Stress detection device, system, and method for detecting mental stress in a person.

Abstract

The present invention relates to a stress detection system, device and method for detecting mental stress in a person. The system 1, 2 comprises an activity sensor 20 configured to acquire activity information related to the person's activity, a heart rate sensor 30 configured to acquire heart rate information indicating or enabling calculation of the person's current heart rate, and a stress detection device 10, 40, 60 for detecting the person's mental stress based on the acquired activity information and the acquired heart rate information. According to this, a change in the person's posture is taken into account and the calculation of the calculated basal heart rate component and / or the basal heart rate component is adapted when a change in posture is detected.

Description

[Technical Field]

[0001] The present invention relates to a stress detection device, system, and method for detecting mental stress in a person. [Background technology]

[0002] To enable personalized programs for stress reduction, it is necessary to measure and monitor an individual's stress level. One method of measuring stress is through questionnaires, which have the advantage of obtaining contextual information, such as the (perceived) causes of stress, current coping strategies, and the (perceived) impact of stress on daily life, and for which validated versions are available. However, questionnaires have the disadvantage of providing subjective data, where the results depend on how the respondent feels at the time the question is asked, making them inconvenient for continuous monitoring. Therefore, it would be desirable to have (and may have) objective measures of stress.

[0003] Stress has been shown to trigger a variety of measurable physiological responses, which can be classified into short-term and long-term responses. While long-term stress can have adverse effects on health, monitoring short-term stress is also of great interest in providing personalized programs for stress reduction. This can provide users with insights into their own stress patterns and help them reduce or prevent long-term stress by changing their daily behaviors.

[0004] The literature identifies several physiological parameters associated with stress, particularly heart rate (HR), heart rate variability (HRV), blood pressure (BP), and cortisol levels. Of these parameters, currently only HR can be measured consistently and accurately without noticeable interference. In addition to stress, heart rate is influenced by many other factors, including physical (exercise) activity, sleep, posture, circadian rhythm, temperature, dehydration, food, caffeine, nicotine, and alcohol.

[0005] Mental stress is a type of stress that arises from how a person perceives events in their external or internal environment, resulting in psychological experiences of distress and anxiety. Mental stress is often accompanied by physiological responses. [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] The object of the present invention is to provide an apparatus, system, and method that can detect and preferably quantify human mental stress with high reliability. [Means for solving the problem]

[0007] In a first aspect of the present invention, a stress detection device for detecting a person's mental stress is provided, the device: - Activity input unit configured to acquire activity information related to human activity; - A heart rate input unit configured to acquire heart rate information that enables the display or calculation of a person's current heart rate; - A processing unit, * Detect changes in a person's posture from the acquired activity information, * Based on the acquired heart rate information, the base heart rate component is calculated considering the results of posture change detection. * Calculate the activity heart rate component from the acquired activity information. * The mental stress heart rate component is calculated by subtracting the calculated activity heart rate component and the calculated basal heart rate component from the person's current heart rate, and * Calculates mental stress information about a person's mental stress from the mental stress heart rate component. A processing unit configured in such a way; and - An output unit configured to output calculated mental stress information; It holds.

[0008] In another aspect of the present invention, a stress detection system for detecting mental stress in a person is provided, the system being: - Activity sensors configured to acquire activity information related to human activity; - A heart rate sensor configured to acquire heart rate information that enables the display or calculation of a person's current heart rate; and - A stress detection device disclosed herein that detects a person's mental stress based on acquired activity information and acquired heart rate information; It holds.

[0009] In yet another aspect of the present invention, a non-temporary computer-readable recording medium is provided which stores a corresponding method, a computer program including program code means for causing a computer to perform steps of the method disclosed herein when executed on a computer, and a computer program product causing the method disclosed herein to be executed when executed by a processor.

[0010] Preferred embodiments of the present invention are defined in the dependent claims. The methods, systems, computer programs and media described in the claims are to be understood to have preferred embodiments similar to and / or identical to those described in the claims, particularly those defined in the dependent claims and disclosed herein.

[0011] This invention is based on the idea that mental stress events can be detected by separating instantaneous heart rate signals into portions (or components) related to basal metabolic function and physical effort, and portions unrelated to both. The latter portion has been found to be directly related to mental stress events.

