Method and system for calculating psychological stress based on wearable device

Abstract

This application discloses a method for calculating psychological stress based on wearable devices, comprising the following steps: acquiring a detection signal; filtering the detection signal; acquiring an RRI sequence from the filtered detection signal; determining the signal quality; calculating several time-domain features of the RRI sequence; performing bandpass filtering on the RRI sequence after extension to obtain low-frequency and high-frequency energy values, and extracting frequency-domain features based on the low-frequency and high-frequency energy values; and calculating psychological stress based on the time-domain features, frequency-domain features, and signal quality. By implementing this scheme, computational resources can be reduced, and dependence on signal quality can be reduced.

Description

Technical Field

[0001] This application relates to wearable devices, and in particular to a method and system for calculating psychological stress based on wearable devices. Background Technology

[0002] In today's rapidly developing society, people are constantly under pressure from all aspects, whether it is from studying, working or family. In the short term, appropriate pressure can give people some motivation, but being under high psychological pressure for a long time will cause serious and irreparable damage to people's physical or mental health. Therefore, it is essential to monitor psychological stress at any time.

[0003] While existing methods for stress assessment using heart rate variability (HRV) offer advantages such as immediacy and low cost per measurement, most suffer from drawbacks including high resource consumption, stringent signal quality requirements, and limited consideration of anomalies specific to wearable devices. Furthermore, most HRV-based stress assessment methods exhibit poor adaptability and may fail to accurately reflect a user's true psychological stress level in certain situations. Summary of the Invention

[0004] This invention aims to address at least one of the technical problems existing in the prior art. To this end, this invention proposes a method and system for calculating psychological stress based on wearable devices, thereby reducing the resource consumption of stress assessment methods based on heart rate variability, lowering signal quality requirements, and increasing the applicability and stability of the algorithm.

[0005] This application provides a method for calculating psychological stress based on wearable devices, including the following steps:

[0006] Acquire detection signals;

[0007] The detection signal is filtered.

[0008] The RRI sequence is obtained from the filtered detection signal;

[0009] Determine signal quality;

[0010] Calculate several time-domain characteristics of the RRI sequence;

[0011] The RRI sequence is extended and then bandpass filtered to obtain low-frequency and high-frequency energy values, and frequency domain features are extracted based on the low-frequency and high-frequency energy values.

[0012] Psychological stress is calculated based on the time-domain characteristics, frequency-domain characteristics, and signal quality.

[0013] In some embodiments, the filtering of the detection signal includes:

[0014] The bandpass signal within a preset range is extracted using a third-order Butterworth bandpass filter;

[0015] Differential filtering is applied to the bandpass-filtered signal;

[0016] The signal after differential filtering is subjected to mean filtering.

[0017] In some embodiments, obtaining the RRI sequence from the filtered detection signal includes:

[0018] Peak points in the filtered detection signal that meet the first preset condition are marked;

[0019] Determine whether the marked peak points are invalid peak points;

[0020] The RRI sequence is updated based on the validity of the peak points;

[0021] The RRI sequence is adjusted for outliers by interpolating the RRI values ​​that meet the outlier conditions to obtain the RRI sequence.

[0022] In some embodiments, satisfying the first preset condition means satisfying the following conditions: the peak point is a maximum point; there are minimum points less than the first threshold between the previous valid peak point and the peak point, and the number of minimum points that meet the requirements is less than the second threshold; the peak point is greater than the third threshold.

[0023] The determination of whether a marked peak point is an invalid peak point specifically involves:

[0024] If the time interval between a suspected peak point and the previous valid peak point is less than the fourth threshold, the peak point is determined to be an invalid peak point; otherwise, the peak point is determined to be a valid peak point.

[0025] The update of the RRI sequence based on the validity of the peak points specifically involves:

[0026] If the time interval is less than the fifth threshold, the time interval is determined to be a valid RRI, and the RRI sequence is updated. Otherwise, it is determined that there is a missed detection between the two peak points, the RRI is determined to be an invalid value, and no RRI update is performed.

