Mental pressure analysis method and equipment based on pulse waves

Abstract

The invention provides a mental stress analysis method and equipment based on pulse waves. The method is applied to physiological signal processing. The method comprises the following steps: acquiring a pulse wave signal of a target object, wherein the pulse wave signal is a signal acquired by pulse wave acquisition equipment; extracting pulse rate change information from the pulse wave signal by using a window function with the length of 0, so as to obtain a forward pulse rate; reversely correcting a problem pulse rate in the forward pulse rate by adopting a path growth mode to obtain a target pulse rate; performing forward heartbeat point search on the target pulse rate, and performing reverse correction on the heartbeat point to obtain a heartbeat interval; calculating a heart rate variability parameter according to the heartbeat interval; and calculating the heart rate variability parameter by using a pre-constructed mental stress model to obtain a mental stress parameter. According to the invention, mental stress monitoring with low burden, high reliability and high accuracy is realized.

Description

Technical Field

[0001] This invention relates to the field of physiological signal processing, specifically to a method and device for analyzing mental stress based on pulse waves. Background Technology

[0002] Mental stress is a common and significant factor affecting people's physical and mental health in modern life. Accurate monitoring and analysis of an individual's mental stress level is of great importance for the prevention and intervention of mental illnesses.

[0003] Current mainstream methods are represented by electrocardiogram (ECG) detection: by attaching electrodes to the chest and limbs to collect cardiac electrical signals and analyzing parameters such as heart rate variability to assess stress. However, this requires the electrodes to be in direct contact with the skin, which increases the physical burden on the target (e.g., they are prone to loosening during exercise, leading to signal interruption), and is also susceptible to electromagnetic interference affecting signal quality. To address this limitation, non-contact devices are gradually being used. These devices do not require direct contact with the human body and mainly include technologies such as radar, WiFi, and mobile phone cameras.

[0004] However, heartbeat signals collected by radar and WiFi devices are easily interfered with by factors such as respiratory rate and body movement, resulting in insufficient data reliability. Mobile phone cameras rely on visual analysis to capture behavioral features such as facial expressions, but they are extremely sensitive to ambient light intensity and the amplitude of facial movements, similarly limiting the accuracy of data collection. All of these issues directly affect the reliability of mental stress detection results. Summary of the Invention

[0005] In view of this, the present invention provides a method for analyzing mental stress based on pulse waves, comprising: Acquire the pulse wave signal of the target object, wherein the pulse wave signal is a signal acquired by a pulse wave acquisition device; Using path growth method with a length of The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. ; The positive pulse rate is corrected in reverse using a path growth method. The problem pulse rate is used to obtain the target pulse rate; Based on the target pulse rate, the pulse wave signal is forward searched for the heartbeat point, and the heartbeat point is reversed to obtain the heartbeat interval. Calculate heart rate variability parameters based on the described heartbeat interval; The heart rate variability parameters were calculated using a pre-constructed mental stress model to obtain the mental stress parameters.

[0006] Optionally, the path growth method utilizes a length of The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. ,include: Determine the set of root nodes of the path in the pulse wave signal. Is it an empty set? If the path root node set If the set is empty, the frequency spectrum of the pulse wave signal within the current window is calculated using the time-frequency transformation method, and a selection is made from the spectrum. The peak points are used as the root nodes of the path at the current time t. and the root node of the path Add to the set of root nodes of the path In, and the path length is +1; If the path root node set If the set is non-empty, the spectrum of the pulse wave signal within the current window is calculated using the time-frequency transformation method, and frequency points are extracted from the spectrum. From the set of root nodes of the path Retrieve the root node of the path extracted from the previous time step t-1. Calculate the frequency point With the root node of the path The distance between them is selected if the distance is less than a distance threshold. Path root node As the path growth direction at the current time t, and the root node of the path Add to the set of root nodes of the path In, and the path length is +1; Determine whether the updated path length is greater than the distance threshold. ; If the updated path length is greater than the distance threshold Then, based on the updated set of root nodes of the path At each time step, a node is selected, and several candidate path growth directions are generated. The confidence score of each candidate path growth direction is calculated, and the node at the earliest time step of the candidate path growth direction with the highest confidence score is selected. As the node at the corresponding time, retain the candidate frequency in the candidate path that has the same starting point as the candidate path with the highest confidence growth direction, and delete the set of path root nodes. The root node at the earliest time step yields the positive pulse rate at the current time t. ; If the updated path length is less than or equal to the distance threshold Then, the positive pulse rate of the pulse wave signal at the next time step t+1 is extracted using a sliding window. .

[0007] Optionally, the positive pulse rate is corrected in reverse using a path growth method. The problem pulse rate is to obtain the target pulse rate, including: Positive pulse rate based on adjacent time points Based on the confidence level, extract the problem pulse rate; The problematic pulse rate is corrected by using a path-reverse growth method to obtain the reverse pulse rate. ; The positive pulse rate With the reverse pulse rate Compare and extract time periods where the difference is greater than the difference threshold. ; Based on each time period The positive pulse rate described in With the reverse pulse rate The target pulse rate is selected based on the confidence level.