[0012] Furthermore, although a person's posture change hardly requires physical effort, it is known to have a significant impact on the heart rate. For example, a change from a stationary sitting position to a stationary standing position can lead to an increased heart rate of 5 - 10 BPM (beats per minute). At the same time, such a posture change causes only a very slight change in the movement signal from, for example, an acceleration sensor, which is used to determine physical effort. For this reason, according to the present invention, posture change detection is provided to improve the detection accuracy of mental stress. That is, the detection of posture change is used as an additional input for the detection of mental stress and the calculation of mental stress information (such as the presence of a mental stress event and / or the level of a person's mental stress).

[0013] In particular, according to the present invention, it is detected whether a person has changed their posture, and this information is then taken into account in the calculation of the basal heart rate component, taking into account the influence of the posture change on the heart rate. Based on the knowledge of the various parts (i.e., components) contributing to the instantaneous (i.e., current) heart rate, the mental stress heart rate component is calculated by subtracting from the person's current heart rate a component representing activities that could have been previously separately calculated or adapted when a posture change was detected and a component representing the basal heart rate. Then, mental stress information can be calculated from this mental stress heart rate component. In this way, the increase or decrease in heart rate due to posture change can be taken into account when detecting mental stress from the measured heart rate.

[0014] In one embodiment, an adaptive filter can be used to cancel out the "activity contribution" from the heart rate signal. The adaptive filter also corresponds to changes in the relationship between activity count, physical effort, and the increase in heart rate due to physical effort. The adaptive filter uses a DC-free input signal for correct operation. Therefore, the basal heart rate component can be subtracted from the heart rate signal before being processed by the adaptive filter. After subtracting the basal heart rate component and canceling out the activity component, the remaining heart rate increases in BPM units due to mental stress.

[0015] There are various options for taking into account the results of posture change detection in the calculation of the basal heart rate, i.e., for taking into account that a posture change has occurred and / or the type and / or amount of the posture change. One of these options is to calculate the basal heart rate component from the acquired heart rate components and adapt the calculated basal heart rate component if a posture change is detected. Another (additional or alternative) option is to adapt the calculation of the basal heart rate component if a posture change is detected, particularly by employing a different calculation method and / or adapting one or more parameters used in the calculation. Generally, if no posture change or large posture change is detected, the calculation of the calculated basal heart rate and the basal heart rate component is not adapted (i.e., it does not change compared to the basal heart rate and its calculation used when no posture change is considered at all).

[0016] In one embodiment, the processing unit is configured to calculate the basal heart rate component as the lower envelope of the heart rate signal of the acquired heart rate information. Thus, the lower envelope of the heart rate signal can be tracked to estimate the basal heart rate. In one embodiment, an average heart rate signal can be calculated from the acquired heart rate information, and the lower envelope of the heart rate signal or the average heart rate signal can be calculated as the basal heart rate component.

[0017] According to another embodiment, the processing unit is configured to adapt the calculation of the basal heart rate component if a posture change is detected by increasing the rate at which the basal heart rate follows the heart rate signal during a time window after the detection of the posture change. In this way, the change in heart rate caused by the change in posture can be tracked much faster compared to normal tracking (when no posture change occurs), so that the contribution of the heart rate to mental stress can be estimated more accurately, and thus a more accurate calculation of a person's mental stress becomes possible.

[0018] Furthermore, the processing unit can be configured to calculate the base heart rate component by taking the difference between the current value of the heart rate signal and the current value of the base heart rate, multiplying this difference by a multiplication constant, and obtaining a new value of the base heart rate by integrating the result. In this way, the base heart rate component can be calculated simply and quickly.

[0019] In a more advanced embodiment, the processing unit is configured to use: a first multiplication constant if the current value of the basal heart rate is greater than the current value of the heart rate signal; a second multiplication constant if the current value of the basal heart rate is less than the current value of the heart rate signal; and a third multiplication constant if a change in posture is detected, where the second multiplication constant is less than the first multiplication constant, and the first multiplication constant is less than the third multiplication constant. In this way, different multiplication constants can be applied to increases or decreases in the basal heart rate. That is, the multiplication constant is selected based on whether the heart rate is greater than or less than the basal heart rate. This may enable faster tracking when the basal heart rate decreases compared to when the basal heart rate increases. Furthermore, in the case of a change in posture, another (third) multiplication constant that is greater than the other multiplication constants is used to enable even faster tracking of the heart rate.

[0020] In another embodiment, the processing unit is configured to calculate a baseline heart rate component from the acquired heart rate component and to adapt the calculated baseline heart rate component by increasing it when the posture changes from sitting or lying down to standing, and by decreasing it when the posture changes from standing to sitting or lying down. This embodiment is based on the idea that heart rate generally increases when standing and generally decreases when sitting or lying down. In this way, the accuracy of detecting mental stress can ultimately be improved.