[0027] In some embodiments, the outlier adjustment of the RRI sequence involves interpolating the RRI values ​​that meet the outlier conditions to obtain the RRI sequence, specifically:

[0028] Calculate the mean of the entire RRI sequence. For each RRI value in the RRI sequence, judge whether the RRI value is less than 0.7 times or greater than 1.3 times the mean of the RRI sequence, and then perform difference processing on the RRI value.

[0029] in,

[0030]

[0031] In the above formula, y is the result after interpolation, f1 is the RRI that meets the threshold requirement before this point, f2 is the first RRI that meets the threshold requirement after this point, m is the number of intervals between this point and the previous RRI that meets the requirement, and n is the number of intervals between the previous RRI that meets the requirement and the first RRI that meets the requirement after this point.

[0032] In some embodiments, determining signal quality specifically includes:

[0033] The quality of the detection signal within a single time window is judged, wherein the detection signal within the window that meets the second preset condition is judged as a signal that meets the quality requirements;

[0034] Signal quality is determined by the number of signals that are deemed to meet quality standards.

[0035] The second preset condition includes:

[0036] The amplitude of the resultant acceleration corresponding to the triaxial acceleration, the baseline of the resultant acceleration, the variance within the resultant acceleration window, the vibration amplitude of the green light PPG, the proportion of effective peak points after filtering that exceed 0.5 to 2 times the mean of the RRI sequence, and the proportion of RRI before anomaly processing that exceed 0.5 to 2 times the mean of the RRI sequence all meet the corresponding threshold requirements.

[0037] In some embodiments, the bandpass filtering of the RRI sequence after extension to obtain low-frequency and high-frequency energy values, and the extraction of frequency domain features based on the low-frequency and high-frequency energy values, specifically includes:

[0038] The RRI sequence is extended multiple times;

[0039] For the extended signal, the bandpass signals of [0.04, 0.15] Hz and [0.15, 0.4] Hz are extracted, and the low-frequency energy LF and high-frequency energy HF are obtained by calculating the sum of squares of the amplitudes of each passband signal.

[0040] The ratio of LF to HF, LF / HF, is calculated as a frequency domain feature;

[0041] The extended signal is g(n):

[0042] g(n)=h(n)*f(n%l)+(1-h(n))*f(ln%l);

[0043] In the above formula, f(*) is the original RRI sequence before extension, l is the length of the original RRI sequence, % is the modulo operation, n is generally selected to be up to 4 times l, and h(n) is calculated as follows:

[0044]

[0045] In the above formula, / / represents the rounding operation.

[0046] In some embodiments, the time-domain features include the mean of the RRI sequence, the root mean square (RMSSD) of the difference between adjacent normal cardiac cycles, the standard deviation (SDNN) of the time interval between heartbeats, and the standard deviation (SDANN) of the mean normal RR interval.

[0047] In some embodiments, calculating psychological stress based on the time-domain features, frequency-domain features, and signal quality includes:

[0048] Obtain the motion impact level, which characterizes the intensity of motion detected by the accelerometer;

[0049] Psychological stress is calculated based on the level of motion impact, time-domain characteristics, frequency-domain characteristics, and signal quality.

[0050] On the other hand, embodiments of this application disclose a psychological stress calculation system based on wearable devices, characterized in that it includes:

[0051] Memory, used to store programs;

[0052] A processor is used to load the program to execute the aforementioned method for calculating psychological stress based on a wearable device.

[0053] In calculating the frequency domain features of heart rate variability, this application embodiment uses an extension of the RRI (heart rate interval sequence signal) and a bandpass filtering method to extract relevant frequency band features. Compared with the traditional Fourier transform correlation method, this method has advantages such as less computational resource consumption and lower requirement for the number of RRI sequences. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a system module block diagram of an embodiment of this application.

[0056] Figure 2This is a flowchart of the psychological stress calculation method according to an embodiment of this application;

[0057] Figure 3 This is a flowchart of the signal processing steps in an embodiment of this application;

[0058] Figure 4 This is a flowchart of the RRI sequence extraction steps in an embodiment of this application;

[0059] Figure 5 This is a flowchart of the signal quality detection steps in an embodiment of this application;

[0060] Figure 6 This is a flowchart of the HRV feature calculation steps in an embodiment of this application. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0062] In the description of this invention, "several" means one or more, "multiple" means two or more, "greater than," "less than," and "exceeding" are understood to exclude the stated number, while "above," "below," and "within" are understood to include the stated number. The use of "first" and "second" in the description is merely for distinguishing technical features and should not be construed as indicating or implying relative importance, or implicitly indicating the number of indicated technical features, or implicitly indicating the order of the indicated technical features.