[0008] Optionally, the step of forward searching for the heartbeat point based on the target pulse rate and reverse correcting the heartbeat point to obtain the heartbeat interval includes: Heartbeat points are extracted sequentially from the pulse wave signals according to time sequence to obtain a set of heartbeat points. ; Determine the set of heartbeat points Is it an empty set? If the set of heartbeat points If the set is empty, then the initial heartbeat point of the first signal cycle in the pulse wave signal is selected. and the initial heartbeat point Add to the set of heartbeat points middle; If the set of heartbeat points If it is a non-empty set, then based on the target pulse rate and the set of heartbeat points... Calculate the location of the first theoretical heartbeat. Select the location at the distance from the first theoretical heartbeat point. The most recent heartbeat point, get the current heartbeat point. and the current heart rate Add to the set of heartbeat points middle; For the updated set of heartbeats The heartbeat point in the middle is corrected in reverse; Based on the revised heartbeat set The time difference between adjacent heartbeats is used to obtain the heartbeat interval.

[0009] Optionally, the step of basing the target pulse rate on the set of heartbeats... Calculate the location of the first theoretical heartbeat. ,include: Calculate the adjacent values ​​of the target pulse rate Average pulse rate over time ; Extract the set of heartbeat points The heartbeat point in the final moments ; Based on the average pulse rate The heartbeat point and pulse wave sampling frequency The theoretical heartbeat point position was calculated to obtain the first theoretical heartbeat point position. .

[0010] Optionally, the set of heartbeat points The heart rate point is corrected in reverse, including: The set of heartbeat points Filter the time range of abnormal heartbeats and delete the set of heartbeats. Abnormal heartbeat points; Calculate the location of the second theoretical heartbeat point based on the stated time range. Select the location at the distance from the second theoretical heartbeat point. The most recent heartbeat point, get the current heartbeat point. The heartbeat points of the pulse wave signal are extracted and added to the heartbeat point set by iterating backwards according to the time range. In order to correct the set of heartbeat points .

[0011] Optionally, the step of calculating the second theoretical heartbeat point location based on the time range... ,include: The average pulse rate is calculated based on the pulse rate within the time range up to time Tw. ; Select the heartbeat point at the last moment within the time range. ; Based on the average pulse rate Heartbeat Point and pulse wave sampling frequency The theoretical heartbeat point position is calculated to obtain the second theoretical heartbeat point position. .

[0012] Optionally, the mental stress parameters include at least one of stress index, fatigue index, stress resistance, autonomic nervous system activity, and autonomic nervous system balance.

[0013] Optionally, after acquiring the pulse wave signal of the target object, the method further includes: The pulse wave signal is bandpass filtered to obtain the initial pulse wave signal. ; The initial pulse wave signal is displayed using a sliding window method. Perform matrix processing to obtain a data matrix. ; For the data matrix Principal component extraction was performed to obtain principal component data signals; The principal component data signal is smoothed and denoised to obtain a standard pulse wave signal.

[0014] A second aspect of the present invention provides a pulse wave-based mental stress analysis device, the device comprising: a processor and a memory connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to cause the processor to perform the aforementioned pulse wave-based mental stress analysis method.

[0015] This invention first acquires the pulse wave signal of the target object using a pulse wave acquisition device. This acquisition method is convenient to operate and significantly improves the comfort and compliance of the target object; then, by utilizing a length of... The window function dynamically slides across the pulse wave signal, analyzing signal features segment by segment and extracting continuous information reflecting pulse rate changes to obtain the positive pulse rate. Then, the positive pulse rate is corrected by reverse correction. This invention addresses the impact of abnormal signals on the accuracy of results by obtaining a more accurate target pulse rate. Then, based on the target pulse rate, a forward search is performed on the heartbeat points in the pulse wave signal, and abnormal points are corrected in reverse, ultimately calculating the accurate time interval between adjacent heartbeat points (inter-heartbeat interval). Based on the corrected inter-heartbeat interval, heart rate variability parameters are calculated. These parameters directly reflect the state of autonomic nervous system function and are the core physiological basis for assessing mental stress. Finally, the heart rate variability parameters are input into a pre-trained neural network or multinomial mental stress model, mapping them into interpretable psychological state indicators such as stress indices. This achieves a precise conversion from physiological signals to psychological stress, supporting long-term monitoring of stress changes in the human body, such as heart rate or cardiovascular targets. This invention, through multi-step correction of pulse wave signals within a path growth framework, combined with the precise calculation of a pre-trained mental stress model, achieves low-burden, high-reliability, and accurate mental stress monitoring based on contactless acquisition using wearable devices. Attached Figure Description