[0021] Various methods exist for detecting changes in a person's posture. In one embodiment, the processing unit is configured to detect changes in a person's posture from acquired activity information by detecting changes in the measured acceleration that exceed an acceleration threshold. For example, when a person stands up, the acceleration is 1g (9.81 m / s²). 2 It will increase beyond ). In other words, the increase in acceleration will be interpreted as information indicating a change in posture.

[0022] In other embodiments, changes in the orientation of a sensor configured to acquire activity information can be detected and used as an indicator of posture changes. For example, if a smartwatch equipped with a motion sensor is used to acquire activity information, the orientation of the smartwatch will change when a person stands up from a seated or reclining position. In most cases, a standing person will have their hands pointing towards the ground, while a seated person will almost always have their arms horizontal.

[0023] In other embodiments, changes in an activity type scale, which indicates the type of activity a person is engaged in, determined from acquired activity information, can be detected and used as an indicator of postural changes. This scale can take values ​​such as "rest," "running," "cycling," "walking," and "other," and changes in this scale can be used as an indicator of postural changes.

[0024] The activity information described above may include one or more of the following: activity count information, activity type information, and accelerometer information. For example, the activity count can be used as a measure of movement and / or effort. The activity count represents the average change in the accelerometer signal over a specific period.

[0025] The processing unit may be configured to calculate the activity heart rate component from the acquired activity information by minimizing the correlation between the activity heart rate component and the mental stress heart rate component. For example, the signal AHR-BHR is used as the input signal to an adaptive filter. According to the model, this signal holds the sum of the activity heart rate component and the mental stress heart rate component. The adaptive filter constructs the physical heart rate component by filtering the activity signal included in the activity information, such as the activity count signal. The mental stress heart rate component can be calculated by subtracting the activity heart rate component from the adaptive filter input. The filter coefficients are adapted to minimize the correlation between the mental stress heart rate component and the activity signal.

[0026] There are various embodiments for implementing a heart rate sensor, which may include one or more of the following: photoplethysmography sensors, pulse oximetry sensors, body-worn cameras, remote cameras (e.g., at a distance of several meters), ECG sensors, wristband pressure sensors, and wristband tension sensors. A smartwatch that can be implemented using the present invention may, for example, use a PPG sensor to measure heart rate, but may also use an ECG sensor, camera image, or other method, depending on the situation, either instead or in addition.

[0027] Various embodiments for implementing activity sensors also exist. These may include one or more of the following: accelerometers, body-worn cameras, remote cameras (e.g., at a distance of several meters), skin electroactivity sensors, gyrometers, and temperature sensors. Accelerometer sensors are commonly used to measure motion. Alternatively, motion may be derived as a motion vector from a series of camera images. However, it should be noted that motion is not the same as "physical effort." An accelerometer in a wrist-worn smartwatch device will show very different signals for running and cycling activity types, although the "physical effort" (e.g., impact on heart rate) may be similar. Nevertheless, for a single activity, it is expected that "more movement" will indicate "more effort."

[0028] These and other aspects of the present invention will become apparent from the embodiments described below and will be clarified by reference to such embodiments. [Brief explanation of the drawing]

[0029] [Figure 1] Figure 1 shows a schematic diagram of a first embodiment of the stress detection system and apparatus according to the present invention. [Figure 2] Figure 2 shows a flowchart of one embodiment of the stress detection method according to the present invention. [Figure 3] Figure 3 shows a schematic diagram of a second embodiment of the stress detection system and apparatus according to the present invention. [Figure 4] Figure 4 shows a schematic diagram of a third embodiment of the stress detection device according to the present invention. [Figure 5] Figure 5 shows graphs of various signals indicating the detection of mental heart rate accompanied by the detection of changes in posture. [Figure 6] Figure 6 shows graphs of various signals indicating the detection of mental heart rate accompanied by the detection of changes in posture. [Figure 7] Figure 7 shows graphs of various signals indicating the detection of mental heart rate accompanied by the detection of changes in posture. [Figure 8] Figure 8 shows graphs of various signals indicating the detection of mental heart rate accompanied by the detection of changes in posture. [Figure 9] Figure 9 shows a flowchart of another embodiment of the stress detection method according to the present invention. [Modes for carrying out the invention]

[0030] Heart rate is responsible for the transport of blood and energy throughout the body. Several biological systems influence heart rate. Firstly, there is the basal heart rate, which occurs when a person is at rest. The basal heart rate ensures sufficient blood flow to support basic biological systems such as respiration, digestion, the heartbeat itself, and body temperature regulation.