[0063] In the description of this invention, unless otherwise explicitly defined, terms such as "setup" should be interpreted broadly, and those skilled in the art can reasonably determine the specific meaning of the above terms in this invention in conjunction with the specific content of the technical solution.

[0064] In the description of this invention, the terms "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., refer to specific features or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0065] The overall block diagram of the system structure of this invention is as follows: Figure 1As shown, it mainly includes modules related to the host computer and modules related to the wearable device. The wearable device modules mainly include a signal acquisition module, a wear detection module, a psychological stress calibration module, and a psychological stress measurement module. S11 in the diagram is the host computer module, which includes a mobile APP, cloud services, and a gateway. Its main functions are user information storage and distribution, initiation of active psychological stress measurement, initiation of active psychological stress calibration, and statistics and display of measurement records. S12 is the device-side signal acquisition module, which is responsible for acquiring green light and infrared PPG signals (including but not limited to 128Hz), as well as three-axis acceleration signals (including but not limited to 25Hz). S13 is the wear detection module, which determines wearability based on the following two conditions: 1) whether the baseline value of the infrared PPG signal meets the threshold requirement; 2) whether a heart rate cycle exists in the green light PPG signal. If the above two requirements are met, it is in the wearing state; otherwise, it is not worn. In the unwearing state, the measurement or calibration process will be interrupted. S14 is the psychological stress calibration module, which is divided into active calibration and passive calibration. This module will adapt the algorithm for the current user based on the questionnaire calibration results, sleep quality, resting heart rate, and measurement signal characteristics during the active calibration process. S15 is the psychological stress calculation module, which is responsible for filtering the green light PPG signal collected by the signal acquisition module, extracting signal feature points, handling signal anomalies, and finally calculating the psychological stress value.

[0066] like Figure 2 As shown, the entire method for calculating psychological stress can be divided into the following steps:

[0067] S21, Algorithm initialization steps.

[0068] Specifically, this step initializes the user's stress level, calibrated stress value, resting heart rate, reference heart rate, characteristic coefficients, and other relevant parameters. The user's stress level and calibrated stress value are obtained from the calibration block, the resting heart rate value is obtained from daily measurement statistics, and the reference heart rate value and characteristic coefficients are obtained from statistics based on the user's personal information and categorized population. This step is an initialization step, completing the initial configuration before acquiring the test data. The test data includes a 128Hz PPG signal and a 25Hz ACC signal.

[0069] S22 is the signal processing step, which involves filtering the detected signal.

[0070] Specifically, such as Figure 3This step consists of the following steps: S31, bandpass filtering is performed on the incoming signal, including but not limited to using a third-order Butterworth bandpass filter to extract the passband signal of [0.3,8] Hz. This step removes low-frequency baseline offset and high-frequency noise. S32, differential filtering is performed on the bandpass filtered signal, including but not limited to first-order differential filtering. The main purpose of this operation is to reduce the interference of low-frequency components and transform the signal into a variable dimension. S33, mean filtering is performed on the differentially filtered signal to reduce the influence of high-frequency components in the signal on the selection of feature points.

[0071] S23 is the RRI extraction step, which obtains the RRI sequence from the filtered detection signal.

[0072] Specifically, such as Figure 4 This step consists of the following steps: S41 This step marks suspicious peak points in the filtered signal. Suspicious peak points must meet the following conditions: 1) The point is a maximum value point; 2) There is a minimum value point less than a threshold between the previous valid peak point and the current peak point, and the number of minimum values ​​meeting the requirements is less than the threshold; 3) The maximum value point is greater than the threshold; S42 If the time interval between the suspicious peak point and the previous valid peak point is less than the threshold 1, the peak point is considered an invalid peak point; otherwise, the peak point is considered a valid peak point. If the time interval is less than the threshold of 2, the time interval is considered a valid RRI, and the RRI sequence is updated; otherwise, it is considered that a missed detection occurred between the two peak points, the RRI is invalid, and no RRI update is performed. S43 performs outlier adjustment on the generated time interval sequence (RRI). This step first calculates the mean of the entire RRI sequence, and then performs a threshold judgment on each RRI value in the sequence. If the RRI is less than 0.7 times the RRI sequence mean or greater than 1.3 times the RRI sequence mean, then interpolation is performed on that RRI value.

