090] The stress monitoring module 408, connected to the feature extraction module 406, is used to determine the current stress monitoring result based on the current heart rate variability characteristics.

[0091] It should be noted that the above modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above modules can be located in the same processor; or the above modules can be located in different processors in any combination.

[0092] It should be noted that the PPG signal acquisition module 400, candidate pulse wave sequence determination module 402, target pulse wave sequence determination module 404, feature extraction module 406, and pressure monitoring module 408 mentioned above correspond to steps S102 to S110 in the embodiments. The instances and application scenarios implemented by the above modules and corresponding steps are the same, but are not limited to the content disclosed in the above embodiments. It should be noted that the above modules, as part of the device, can run on a computer terminal.

[0093] It should be noted that the optional or preferred implementation methods of this embodiment can be found in the relevant descriptions in the embodiments, and will not be repeated here.

[0094] The aforementioned pressure monitoring device based on wearable devices may also include a processor and a memory. The PPG signal acquisition module 400, the candidate pulse wave sequence determination module 402, the target pulse wave sequence determination module 404, the feature extraction module 406, the pressure monitoring module 408, etc., are all stored in the memory as program modules, and the processor executes the aforementioned program modules stored in the memory to realize the corresponding functions.

[0095] The processor contains a core that retrieves the corresponding program modules from memory. One or more cores may be configured. Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory includes at least one memory chip.

[0096] According to an embodiment of this application, an embodiment of a non-volatile storage medium is also provided. Optionally, in this embodiment, the non-volatile storage medium includes a stored program, wherein, when the program is running, it controls the device containing the non-volatile storage medium to execute any of the aforementioned pressure monitoring methods based on wearable devices.

[0097] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals, and the non-volatile storage medium includes stored programs.

[0098] Optionally, during program execution, a program controls the device containing the non-volatile storage medium to execute any of the above-described pressure monitoring method steps based on wearable devices.

[0099] According to an embodiment of this application, an embodiment of a processor is also provided. Optionally, in this embodiment, the processor is used to run a program, wherein the program executes any of the aforementioned stress monitoring methods based on wearable devices.

[0100] According to an embodiment of this application, an embodiment of a computer program product is also provided, which, when executed on a data processing device, is adapted to execute a program that initializes the stress monitoring method steps based on a wearable device, having any of the steps described above.

[0101] This invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of any of the above-described pressure monitoring methods based on wearable devices.

[0102] The order of the above embodiments of the present invention is merely for description and does not represent the superiority or inferiority of the embodiments.

[0103] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0104] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of modules described above can be a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between modules, and may be electrical or other forms.

[0105] The modules described above as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0106] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated modules described above can be implemented in hardware or as software functional modules.

[0107] If the aforementioned integrated modules are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable non-volatile storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a non-volatile storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned non-volatile storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0108] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A pressure monitoring method based on wearable devices, characterized in that, include: The wearable device acquires the current photoplethysmography (PPG) signals corresponding to each of the multiple channels, wherein the current PPG signals corresponding to each of the multiple channels are acquired for the target object in the current time period. Based on the current PPG signal corresponding to each of the multiple channels, a candidate pulse wave sequence corresponding to each of the multiple channels is determined, wherein the candidate pulse wave sequence includes multiple pulse waves of the corresponding channel; Based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, a target pulse wave sequence is determined from the candidate pulse wave sequences corresponding to each of the multiple channels. The pulse wave continuity evaluation results are used to indicate the continuity of the pulse waves in the corresponding candidate pulse wave sequence, and the pulse wave quality evaluation results are used to indicate the template matching degree of each of the multiple pulse waves included in the corresponding candidate pulse wave sequence. The template matching degree is used to indicate the matching degree between the corresponding pulse wave and the template waveform. Feature extraction is performed on the target pulse wave sequence to obtain the current heart rate variability features; Based on the current heart rate variability characteristics, determine the current stress monitoring results.

2. The method according to claim 1, characterized in that, The step of determining the candidate pulse wave sequence corresponding to each of the multiple channels based on the current PPG signal of each channel includes: The pulse wave is split into the current PPG signals corresponding to each of the multiple channels to obtain the initial pulse wave sequences corresponding to each of the multiple channels. Based on the current acceleration signal collected by the wearable device in the current time period, abnormal fluctuation segments in the initial pulse wave sequence corresponding to each of the multiple channels are identified, wherein the abnormal fluctuation segment is a segment between peaks where the relative fluctuation intensity of the acceleration signal is greater than a preset intensity threshold, and the segment between peaks is a pulse wave segment between the peaks corresponding to two pulse waves. By removing the pulse wave corresponding to the end peak of the abnormal fluctuation segment from the initial pulse wave sequence corresponding to each of the multiple channels, candidate pulse wave sequences corresponding to each of the multiple channels are obtained, wherein the end peak is the peak corresponding to the end position of the abnormal fluctuation segment.