[0016] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is an application scenario diagram of the pulse wave-based mental stress analysis device in this embodiment of the invention; Figure 2 This is a flowchart of a pulse wave-based mental stress analysis method according to an embodiment of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] In the description of this invention, it should be noted that the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0020] Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0021] like Figure 1 As shown in the figure, an application scenario of a pulse wave-based mental stress analysis device provided by an embodiment of the present invention is as follows: This invention comprises a data acquisition end and a signal processing end. The data acquisition end is a pulse wave acquisition device 11. Acquisition devices include, but are not limited to, watches, bracelets, rings, ear clips, etc., capable of acquiring human pulse waves. The pulse wave acquisition device is worn on a suitable part of the subject's body. The pulse wave acquisition device acquires the target object's pulse wave data and transmits it to the signal processing end 12 for analysis. Transmission methods include, but are not limited to, networks, WiFi, Bluetooth, etc. After the human pulse wave data acquisition is completed, the signal processing end 12 analyzes all the received pulse wave data to obtain information such as heart rate interval and mental stress.

[0022] like Figure 2 As shown, this embodiment of the invention provides a method for analyzing mental stress based on pulse waves. This method is applied to a signal processing terminal and specifically includes: S1, acquire the pulse wave signal of the target object. The pulse wave signal is the signal acquired by the pulse wave acquisition device.

[0023] Taking a wristwatch as an example, a pulse wave acquisition device, such as Figure 1 In the process, the pulse wave acquisition device 11 is worn on the wrist of the person being tested, and the pulse data collected by the watch for a certain period of time will be transmitted to the signal processing terminal 12 for data processing.

[0024] S2, using a path growth method with a length of... The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. .

[0025] This embodiment is based on the pulse wave signal obtained in step S1, and uses a path growth framework to progressively extract pulse rate change information. This can be achieved using a sliding window technique (the window function length is...). By dynamically sliding across the pulse wave signal and analyzing its characteristics segment by segment, continuous information reflecting pulse rate changes is extracted in a forward (time-sequential) manner to obtain the forward pulse rate. .

[0026] S3 uses a path growth method to reverse-correct the positive pulse rate. The problem pulse rate is obtained by finding the target pulse rate.

[0027] By using the reverse verification logic of path growth, the positive pulse rate is verified. Anomaly detection and correction are performed to ultimately obtain a more accurate target pulse rate.

[0028] S4. Based on the target pulse rate, perform a forward search for the heartbeat point on the pulse wave signal and perform a reverse correction on the heartbeat point to obtain the heartbeat interval.

[0029] This embodiment calculates the heartbeat interval (the time interval between adjacent heartbeats) by forward searching the pulse wave signal (from front to back), such as the pulse wave peak point, and reverse correcting abnormal points.

[0030] S5 calculates heart rate variability parameters based on the heart rate interval.

[0031] Based on the heart rate interval, heart rate variability parameters can be calculated, such as time-domain parameters like the standard deviation of the heart rate interval (SDNN), the root mean square of the difference between consecutive heart rate intervals (RMSSD), and the standard deviation of the difference between consecutive heart rate intervals (SDSD); frequency-domain parameters like the total energy (TP) of all frequency components in the spectrum, the low-frequency energy (LF) in the spectrum, and the high-frequency energy (HF) in the spectrum; and geometric parameters like the dispersion along the short axis (SD1) and the dispersion along the long axis (SD2) in the Poincaré scatter plot. The specific calculation methods are as follows: , , , , , , , , , , , Where, hi represents the inter-heartbeat period. For difference operations, For mean calculation, For square root operation, For standard deviation calculation, For absolute value operations, hif represents the frequency domain distribution of the inter-heartbeat period, and f1, f2, and f3 are the frequency domain subscripts for extremely low frequency, low frequency, and high frequency, respectively.

[0032] S6. The pre-built mental stress model is used to calculate the heart rate variability parameters to obtain the mental stress parameters.

[0033] The mental stress model is pre-trained based on heart rate variability parameters, and its specific structure can be a neural network, a polynomial structure, etc.

[0034] This embodiment first acquires the pulse wave signal of the target object using a pulse wave acquisition device. This acquisition method is convenient to operate and significantly improves the comfort and compliance of the target object; then, by utilizing a length of... The window function dynamically slides across the pulse wave signal, analyzing signal features segment by segment and extracting continuous information reflecting pulse rate changes to obtain the positive pulse rate. Then, the positive pulse rate is corrected by reverse correction. This invention addresses the impact of abnormal signals on the accuracy of results by obtaining a more accurate target pulse rate. Then, based on the target pulse rate, a forward search is performed on the heartbeat points in the pulse wave signal, and abnormal points are corrected in reverse, ultimately calculating the accurate time interval between adjacent heartbeat points (inter-heartbeat interval). Based on the corrected inter-heartbeat interval, heart rate variability parameters are calculated. These parameters directly reflect the state of autonomic nervous system function and are the core physiological basis for assessing mental stress. Finally, the heart rate variability parameters are input into a pre-trained neural network or multinomial mental stress model, mapping them into interpretable psychological state indicators such as stress indices. This achieves a precise conversion from physiological signals to psychological stress, supporting long-term monitoring of stress changes in the human body, such as heart rate or cardiovascular targets. This invention, through multi-step correction of pulse wave signals within a path growth framework, combined with the precise calculation of a pre-trained mental stress model, achieves low-burden, high-reliability, and accurate mental stress monitoring based on contactless acquisition using wearable devices.