[0073]

[0074] In the above formula, y represents the interpolated result, f1 is the previous RRI that meets the threshold requirement, f2 is the first RRI that meets the threshold requirement after the point, m is the number of intervals between the point and the previous RRI that meets the requirement, and n is the number of intervals between the previous RRI that meets the requirement and the first RRI that meets the requirement after the point. The RRI threshold requirements in this step include, but are not limited to, 0.7 and 1.3, and the interpolation method includes, but is not limited to, the linear interpolation described above.

[0075] S24 is the signal quality judgment step, which determines the signal quality.

[0076] Specifically, such as Figure 5This step consists of the following steps: S51. Signal quality is assessed for the signal within a single time window, which includes, but is not limited to, 6 seconds. The signal must meet the following requirements: 1) Is the amplitude of the resultant acceleration corresponding to the triaxial acceleration (ACC) less than the threshold 3? 2) Does the baseline of the resultant acceleration meet the threshold requirement? 3) Is the variance within the resultant acceleration time window less than the threshold? 4) Does the amplitude of the green light PPG vibration meet the threshold requirement? 5) Is the proportion of effective peak points exceeding 0.5-2 times the mean after filtering less than the threshold? 6) Is the proportion of RRI exceeding 0.5-2 times the mean of the RRI sequence before anomaly processing less than the threshold? If the above requirements are met, the signal quality within the time window is satisfactory; otherwise, the signal quality is unsatisfactory. S52. The entire measurement process takes, but is not limited to, 60 seconds. During the entire measurement process, signal quality assessment does not involve sliding window operations, allowing the acquisition of signal quality for multiple time windows and counting the number of time windows that meet the requirements. S53. If the number of signals meeting the requirements is less than the threshold 1, the signal quality level is determined. signal If the number of signals is greater than threshold 1 but less than threshold 2, the signal quality level is 1; otherwise, the signal quality level is 2.

[0077] S25 is the feature extraction step, which calculates several time-domain features of the RRI sequence; after the RRI sequence is extended, bandpass filtering is performed to obtain low-frequency energy values ​​and high-frequency energy values, and frequency-domain features are extracted based on the low-frequency energy values ​​and high-frequency energy values.

[0078] Specifically, such as Figure 6 The module first calculates the time-domain features of the RRI sequence in step S61, including the mean, root mean square (RMSSD) of the difference between adjacent normal cardiac cycles, standard deviation (SDNN) of the time interval between heartbeats, and standard deviation (SDANN) of the mean normal RR interval. Then, in step S62, the RRI sequence is extended multiple times, and the extended signal is g(n).

[0079] g(n)=h(n)*f(n%l)+(1-h(n))*f(ln%l)

[0080] In the above formula, f(*) is the original RRI sequence before extension, l is the length of the original RRI sequence, % is the modulo operation, n is generally selected to be up to 4 times l, and h(n) is calculated as follows:

[0081]

[0082] In the above formula, / / represents the rounding operation. This extension operation is essentially a time shift after even extension of the RRI sequence. Odd extension is not used here to avoid discontinuous signal transitions after extension. For the extended signal, bandpass signals of [0.04, 0.15] Hz and [0.15, 0.4] Hz are extracted. The low-frequency energy LF and high-frequency energy HF are obtained by calculating the sum of squares of the amplitudes of each passband signal. LF and HF are divided by the signal length to reduce the influence of different RRI signal lengths on this value. The ratio of LF to HF, LF / HF, is calculated. This frequency domain feature extraction method can greatly reduce the amount of computing resources used while ensuring feature accuracy. Taking the actual comparison data of the chip Apollo 3.0 as an example, the Fourier correlation spectrum method requires 2.5 times more flash memory, 6 times more RAM, and 25 times more overall computation. After obtaining the relevant features, step S63 will make a reasonable judgment on the orientation of each feature. If the contribution of a certain feature to the final pressure value exceeds the threshold range, the feature will be truncated.