[0031] When a person starts physical (athletic) activity, the heart rate increases to a certain level higher than the basal heart rate, and in that case, the total increase has a certain relationship with the amount of work performed. This also applies to walking, running, carrying, etc.

[0032] Mental stress can also cause an increase in heart rate. This can be seen as part of the body's preparation for the fight-or-flight response. According to this, a "mental stress event" is defined as an event that causes deep feelings of misfortune or dissatisfaction experienced by a human. Well-known tests from psychological research are the "tone avoidance test", the "stress test of singing a song", the "Raven's Progressive Matrices", and the "pace auditory serial addition test", etc.

[0033] These heart rate dependencies are: HR(t)=HR Basal (t)+HR PhysicalActivity (t)+HR MentalStress (t) and can be modeled. Thus, the measured heart rate HR(t) is a combination of three components (or parts): namely, the basal heart rate component HR Basal (t), the activity heart rate component HR PhysicalActivity (t), and the mental stress heart rate component HR MentalStress (t). This model is: HR MentalStress (t)=HR(t)-HR Basal (t)-HR PhysicalActivity (t) and can be rewritten as. This rewritten model will be applied to determine the mental stress heart rate component HR MentalStress (t).

[0034] Figure 1 shows a schematic diagram of a first embodiment of the stress detection system 1 and stress detection device 10 according to the present invention. System 1 comprises an activity sensor 20 for acquiring activity information relating to a person's activities, a heart rate sensor 30 for acquiring heart rate information that can indicate or calculate a person's current heart rate, and a stress detection device 10 for detecting a person's mental stress based on the acquired activity information and heart rate information.

[0035] In one embodiment, the activity sensor 20 may be attached to a person's body, for example, the wrist, arm, leg, or torso, and may include one or more of an accelerometer, camera, skin electroactivity sensor, gyrometer, and temperature sensor. In other embodiments, a device located away from the person, such as a remote camera observing the person from a distance of, for example, one meter or several meters, may also be used as the activity sensor 20 to detect the person's activity. The activity sensor 20 is generally configured to detect movement and enable the recognition of the person's activity, and optionally the intensity and / or type of activity.

[0036] In one embodiment, the heart rate sensor 30 may be attached to a person's body, for example, the wrist, arm, leg, or torso, and may include one or more of the following: a photoplethysmograph sensor, a pulse oximetry sensor, a camera, an ECG sensor, a wristband pressure sensor, and a wristband tension sensor. In other embodiments, the heart rate sensor 30 may also be a device located away from the person, such as a remote camera that observes the person from a distance of, for example, one meter or several meters, and evaluates radiation transmitted through or reflected from the skin to detect minute changes in light absorption of the skin caused by periodic color changes of the person's skin induced by the pulsating blood volume, i.e., by blood volume pulses. This technique is widely known and is described, for example, in the literature “Remote plethysmographic imaging using ambient light”, Optics Express, 16(26), 22 December 2008, pp. 21434-21445 by Verkruijsse et al.

[0037] The stress detection device 10 can be implemented in hardware and / or software. For example, the device 10 may be implemented as a appropriately programmed computer or processor. Depending on the application, the device 10 may be, for example, a computer or workstation, or a mobile user device such as a smartphone, laptop, tablet, or smartwatch. For example, in an actual implementation, all elements of the system, including sensors 20, 30 and device 10, may be part of a smartwatch or other wearable. In another practical implementation, only sensors 20, 30 may be part of such a smartwatch or other wearable, while device 10 is implemented on another device such as a computer, laptop, or smartphone, and a wireless connection (e.g., via Bluetooth® or WiFi) is used to transmit data from sensors 20, 30 to device 10.

[0038] The stress detection device 10 generally comprises an activity input unit 11, a heart rate input unit 12, a processing unit 13, and an output unit 14. Figure 2 shows a flowchart of a stress detection method 100 that can be performed by the stress detection device.

[0039] The activity input unit 11 acquires (i.e., receives or takes in; step 101) activity information relating to human activity, preferably via a direct connection to the activity sensor 20 (or a device including the sensor), or via other entities such as memory or a preprocessing unit. The activity information may include, for example, one or more of activity count information, activity type information, and accelerometer information. The activity input unit 11 can be configured as a normal wired or wireless data interface such as Bluetooth®, WiFi, or a LAN interface.