[0035] In some optional embodiments of this example, in step S6, the pre-constructed mental stress model is used to calculate the heart rate variability parameters to obtain mental stress parameters, including at least one of stress index, fatigue index, stress resistance, autonomic nervous system activity, and autonomic nervous system balance.

[0036] In some optional embodiments of this example, step S2 uses path growth to utilize a length of The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. Specifically, it includes: S21, Determine the set of root nodes in the pulse wave signal. Is it an empty set? If it is the set of root nodes of the path? If the set is empty, then proceed to step S22. If the set of root nodes of the path is empty... If it is a non-empty set, then proceed to step S23.

[0037] This process is a loop. The root node of the path at the first moment of the sliding window processing is an empty set, so the operation of step S22 is executed. The root node of the path at subsequent moments is a non-empty set, so the operation of step S23 is executed.

[0038] S22, Calculate the spectrum of the pulse wave signal within the current window using the time-frequency transformation method, and select from the spectrum... The peak points are used as the root nodes of the path at the current time t. and the root node of the path Add to the path root node set In, and the path length is +1.

[0039] The current time t is the window size for each sliding window processing step. At any given moment within the timeframe, this could be an intermediate moment. During the initial processing (the first moment of the loop), the set of root nodes of the path... If empty, the current window spectrum needs to be calculated using time-frequency transformation and then selected. The peak points are used as the initial root node. Add to set and path length Add 1; This is because there is no historical root node in the initial state, and the initial node needs to be extracted through the spectral features of the current window to provide the necessary starting point data for subsequent path growth.

[0040] S23, Calculate the spectrum of the pulse wave signal within the current window using the time-frequency transformation method, and extract frequency points from the spectrum. From the set of root nodes of the path Retrieve the root node of the path extracted from the previous time step t-1. Calculate frequency points With the root node of the path The distance between them is selected if the distance is less than a distance threshold. Path root node As the path growth direction at the current time t, and the root node of the path... Add to the path root node set In, and the path length is +1.

[0041] Extracting frequency points Frequency points within preset conditions can be extracted based on energy or peak width. This embodiment uses the path root node set. When the set is non-empty, frequency points of the current window spectrum are extracted by time-frequency transformation. The distance between the frequency points and the root node of the path at the previous time step is calculated. The point with the smallest distance is selected as the current root node, added to the set, and the path length is extended. The purpose is to use the spatial proximity (smallest distance) of frequency points at adjacent time steps to ensure the continuity of path growth. Since the frequency characteristics of the real pulse wave signal usually do not fluctuate drastically in a short period of time, the frequency point with the smallest distance is more likely to belong to the continuation of the same physiological process, rather than an isolated point caused by noise or random interference. This avoids the path being interrupted by accidental errors and ensures that subsequent analysis is based on stable signal characteristics.

[0042] S24, Determine the path length Is it greater than the distance threshold? If the path length Greater than the distance threshold Then proceed to step S25, if the path length Less than or equal to the distance threshold Then proceed to step S26.

[0043] After each sliding window operation, the updated path length needs to be determined. Does it exceed the distance threshold? This is to control the length of the root node set and avoid data redundancy or noise accumulation.

[0044] S25, based on the updated set of root nodes of the path At each time step, a node is selected, and several candidate path growth directions are generated. The confidence score of each candidate path growth direction is calculated, and the node at the earliest time step of the candidate path growth direction with the highest confidence score is selected. As the node at the corresponding time, retain the candidate frequency in the candidate path that has the same starting point as the candidate path with the highest confidence growth direction, and delete the set of path root nodes. The root node at the earliest time step yields the positive pulse rate at the current time t. .

[0045] In path length Greater than the distance threshold At the same time, by generating multiple candidate paths and calculating confidence, the optimal growth direction is selected, redundant old nodes (such as the root node at the earliest moment) are deleted, and high-reliability nodes are retained to ensure that the pulse rate calculation is based on the most stable physiological signal characteristics, thereby improving accuracy.

[0046] S26, Sliding window extracts the positive pulse rate of the pulse wave signal at the next time step t+1. .

[0047] The updated path length did not exceed the distance threshold. At that time, the sliding window extracts the pulse wave signal at the next time t+1, and continues to execute the positive pulse rate calculation in steps S21-S26.