[0083] S26 is the stress calculation step, which calculates psychological stress based on the level of motion influence, time domain characteristics, frequency domain characteristics, and signal quality.

[0084] Specifically, this step requires obtaining, in addition to heart rate variability characteristic values ​​and signal quality levels, the user's motion impact level.

[0085]

[0086] In the above formula, t represents the time interval between effective strenuous exercise and the measurement of psychological stress. Effective strenuous exercise is defined as the vibration amplitude exceeding threshold 1 and the duration exceeding threshold 2 detected by the accelerometer. When the exercise influence level is not 2, there is an effect of exercise on heart rate variability, and psychological stress is measured at this time. cur The calculation requires reference to the user's calibration values:

[0087] stress cur =γ*(∑θ) i *f i +level stress *ε+β*(hr rest -hr ref ))+(1-γ)*stress calib ;

[0088] In the above formula, θ i f represents the weights corresponding to each heart rate variability feature under the influence of exercise. i For heart rate variability characteristics, level stressThe pressure level is represented by ε, which is the pressure value correction between each pressure level, derived statistically. β is the heart rate calibration coefficient, which varies among different populations. hr rest This is the resting heart rate value, hr ref This refers to the reference heart rate value obtained by statistical analysis of this population at rest. calib This is a calibration value for the user's heart rate and stress.

[0089] In the above formula, γ represents the confidence level of this measurement.

[0090]

[0091] The test fails when the signal quality is 0. (This refers to the motion impact level.) sport When the value is 2, exercise has a relatively small impact on heart rate variability, and the formula for calculating psychological stress is:

[0092] stress cur =∑α i *f i +level stress *ε+β*(hr rest -hr ref );

[0093] Where θ i The weights are the values ​​corresponding to each heart rate variability feature under no-exercise conditions.

[0094] To address psychological stress values ​​affected by motion and signal quality, this invention returns a flag indicating that the value may be inaccurate, along with the stress value. The firmware's UI and the app will display the returned flag accordingly.

[0095] In summary, to reduce the resource consumption of the heart rate variability (HRV) stress assessment method, this invention employs an extension of the RRI (heart rate interval sequence) signal and a bandpass filtering method to extract relevant frequency band features when calculating the HRV frequency domain features. Compared with the traditional Fourier transform correlation method, this method has advantages such as lower computational resource consumption and lower requirement for the number of RRI sequences. To reduce the signal quality requirements of the HRV stress assessment method, this invention adopts a method of assigning different weights to different variability features based on different signal qualities, which increases the applicability of the algorithm to a certain extent. Since exercise has a significant impact on HRV-related features, to distinguish between HRV changes caused by exercise and those caused by psychological stress, this invention adds an exercise influence level feature, which increases the stability of the algorithm.

[0096] If the integrated units described in this application are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, 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 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 described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0097] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the appended claims.