[0040] The heart rate input unit 12 acquires (i.e., receives or takes in; step 102) heart rate information that enables it to indicate or calculate a person's current heart rate. The heart rate information may preferably be acquired via a direct connection to a heart rate sensor 30 (or a device including the sensor) or via other entities such as memory or a preprocessing unit. The heart rate information may include, for example, a pulse signal or photoplethysmography (PPG) signal that enables the calculation of the heart rate, or it may be a signal that directly indicates the heart rate. For example, a heart rate signal (a heart rate value over time) may be calculated from the acquired heart rate information or may be included in the acquired heart rate information. The heart rate input unit 12 can be configured as a normal wired or wireless data interface such as Bluetooth®, WiFi, or a LAN interface.

[0041] Processor 13 processes the acquired activity information and acquired heart rate information and calculates mental stress information regarding the person's mental stress (steps 103-107). In particular, in the first step 103, the processor detects changes in the person's posture (for example, from sitting or lying down to standing, or from standing to sitting or lying down) from the acquired activity information. The result of this posture change detection is considered in the subsequent second step 104, in which the base heart rate component is calculated from the acquired heart rate information. For example, the base heart rate component can be calculated from the acquired heart rate component, taking into account the result of the posture change detection, and the calculated heart rate component is applied when a posture change is detected. If no posture change is detected, the calculated heart rate component is not applied. In other embodiments, the calculation of the basal heart rate component may be adapted by employing a different calculation method when a change in posture is detected, particularly compared to when no change in posture is detected, and / or by adapting one or more parameters used in the calculation when a change in posture is detected.

[0042] Before, after, or concurrently with that, the processor calculates the activity heart rate component from the acquired activity information in step 105.

[0043] Next, in step 106, the processor 13 calculates the mental stress heart rate component by subtracting the calculated activity heart rate component and the calculated base heart rate from the person's current heart rate, which is included in or derived from the acquired heart rate information. Finally, in step 107, the processor calculates mental stress information related to the person's mental stress from the mental stress heart rate component.

[0044] The output unit 14 outputs the calculated mental stress information (step 108). For example, the processor can output this information visually on a display or audibly through speakers, or it can supply this information to other entities for further processing or rendering. The mental stress information may include information indicating whether a person is experiencing mental stress and / or the level / intensity of that mental stress.

[0045] Figure 3 shows a schematic diagram of a second embodiment of the stress detection system 2 and stress detection device 40 according to the present invention. In addition to the elements of system 2, a human body model 200 is also shown in this figure for illustrative purposes.

[0046] Human body model 200 shows the three components of the measured heart rate HR(t) as described above, and a term h(n) that models the body's response to physical activity in relation to heart rate. For example, when a person starts some physical activity, it affects their heart rate. This effect is almost always not instantaneous, but involves some delay at the start and some delay when the activity ends. The temporal behavior of these delay effects can be modeled by term h(n).

[0047] The activity sensor 20 and heart rate sensor 30 shown in Figure 1 are included in a sensor and signal processing module 50 that can be implemented as a wrist-based activity sensor module. This module 50 may include, for example, a PPG sensor (as a heart rate sensor) and an accelerometer sensor (as an activity sensor). Furthermore, in this embodiment, the sensor signals acquired by the sensors are preprocessed to provide higher-level signals that include at least a heart rate AHR (which may be an estimation measure of a person's heart rate, such as instantaneous heart rate or average heart rate), an activity count metric ACN, and optionally an activity type ACT. That is, this preprocessing extracts higher-level signals from the raw sensor data. Furthermore, the raw accelerometer signal ACC may also be provided by the module 50.

[0048] During a user's performance of a specific activity type, heart rate changes with activity. The change relative to the baseline generally depends (linearly) on the activity count. The actual relationship will differ depending on the type of activity (sitting, walking, running, etc.). As a person changes their activity pattern, there will be continuous fluctuations in the baseline heart rate, and it can be expected that the relationship between activity count and heart rate will increase with physical activity. However, in the short term, a "linear" relationship will still exist between activity and the increase in heart rate relative to the baseline. The adaptive filter updates its FIR coefficients in response to changes in the relationship between physical activity and physical heart rate due to changes in activity type, so that the suppression of physical activity in the adaptive filter output is optimal for the type of activity being performed at that time.

[0049] The mental stress detection algorithm applied by the present invention generates separate estimates for the three different heart rate components of the heart rate model described above.