3. The method according to claim 2, characterized in that, The step of splitting the current PPG signal corresponding to each of the multiple channels into pulse waves to obtain the initial pulse wave sequence corresponding to each of the multiple channels includes: Identify amplitude aberration segments in the current PPG signal corresponding to each of the multiple channels, wherein the amplitude aberration segments include aberration segments in the current PPG signal whose amplitude is greater than a first preset amplitude, and aberration segments in the current PPG signal whose consecutive predetermined number of differential values ​​are all less than a second preset amplitude, wherein the differential values ​​are obtained by performing a differential operation on the corresponding current PPG signal; The amplitude anomaly segments in the current PPG signals corresponding to each of the multiple channels are replaced with predetermined values ​​to obtain the preprocessed PPG signals corresponding to each of the multiple channels. Based on the peaks of the preprocessed PPG signals corresponding to each of the multiple channels, the preprocessed PPG signals corresponding to each of the multiple channels are split into pulse waves to obtain candidate pulse wave sequences corresponding to each of the multiple channels, wherein each pulse wave is a signal segment within a predetermined neighborhood range of the corresponding peak.

4. The method according to claim 3, characterized in that, Before determining the target pulse wave sequence from the candidate pulse wave sequences corresponding to each of the multiple channels based on the pulse wave continuity assessment results and pulse wave quality assessment results corresponding to each of the multiple channels, the method further includes: In the candidate pulse wave sequences corresponding to the multiple channels, the pulse wave positions corresponding to the amplitude abnormal segments are marked as discontinuous positions, and the pulse wave positions corresponding to the end peaks of the fluctuation abnormal segments are marked as discontinuous positions, thereby obtaining the pulse wave continuity evaluation results corresponding to the multiple channels.

5. The method according to claim 1, characterized in that, Before determining the target pulse wave sequence from the candidate pulse wave sequences corresponding to each of the multiple channels based on the pulse wave continuity assessment results and pulse wave quality assessment results corresponding to each of the multiple channels, the method further includes: The pulse wave quality assessment result for any one of the multiple channels is obtained as follows: A point-by-point averaging operation is performed on multiple pulse waves in the candidate pulse wave sequence of any channel to obtain the template waveform corresponding to any channel. The point-by-point averaging operation is used to calculate the average value of the values ​​corresponding to the same point in the multiple pulse waves. Calculate the template matching degree between multiple pulse waves in the candidate pulse wave sequence of any channel and the template waveform respectively, and obtain the pulse wave quality evaluation result of any channel; The pulse wave quality assessment results for each of the multiple channels are obtained by using the method of obtaining the pulse wave quality assessment results for any one of the channels.

6. The method according to claim 1, characterized in that, The step of determining the target pulse wave sequence from the candidate pulse wave sequences corresponding to each of the multiple channels based on the pulse wave continuity assessment results and pulse wave quality assessment results of each of the multiple channels includes: Based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to each of the multiple channels, effective pulse wave segments in the candidate pulse wave sequences corresponding to each of the multiple channels are determined, and other pulse waves except for the effective pulse wave segments are marked as discontinuous positions, thereby obtaining the optimized pulse wave sequences corresponding to each of the multiple channels. The effective pulse wave segments include high-quality pulse waves with a template matching degree greater than the matching degree threshold, and pulse wave segments composed of more than a specified number of undetermined pulse waves. The undetermined pulse waves are pulse waves with a template matching degree within a preset threshold range, and the upper limit of the preset threshold range is the matching degree threshold. Based on the number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels, the target pulse wave sequence is determined from the optimized pulse wave sequences corresponding to each of the multiple channels, wherein the average matching degree is the average value of the template matching degree of each pulse wave included in the corresponding optimized pulse wave sequence.