Claims

1.A method for calculating mental stress based on a wearable device, characterized by, The method comprises the following steps: obtaining a detection signal; filtering the detection signal; obtaining an RRI sequence from the filtered detection signal; determining signal quality; calculating a plurality of time domain features of the RRI sequence; carrying out band-pass filtering on the RRI sequence after extension to obtain low-frequency energy and high-frequency energy, and extracting frequency domain features according to the low-frequency energy and the high-frequency energy; calculating psychological stress according to the time domain features, the frequency domain features and the signal quality; wherein the determination of signal quality specifically comprises: judging the quality of the detection signal in a single time window, wherein the detection signal in the window is determined to be a signal meeting the quality when the detection signal meets a second preset condition; determining the signal quality according to the number of signals determined to meet the quality; wherein the second preset condition comprises: the amplitude of the resultant acceleration corresponding to the three-axis acceleration, the baseline of the resultant acceleration, the variance of the resultant acceleration in the window, the vibration amplitude of the green PPG, the proportion of the filtered effective peak points exceeding 0.5-2 times the mean value of the RRI sequence, and the proportion of the RRI before abnormal processing exceeding 0.5-2 times the mean value of the RRI sequence all meet the corresponding threshold requirements. 2.The mental stress calculation method based on a wearable device of claim 1, wherein, The filtering of the detection signal comprises: extracting a band-pass signal in a preset range through a third-order Butterworth band-pass filter; carrying out differential filtering on the band-pass filtered signal; carrying out mean value filtering on the differential filtered signal. 3.The mental stress calculation method based on a wearable device of claim 1, wherein, The obtaining of the RRI sequence from the filtered detection signal comprises: labeling the peak points in the filtered detection signal that meet a first preset condition; judging whether the labeled peak points are invalid peak points; updating the RRI sequence according to the validity of the peak points; adjusting the abnormal values of the RRI sequence, interpolating the RRI values meeting the abnormal conditions to obtain the RRI sequence. 4.The mental stress calculation method based on a wearable device according to claim 3, wherein, The first preset condition is that the peak point is a maximum value point, there is a minimum value point between the previous valid peak point and the peak point, the number of minimum value points meeting the requirement is less than a second threshold, and the peak point is greater than a third threshold. The judgment of whether the labeled peak points are invalid peak points specifically comprises: if the time interval between the suspicious peak point and the previous valid peak point is less than a fourth threshold, the peak point is determined to be an invalid peak point, otherwise the peak point is determined to be a valid peak point. The updating of the RRI sequence according to the validity of the peak points specifically comprises: if the time interval is less than a fifth threshold, the time interval is determined to be a valid RRI, the RRI sequence is updated, otherwise it is determined that the RRI between the two peak points is missed, the RRI is determined to be an invalid value, and the RRI is not updated. 5.The mental stress calculation method based on a wearable device according to claim 4, wherein, The adjustment of the abnormal values of the RRI sequence, the interpolation of the RRI values meeting the abnormal conditions to obtain the RRI sequence specifically comprises: calculating the mean value of the entire RRI sequence, judging each RRI value in the RRI sequence, if the RRI value is less than 0.7 times or greater than 1.3 times the mean value of the RRI sequence, the RRI value is interpolated; wherein, ； In the above formula is the interpolated result, is the RRI before the point that meets the threshold requirement, is the RRI after the point that meets the threshold requirement, is the interval number between the point and the RRI before the point that meets the requirement, is the interval number between the RRI before the point that meets the requirement and the RRI after the point that meets the requirement. 6.The mental stress calculation method based on a wearable device of claim 1, wherein, The RRI sequence is extended and then band-pass filtered to obtain low-frequency energy and high-frequency energy, and frequency domain features are extracted according to the low-frequency energy and the high-frequency energy, specifically comprising: The RRI sequence is extended multiple times; For the extended signal, [0.04, 0.15] Hz and [0.15, 0.4] Hz band-pass signals are extracted, and the low-frequency energy LF and the high-frequency energy HF are obtained by calculating the sum of squares of the amplitudes of the band-pass signals; The ratio of LF to HF, LF / HF, is calculated as a frequency domain feature; wherein the extrapolated signal is : ； In the above formula is the original RRI sequence before extension, is the length of the original RRI sequence, where is the modulo operation, The general maximum is selected to be 4 times of the maximum of the original RRI sequence, The calculation is as follows: ； In the above formulae is a rounding operation. 7.The mental stress calculation method based on a wearable device of claim 1, wherein, The time domain features include the mean of the RRI sequence, the root mean square of the difference between adjacent normal cardiac cycles RMSSD, the standard deviation of the time interval between heartbeats SDNN, and the standard deviation of the average normal RR interval SDANN. 8.The mental stress calculation method based on a wearable device of claim 1, wherein, The psychological stress is calculated according to the time domain features, the frequency domain features, and the signal quality, comprising: An exercise influence level is obtained, which is used to represent the intensity of the motion detected by the acceleration sensor; The psychological stress is calculated according to the exercise influence level, the time domain features, the frequency domain features, and the signal quality. 9.A mental stress calculation system based on a wearable device, characterized by, Comprise: A memory for storing a program; A processor for loading the program to execute a psychological stress calculation method based on a wearable device according to any one of claims 1-8.

Images