[0048] This embodiment achieves effective processing of pulse wave signals through phased root node management: First, it determines whether the root node set is empty; if empty, it initializes with spectral peak values. If not empty, it filters the growth direction based on the distance between the spectral frequency point and the previous node. After updating the path length, if the updated path length exceeds a threshold, high-confidence paths are selected and nodes are simplified; if the updated path length does not exceed the threshold, a sliding window is used for advancement, ultimately outputting a positive pulse rate. Through the synergistic effect of initialization assurance, dynamic growth correction, redundancy control, and real-time advancement, an accurate and stable positive pulse rate is ultimately output, providing a high-quality physiological signal foundation for subsequent psychological stress analysis.

[0049] In some optional embodiments of this example, step S3 uses path growth to reverse-correct the positive pulse rate. The problem pulse rate is to obtain the target pulse rate, which specifically includes: S31, based on the positive pulse rate of adjacent time points The confidence level is used to extract the problem pulse rate.

[0050] Calculate the positive pulse rate at adjacent time points The confidence levels between the points are used to extract the time periods with lower confidence levels (below the confidence threshold) as the problem pulse rate.

[0051] S32, using a path reverse growth method to correct the problem pulse rate, obtain the reverse pulse rate. .

[0052] For each problematic pulse rate segment, reverse correction is performed. Referring to the method shown in step S2, reverse pulse rate is extracted using path reverse growth. .

[0053] S33, positive pulse rate With reverse pulse rate Compare and extract time periods where the difference is greater than the difference threshold. .

[0054] Compare positive pulse rates With reverse pulse rate The results extracted the time periods where the two differed significantly.

[0055] S34, based on various time periods Neutral pulse rate With reverse pulse rate Based on the confidence level, the target pulse rate is selected.

[0056] Calculate the positive pulse rate for each time period. With reverse pulse rate The confidence level is determined, and the pulse rate with the highest confidence level is selected as the final target pulse rate.

[0057] This embodiment filters problematic pulse rates by assessing the confidence level of positive pulse rates at adjacent time points; then corrects the problematic pulse rates using a path-reverse growth method; finally, it compares the positive pulse rates. With reverse pulse rate And extract the difference time periods. Finally, the target pulse rate is determined by combining the confidence scores of both methods during the difference period. Problem pulse rates are processed using two different calculation methods (forward and reverse growth), and the confidence scores of both methods are combined during the difference period. This multi-directional processing of problem pulse rates effectively improves the accuracy of pulse rate extraction.

[0058] In some optional embodiments of this example, step S4 involves forward searching for the heartbeat point based on the target pulse rate and reverse correcting the heartbeat point to obtain the heartbeat interval. Specifically, this includes: S41, extract heartbeat points from the pulse wave signal sequentially according to time sequence to obtain a set of heartbeat points. .

[0059] Heartbeat points are extracted in chronological order, from front to back. These heartbeat points can be the highest point, lowest point, or zero-crossing point of the pulse wave signal.

[0060] S42, Determine the set of heartbeat points Is it an empty set, if it is the heartbeat point set? If it is an empty set, then proceed to step S43. If the heartbeat point set is empty... If it is a non-empty set, then proceed to step S44.

[0061] S43, Select the initial heartbeat point of the first signal cycle in the pulse wave signal. and the initial heartbeat point Add to heartbeat set middle.

[0062] If no heartbeat point is extracted in step S41, it means that the heartbeat point to be extracted is the initial point. Select the initial heartbeat point within the first pulse wave signal cycle. And the initial heartbeat point Add to heartbeat set middle.

[0063] S44, based on the target pulse rate and heart rate set Calculate the location of the first theoretical heartbeat. Select the location at the distance from the first theoretical heartbeat point The most recent heartbeat point, get the current heartbeat point. and the current heart rate Add to heartbeat set middle.

[0064] If a heartbeat point is extracted in step S41, then heartbeat points other than the initial point need to be extracted, and the position P̂ of the first theoretical heartbeat point is calculated based on the target pulse rate. t， Select a set of heartbeat points The closest location to the first theoretical heartbeat point The heartbeat point, and add it to the heartbeat point set. .

[0065] S45, for the updated heartbeat set The heartbeat point in the middle is corrected in reverse.

[0066] For the updated set of heartbeats The heartbeat points in the data are verified and adjusted in reverse (such as removing outliers and calibrating positions) to complete the reverse correction.

[0067] S46, based on the corrected heartbeat set The time difference between adjacent heartbeats is used to obtain the interval between heartbeats.