7. The method according to claim 6, characterized in that, The step of determining the target pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels, based on the average number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels, includes: Based on the average number of pulse waves and the matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels, the optimal pulse wave sequence is determined from the optimized pulse wave sequences corresponding to each of the multiple channels. Detect whether there are inferior pulse waves in the optimal pulse wave sequence, wherein the inferior pulse wave is a pulse wave with a template matching degree less than or equal to the matching degree threshold; If a poor-quality pulse wave exists in the optimal pulse wave sequence, a specified pulse wave sequence is determined from other pulse wave sequences based on the specified position of the poor-quality pulse wave. These other pulse wave sequences are pulse wave sequences other than the target pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels. The pulse wave at the specified position in the specified pulse wave sequence is a high-quality pulse wave. The optimal pulse wave sequence and the specified pulse wave sequence are then fused according to the specified position to obtain the target pulse wave sequence; or... If the inferior pulse wave is not present in the optimal pulse wave sequence, the optimal pulse wave sequence shall be used as the target pulse wave sequence.

8. The method according to claim 6, characterized in that, The step of determining the target pulse wave sequence from the optimized pulse wave sequences corresponding to each of the multiple channels, based on the average number of pulse waves and the average matching degree of the optimized pulse wave sequences corresponding to each of the multiple channels, includes: Determine the first weight corresponding to the number of pulse waves and the second weight corresponding to the mean matching degree; Based on the first weight and the second weight, the pulse wave count and the mean matching degree of the optimized pulse wave sequence of any channel are weighted and summed to obtain the comprehensive quality score of the pulse wave of any channel. Based on the first weight and the second weight, the pulse wave quality comprehensive score corresponding to each of the multiple channels is obtained by using the method of obtaining the pulse wave quality comprehensive score of any channel. Based on the comprehensive pulse wave quality score corresponding to each of the multiple channels, the target pulse wave sequence is determined from the optimized pulse wave sequences corresponding to each of the multiple channels.

9. The method according to any one of claims 1 to 8, characterized in that, The feature extraction of the target pulse wave sequence to obtain the current heart rate variability features includes: Detect whether there are discontinuous pulse waves in the target pulse wave sequence; If the target pulse wave sequence contains discontinuous pulse waves, the discontinuous pulse waves in the target pulse wave sequence are removed to obtain the processed pulse wave sequence. Feature extraction is performed on the processed pulse wave sequence to obtain the current heart rate variability features.

10. The method according to any one of claims 1 to 8, characterized in that, The step of determining the current stress monitoring result based on the current heart rate variability characteristics includes: Based on the current heart rate variability characteristics and the current object information of the target object, a stress monitoring model is used to obtain the current stress monitoring result. The stress monitoring model is obtained through machine learning based on multiple sets of historical data. Each set of historical data includes historical heart rate variability characteristics, corresponding historical object information, and historical stress monitoring results.

11. The method according to claim 10, characterized in that, The current stress monitoring result is obtained by using a stress monitoring model based on the current heart rate variability characteristics and the current object information of the target object, including: Based on the current heart rate variability characteristics and the current object information, the stress monitoring model is used to obtain the predicted stress monitoring results; Obtain the historical pressure range of the target object during a preset reference time period; Determine the offset between the historical pressure range and the reference pressure range; If the offset is greater than a preset offset threshold, the predicted pressure monitoring result is corrected based on the offset to obtain the current pressure monitoring result.

12. A pressure monitoring device based on a wearable device, characterized in that, include: The PPG signal acquisition module is used to acquire the current photoplethysmography (PPG) signals corresponding to each of the multiple channels collected by the wearable device, wherein the current PPG signals corresponding to each of the multiple channels are acquired for the target object in the current time period. The candidate pulse wave sequence determination module is used to determine the candidate pulse wave sequence corresponding to each of the multiple channels based on the current PPG signal corresponding to each of the multiple channels, wherein the candidate pulse wave sequence includes multiple pulse waves of the corresponding channel; The target pulse wave sequence determination module is used to determine the target pulse wave sequence from the candidate pulse wave sequences corresponding to the multiple channels based on the pulse wave continuity evaluation results and pulse wave quality evaluation results corresponding to the multiple channels respectively. The pulse wave continuity evaluation results are used to indicate the continuity of the pulse waves in the corresponding candidate pulse wave sequence, and the pulse wave quality evaluation results are used to indicate the template matching degree of the multiple pulse waves included in the corresponding candidate pulse wave sequence. The template matching degree is used to indicate the matching degree between the corresponding pulse wave and the template waveform. The feature extraction module is used to extract features from the target pulse wave sequence to obtain the current heart rate variability features; The stress monitoring module is used to determine the current stress monitoring result based on the current heart rate variability characteristics.

13. An electronic device, characterized in that, It includes one or more processors and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the stress monitoring method based on any one of claims 1 to 11.

Images