[0050] Module 50 provides an AHR scale as an estimate of the current heart rate. In this embodiment, the basal heart rate tracking module 41 provides the HR BasalIn other words, the baseline heart rate component (BHR) is estimated as the lower envelope of the mean heart rate signal (AHR). For example, as will be explained in more detail below, BHR tracks AHR with different time constants depending on whether AHR or BHR has the maximum value.

[0051] Module 50 also provides an ACN scale with values ​​that increase with the amount of “physical activity.” This assumption is valid as long as the user continues similar activities, such as sitting, walking, running, thinking, etc. In such situations, HR PhysicalActivity A linear dependency is assumed between (t) and ACN. An adaptive least-squares adaptive filter 42 is used to determine the ACN's contribution to the mental heart rate contribution (MHR) and to ensure that the physical contribution cancels out in the mental heart rate contribution (MHR). In other words, the adaptive filter 42 removes the ACN-correlated content from the mental heart rate contribution by correlating a set of ACN samples with the MHR, so that the filtered output PHR (physical heart rate) simulates the activity-induced activity heart rate contribution.

[0052] The mental stress heart rate contribution (MHR) is obtained by subtracting both the heart rate heart rate (BHR) and heart rate heart rate (PHR) from the AHR using common or separate subtraction modules 43 and 44. The scaler module 45 (for example, applying a logarithmic scale) converts the mental stress heart rate contribution (MHR) into a presentable value representing mental stress information, such as the mental stress level (MSL).

[0053] It is a known effect that heart rate depends on a person's posture. Even when the user is completely still, a variation of 10 BPM can be expected between lying, sitting, and standing positions. This affects the determination of mental stress information, because standing up causes a relatively large change in heart rate, although it hardly appears in the ACN activity signal. This problem is solved according to the present invention by adding a posture change detection module 46 that detects whether or not there is a change in posture. Its output PCD (Posture Change Detection) is supplied to the basal heart rate tracking module 41. If no change in posture is detected, the basal heart rate and / or its calculation do not change. If a change in posture is detected, the basal heart rate and / or its calculation are modified by the basal heart rate tracking module 41.

[0054] Figure 4 shows a schematic diagram of a third embodiment of the stress detection device 60 according to the present invention. In this embodiment, AHR, ACN, and ACC are used as inputs. The sample rates of these signals may be higher than the minimum required in terms of signal bandwidth. The input signals AHR and ACN may have, for example, a sample rate of 1 Hz (N1=1). The ACC signal may be sampled at 128 Hz (N2=4) or 32 Hz (N2=1) to both be 32 Hz. Low-pass filters (LPFs) 61, 62, and 63 are used to reduce noise and to prepare the decimation stage. Thus, these LPFs prevent aliasing when the signal is decimated to a lower sample rate. The LPFs 61, 62, and 63 can be designed as a Hanning window with n LPF taps and a DC gain of 1. The output signals of LPF61, 62, and 63 are shown as ACNlpd, AHRlpd, and ACClpd, respectively, meaning they have been low-pass filtered and decimated.

[0055] In this embodiment, a height change detection (HCD) module 64 is provided for detecting changes in posture. The absolute value module (ABS) 641 calculates the absolute value of the accelerometer value, for example, abs(ax,ay,az)=sqrt(ax 2 +ay2 +az 2 The posture change detection unit (ABS(xg)) 642 calculates the following. The posture change detection unit (ABS(xg)) 642 compares the absolute value of the accelerometer measurement with the gravity of 1g. If the difference is large, the person may be standing or sitting, which is interpreted as a change in height, or more generally, a change in posture. After posture change detection, the window unit 643 starts the window timer. The window signal is decimated to 1Hz by the decimation coefficient N3=32 in the decimation unit 644.

[0056] If no postural change is detected, the basal heart rate estimation module 65 tracks increases in heart rate slowly (with a large time constant) and decreases much faster (with a smaller time constant). In this way, it can track the "lower envelope" of the heart rate signal, called the basal heart rate. If the window signal is activated (i.e., if a postural change is detected), the basal heart rate estimation module 65 is controlled to track the heart rate signal very quickly (e.g., with a small time constant). In this way, the basal heart rate adapts very quickly to a value that matches the new posture.