[0068] This embodiment extracts heartbeat points in chronological order to form a heartbeat point set. Then determine the set of heartbeat points. Is it empty; if it is the heartbeat point set If the set of heartbeats is empty, then fill in the initial heartbeat points; if the heartbeat point set is empty... If the heartbeat point is not empty, it is calibrated by matching the theoretical position (to improve accuracy); then the updated heartbeat point set is corrected in reverse. The heartbeat point; finally, the time difference between adjacent points is calculated to obtain the heartbeat interval. Through the set of heartbeat points... The judgment and initial supplementation solved problems such as empty sets and initial missing values ​​in the heartbeat point set. Furthermore, the accuracy of the heartbeat points was improved through theoretical matching and reverse correction. Finally, the heartbeat interval was reliably extracted, providing high-quality input for subsequent interval-based mental stress analysis.

[0069] In some optional embodiments of this example, step S44 is based on the target pulse rate and heart rate set. Calculate the location of the first theoretical heartbeat. Specifically, it includes: S441, Calculate adjacent values ​​in the target pulse rate Average pulse rate over time .

[0070] S442, Extract the heartbeat set The heartbeat point in the final moments .

[0071] S443, based on average pulse rate Heartbeat Point and pulse wave sampling frequency The theoretical heartbeat point position was calculated to obtain the first theoretical heartbeat point position. .

[0072] For example, the location of the first theoretical heartbeat point is calculated using the following method. : , This embodiment calculates adjacent... The average pulse rate over a given time period smooths out short-term fluctuations, resulting in a stable baseline average pulse rate. Then extract the most recent heartbeat. Maintaining timeline continuity, the theoretical heartbeat location is calculated by combining the average pulse rate, the last heartbeat point, and the sampling frequency. Through the coordinated operation of smoothing the pulse rate and anchoring to the most recent heartbeat point, the interference of instantaneous pulse rate fluctuations on the theoretical location is avoided, while maintaining time continuity using the latest heartbeat point, ultimately yielding a theoretical heartbeat location that better conforms to physiological patterns. This significantly improves the accuracy of heartbeat location.

[0073] In some optional embodiments of this example, step S45 involves updating the set of heartbeat points. The heart rate point is corrected in reverse, specifically including: S451, at the heartbeat point set Filter the time range of abnormal heartbeats and delete the set of heartbeats. Abnormal heartbeat points in the heart.

[0074] Abnormal heartbeat points are those that do not conform to the normal heartbeat pattern, such as a significant difference in heart rate between two consecutive heartbeats exceeding a certain threshold, like jumping from 30 bpm to 300 bpm. Abnormal heartbeat points within this range are removed to resolve the issue of residual error points.

[0075] S452, Calculate the location of the second theoretical heartbeat point based on the time range. Select the location at the distance from the second theoretical heartbeat point. The most recent heartbeat point, get the current heartbeat point. The heartbeat points of the pulse wave signal are extracted and added to the heartbeat point set by iterating backwards according to the time range. In order to correct the set of heartbeat points .

[0076] Assuming the time range is t1 to t2, the process is reversed, for example, starting from t2, the heartbeat points from which the pulse wave signal is extracted are added to the heartbeat point set. In the middle, calculate the location of the second theoretical heartbeat point. Select the location at the distance from the second theoretical heartbeat point The most recent heartbeat point, get the current heartbeat point. This process continues until the heartbeat point at time t1 is obtained. In this embodiment, for the missing time range after removing outliers, the theoretical heartbeat point position is calculated based on information such as the average pulse rate. The effective heartbeat point closest to this position is selected as the current correct point, and the pulse wave signal is re-detected and corrected along the time backward to ensure the accuracy of the heartbeat points in the set.

[0077] This embodiment resolves residual errors by removing outliers, and then corrects the abnormal heartbeat points extracted in the forward process through theoretical calculation and reverse traversal. The two work together to remove noise interference and correct abnormal heartbeat points, significantly improving the completeness and accuracy of the heartbeat point set, and providing a more reliable physiological signal basis for subsequent heartbeat interval calculation.

[0078] In some optional embodiments of this example, step S452 calculates the location of the second theoretical heartbeat point based on the time range. Specifically, it includes: S4521, calculate the average pulse rate based on the pulse rate within the time range up to time Tw, and obtain the average pulse rate. .

[0079] If we iterate in reverse over the time range t1~t2, we start from t2 and calculate the average pulse rate during the time interval Tw before t2. .

[0080] S4522, Select the heartbeat point at the last moment within the time range. .

[0081] Then select the heartbeat point at the last moment t2 in the range t1~t2. .

[0082] S4523, based on average pulse rate Heartbeat Point and pulse wave sampling frequency The theoretical heartbeat point position is calculated to obtain the second theoretical heartbeat point position. .

[0083] For example, the location of the second theoretical heartbeat point is calculated using the following method. : , This embodiment calculates the average pulse rate at time Tw after the time range to smooth out short-term fluctuations and obtain a stable baseline average pulse rate. (Reduce transient noise interference); then select the heartbeat point at the last moment of the time range. As a reference, the theoretical heartbeat location is calculated by combining the average pulse rate, the last heartbeat point, and the sampling frequency using a formula. This allows for the acquisition of normal heartbeat points within the abnormal time range, ensuring a complete set of heartbeat points. The integrity and accuracy of the data provide a reliable physiological signal basis for subsequent calculations of the heartbeat interval.

[0084] In some optional embodiments of this example, after acquiring the pulse wave signal of the target object, step S1 further includes: Step 1: Perform bandpass filtering on the pulse wave signal to obtain the initial pulse wave signal. .

[0085] In this embodiment, bandpass filtering can be performed using an IIR filter or an FIR filter.

[0086] Step two, use a sliding window to capture the initial pulse wave signal. Perform matrix processing to obtain a data matrix. .

[0087] For example, the data matrix is ​​calculated using the following method. : , in, The length of the sliding window. Step size, Indicates the initial pulse wave signal The signal length.

[0088] Step 3, process the data matrix Principal component extraction was performed to obtain principal component data signals.

[0089] Specifically, firstly, the data matrix is ​​processed using the following formula. Perform singular value decomposition: , in, The singular value decomposition function is a well-known and publicly disclosed method in this field. It is a left singular value matrix. It is a right singular value matrix. and All are unitary matrices. It is a singular value matrix, and its diagonal elements are eigenvalues.

[0090] Then, the singular value matrix is ​​extracted using the following formula. Eigenvalues ​​in: , And select its k largest eigenvalues ​​using the following formula. This ensures that the sum of its eigenvalues ​​accounts for a proportion greater than the eigenvalue threshold. : , in, It is the set of eigenvalues.

[0091] Next, extract the following formula with... The corresponding k sets of feature vectors are used to recover the principal component data matrix. : , Finally, the principal component data matrix is ​​converted into principal component pulse wave signals using the time-frequency transformation method, as shown in the following formula: , The time-frequency transformation method converts the principal component data matrix into principal component data signals according to the following formula: , Where F is the frequency domain transformation function, and IF is the inverse frequency domain transformation method. The frequency domain transformation function can be Fourier transform, music, Welch transform, STRFT transform, etc., and the inverse frequency domain transformation method is the inverse transform corresponding to the frequency domain transformation function.

[0092] Step four: Smooth and denoise the principal component data signal to obtain the standard pulse wave signal.

[0093] Specifically, assuming the width of the denoising window is 2m+1, and the i-th data point of the principal component data signal is... The corresponding time is The data points within the window are fitted using an (n-1)th degree polynomial, i.e.: , in, , , ... These correspond to the weights of each term in the polynomial.

[0094] For ease of calculation, the above formula can be expanded into matrix form as follows: , The expanded formula is expressed as a matrix. The form is: , The solution obtained using the least squares method is as follows: , , in, express The least-multiplier estimate, For use The calculated fitted data vector.

[0095] The standard pulse wave signal is obtained by smoothing and denoising the principal component data signal using the two formulas obtained from the solution.

[0096] This embodiment removes noise interference from the pulse wave signal using bandpass filtering while retaining effective frequency components. Then, sliding window matrixing converts the continuous signal into local data blocks for easier batch analysis. Principal component extraction separates and retains principal components reflecting physiological characteristics (components corresponding to large eigenvalues) while suppressing secondary noise. Polynomial smoothing further optimizes the principal component signal and eliminates residual fluctuations. These steps work synergistically to improve the signal-to-noise ratio while preserving key physiological characteristics of the pulse wave (such as periodicity and amplitude variation), ultimately yielding a stable and reliable standard pulse wave signal. This provides a high-quality input data foundation for subsequent pulse wave-based mental stress analysis and calculations.

[0097] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0098] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0099] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0100] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0101] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for analyzing mental stress based on pulse waves, characterized in that, include: Acquire the pulse wave signal of the target object, wherein the pulse wave signal is a signal acquired by a pulse wave acquisition device; Using path growth method with a length of The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. ; The positive pulse rate is corrected in reverse using a path growth method. The problem pulse rate is used to obtain the target pulse rate; Based on the target pulse rate, the pulse wave signal is forward searched for the heartbeat point, and the heartbeat point is reversed to obtain the heartbeat interval. Calculate heart rate variability parameters based on the described heartbeat interval; The heart rate variability parameters were calculated using a pre-constructed mental stress model to obtain the mental stress parameters.