[0057] Specifically, the subtraction unit 651 takes the difference between the actual heart rate AHR1pd and the current baseline heart rate. The coding unit 652 determines the sign of the result of this subtraction. Depending on the sign and the HCD attitude detection signal supplied by the HCD module 64, the gain control unit 653 proposes different multiplication constants for the lower envelope tracking system. The multiplier 654 multiplies the difference result from the subtraction unit 651 by the multiplication constant proposed by the gain control unit 653. The result is integrated (for example, added to the current baseline heart rate value by the addition unit 655) and delayed by the delay unit 656, which imposes a single-cycle delay.

[0058] In other words, the BHR estimation module 65 forms a first-order IIR low-pass filter with a variable control time constant. Basal heart rate tracking takes the difference between the current basal heart rate value BHR and the actual (filtered and downsampled) heart rate value AHR. In the next clock cycle, a new BHR is calculated taking into account the value of the above difference, the sign of the difference, and the window signal (HCD).

[0059] The output BHR from the BHR estimation module 65 is subtracted from the actual heart rate AHR1pd by the subtraction unit 66. The resulting difference is then subjected to adaptive filtering. Finally, the adaptive filter 67 removes content correlated with ACN from the mental heart rate MHR signal. For this purpose, the adaptive filter unit 672 (e.g., an adaptive LMS (least squares mean) filter) correlates a set of sample ACNs with the MHR. The FIR (finite impulse response) coefficient of the adaptive filter unit 672 is updated or adjusted in the update unit 671 so that the correlation value decreases in the next step. The FIR coefficient is used to filter the ACN signal so that the filtered output PHR simulates the activity-based heart rate contribution.

[0060] A user's heart rate does not respond immediately to changes in activity. A delay of several seconds is expected, so the physical (exercise) contribution to heart rate is calculated using a linear filter (FIR): HR PhysicalActivity (t)~=FIR(ACN) It can be estimated as follows: Regarding the scale, the calculation is:

number

[0061] signal HR MentalStress (t) and signal HR PhysicalActivity (t) is assumed to be uncorrelated. The adaptive LMS filter 671 helps to continuously estimate the FIR coefficient 672. The filter determines the contribution (correlation, inner product) of ACN in MHR and adapts the filter coefficient so that the correlation result is least squares minimized. As a result, the physical contribution is canceled out in the mental heart rate signal. The new FIR coefficient is adjusted for each sample period. Thus, FIR(n,1:nFIR) is assumed, where n is the actual sample index and 1:nFIR is the filter coefficient.

[0062] Next, during initialization, FIR(1,1:nFIR) is set to [0 0 0 … 0] (nFIR zero). In the following sample, calculations are performed for m=1:nFIR. FIR(n+1,m)=FIR(n,m)+kAdaptive* MHR(n)*ACN(nm) Here, kAdaptive is a constant used to adjust the filter's convergence, tracking speed, and stability.

[0063] In the subtraction unit 673, the PHR is obtained by subtracting the BHR from the AHR, and this is converted into mental stress information (MSL) by the scaler 68.

[0064] Figure 5 shows |ACC|70 and the posture correction signal HCD71 which is active during the periods 23:07-23:08 and 23:35-23:36. Furthermore, the heart rate signal 72 and the basal heart rate signal 73 are also shown. The fast tracking mode with kfast is active during the period when the HCD is active. The slow tracking mode with krise is active from 00:13 to 00:23. The normal tracking mode with kfall is active from 00:23 to 00:26. Figure 30135_pcor_HRcomponents.png shows the activity signal ACN and the reconstructed physical heart rate PHR.

[0065] Figure 6 provides an impression of the relative signal amplitudes of the HR signal 74, BHR signal 75, PHR signal 76, and MHR signal 77, as well as showing the ACT activity type signal 78.

[0066] Figure 7 shows |ACC|80 and the posture correction signal HCD81, which is active and tracks rapidly between 11:14 and 11:15. Slow tracking is active from 11:25 to 11:37, and normal tracking is active from 11:37 to 11:38 and from 11:43 to 11:50. Furthermore, the heart rate signal 82 and the basal heart rate signal 83 are also shown.

[0067] Figure 8 provides an impression of the relative signal amplitudes of the HR signal 84, BHR signal 85, PHR signal 86, and MHR signal 87, as well as showing the ACT activity type signal 88.

[0068] Figure 9 shows a flowchart of another embodiment of the stress detection method 120 according to the present invention. This method partially includes the same steps as the embodiment of method 100 shown in Figure 2, these steps are indicated by the same reference numerals and will not be described. Unlike method 100, method 120 includes a step 109 in which parameters and / or a calculation method for calculating the basal heart rate in step 104 if a change in posture is detected. This step 109 is omitted if no change in posture is detected.