2. The method according to claim 1, characterized in that, The path growth method utilizes a length of... The window function is used to extract pulse rate change information from the pulse wave signal to obtain the positive pulse rate. ,include: Determine the set of root nodes of the path in the pulse wave signal. Is it an empty set? If the path root node set If the set is empty, the frequency spectrum of the pulse wave signal within the current window is calculated using the time-frequency transformation method, and a selection is made from the spectrum. The peak points are used as the root nodes of the path at the current time t. and the root node of the path Add to the set of root nodes of the path In, and the path length is +1; If the path root node set If the set is non-empty, the spectrum of the pulse wave signal within the current window is calculated using the time-frequency transformation method, and frequency points are extracted from the spectrum. From the set of root nodes of the path Retrieve the root node of the path extracted from the previous time step t-1. Calculate the frequency point With the root node of the path The distance between them is selected if the distance is less than a distance threshold. Path root node As the path growth direction at the current time t, and the root node of the path Add to the set of root nodes of the path In, and the path length is +1; Determine whether the updated path length is greater than the distance threshold. ; If the updated path length is greater than the distance threshold Then, based on the updated set of root nodes of the path At each time step, a node is selected, and several candidate path growth directions are generated. The confidence score of each candidate path growth direction is calculated, and the node at the earliest time step of the candidate path growth direction with the highest confidence score is selected. As the node at the corresponding time, retain the candidate frequency in the candidate path that has the same starting point as the candidate path with the highest confidence growth direction, and delete the set of path root nodes. The root node at the earliest time step yields the positive pulse rate at the current time t. ; If the updated path length is less than or equal to the distance threshold Then, the positive pulse rate of the pulse wave signal at the next time step t+1 is extracted using a sliding window. .

3. The method according to claim 1, characterized in that, The positive pulse rate is corrected in reverse using a path growth method. The problem pulse rate is to obtain the target pulse rate, including: Positive pulse rate based on adjacent time points Based on the confidence level, extract the problem pulse rate; The problematic pulse rate is corrected by using a path-reverse growth method to obtain the reverse pulse rate. ; The positive pulse rate With the reverse pulse rate Compare and extract time periods where the difference is greater than the difference threshold. ; Based on each time period The positive pulse rate described in With the reverse pulse rate The target pulse rate is selected based on the confidence level.

4. The method according to claim 1, characterized in that, The step of forward searching for the heartbeat point based on the target pulse rate and reverse correcting the heartbeat point to obtain the heartbeat interval includes: Heartbeat points are extracted sequentially from the pulse wave signals according to time sequence to obtain a set of heartbeat points. ; Determine the set of heartbeat points Is it an empty set? If the set of heartbeat points If the set is empty, then the initial heartbeat point of the first signal cycle in the pulse wave signal is selected. and the initial heartbeat point Add to the set of heartbeat points middle; If the set of heartbeat points If it is a non-empty set, then based on the target pulse rate and the set of heartbeat points... Calculate the location of the first theoretical heartbeat. Select the location at the distance from the first theoretical heartbeat point. The most recent heartbeat point, get the current heartbeat point. and the current heart rate Add to the set of heartbeat points middle; For the updated set of heartbeats The heartbeat point in the middle is corrected in reverse; Based on the revised heartbeat set The time difference between adjacent heartbeats is used to obtain the heartbeat interval.

5. The method according to claim 4, characterized in that, The target pulse rate and the set of heartbeats are used. Calculate the location of the first theoretical heartbeat. ,include: Calculate the adjacent values ​​of the target pulse rate Average pulse rate over time ; Extract the set of heartbeat points The heartbeat point in the final moments ; Based on the average pulse rate The heartbeat point and pulse wave sampling frequency The theoretical heartbeat point position was calculated to obtain the first theoretical heartbeat point position. .

6. The method according to claim 4, characterized in that, The set of heartbeat points The heart rate point is corrected in reverse, including: The set of heartbeat points Filter the time range of abnormal heartbeats and delete the set of heartbeats. Abnormal heartbeat points; Calculate the location of the second theoretical heartbeat point based on the stated time range. Select the location at the distance from the second theoretical heartbeat point. The most recent heartbeat point, get the current heartbeat point. The heartbeat points of the pulse wave signal are extracted and added to the heartbeat point set by iterating backwards according to the time range. In order to correct the set of heartbeat points .

7. The method according to claim 6, characterized in that, The second theoretical heartbeat point position is calculated based on the time range. ,include: The average pulse rate is calculated based on the pulse rate within the time range up to time Tw. ; Select the heartbeat point at the last moment within the time range. ; Based on the average pulse rate Heartbeat Point and pulse wave sampling frequency The theoretical heartbeat point position is calculated to obtain the second theoretical heartbeat point position. .

8. The method according to claim 1, characterized in that, The mental stress parameters include at least one of the following: stress index, fatigue index, stress resistance, autonomic nervous system activity, and autonomic nervous system balance.

9. The method according to claim 1, characterized in that, After acquiring the pulse wave signal of the target object, the process also includes: The pulse wave signal is bandpass filtered to obtain the initial pulse wave signal. ; The initial pulse wave signal is displayed using a sliding window method. Perform matrix processing to obtain a data matrix. ; For the data matrix Principal component extraction was performed to obtain principal component data signals; The principal component data signal is smoothed and denoised to obtain a standard pulse wave signal.

10. A mental stress analysis device based on pulse waves, characterized in that, include: A processor and a memory connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to cause the processor to perform the pulse wave-based mental stress analysis method as described in any one of claims 1-9.

Images