[0069] Based on the baseline heart rate calculated in step 104, the acquired heart rate information, and the calculated activity heart rate component, the mental stress heart rate component is calculated in step 106.

[0070] In addition to calculating mental stress information in step 107, the calculated mental stress heart rate component can be further used in step 110 to adapt the filter parameters of an adaptive filter (42 in Figure 3, 67 in Figure 4) which can be used to cancel out the activity contribution from the heart rate signal as described above. This adaptive filter can be implemented in steps 105 and 110, in which case step 105 includes filtering by an FIR filter to generate an activity heart rate contribution from the activity information, and step 110 updates (i.e., adapts) the FIR coefficient using the current value of the adaptive filter coefficient, the activity information, and the mental heart rate component.

[0071] As mentioned earlier, heart rate is posture-dependent. To detect changes in posture, ultrafast tracking of the BHR signal after height change detection may be performed. Here, the term "ultrafast tracking" is used to distinguish it from the fast tracking mentioned above.

[0072] A simple way to detect changes in height is to monitor the norm of the accelerometer signal. When the user is stationary, the accelerometer measurement is close to gravity (g). As soon as the user changes posture, the accelerometer measurement is expected to deviate more significantly from gravity. This detection works not only when the user stands but also when they are seated. Statistical analysis of the results from this simple posture influence reduction system showed a clear improvement compared to the original algorithm.

[0073] Any height change detection event will restart a specific duration window. During this window, the base heart rate estimation function will operate in high-speed tracking mode. This detection can be described with the following pseudocode: # During initialization it sets: BHR(1) = 60 # Then for all samples in the future If BHR(n) > AHR(n) % AHR is lower BHR(n+1) = BHR(n) + kfall * ( AHR(n) - BHR(n) ) % -> Fast tracking Else if HCD_window % Posture change detection window BHR(n+1) = BHR(n) + kfast * (AHR(n) - BHR(n)) % -> SuperFast tracking Otherwise, % AHR is higher BHR(n+1) = BHR(n) + krise * (AHR(n) - BHR(n)) % -> Slow tracking End

[0074] In other embodiments, more constants can be used to address different states, postural changes, and activity changes. For example, ((ACT==sitting)&(HR>BHR)), ((ACT==sitting)&postural change&(BHR>HR)), ((previous ACT==sitting)->(ACT==walking)). The output scale ARL MSL is a scaled version of the MHR scale. Scaling is performed as follows: ARL=min(R-1,floor(R / log2(51)*(log2(floor(2) M *(1+kscale*MHR)))-M))) The calculation can be performed according to the following formula, where the number of scale levels R = 1000, the scaling constant kscale = 1 (which can be changed to adjust the results for stress levels), and the precision M = 10. The results are shown in the table below.

[0075] [Table 1]

[0076] Changes in heart rate due to changes in posture can be considered as different basal metabolic heart rates depending on posture. According to the present invention, changes in posture are detected from an accelerometer signal, and the detection of the change in posture is performed by a tracking mechanism that applies a short time constant during a time window that starts after the detection of the change in posture.

[0077] There are various simple and complex methods available for detecting changes in posture from accelerometer data.

[0078] According to a specific embodiment, when stationary, the 3-axis accelerometer measures 1g⁻¹ = 9.81 m / s². 2 This measures the following: When a person stands up from a seated position, this value temporarily increases to more than 1g. In this case, the basal heart rate ultrafast tracking window is Trigg1=(|acc|-9.81)>Threshold1 It is activated in the following case.

[0079] According to other embodiments, the change in posture from standing to sitting is, Trigg2=||acc|-9.81|>Threshold2 This can be detected by [method / function]. This embodiment does not distinguish between standing and sitting. Therefore, it may enable a faster response in detecting mental stress after sitting.

[0080] When a wearable device is worn on the wrist, most people will point their hands downwards when standing and horizontally when sitting. For example, the orientation of an accelerometer depends on posture. This alternative configuration uses orientation information to detect changes in posture to improve the accuracy of mental stress detection.

[0081] The "Activity Type" scale can take values ​​such as "Rest," "Running," "Cycling," "Walking," and "Other." Changes in this scale value can also indicate changes in posture.

[0082] In one embodiment, the new basal heart rate value is estimated by taking the difference between the actual heart rate and the current basal heart rate, multiplying this by a constant, and adding this to the current basal heart rate value, which can be regarded as a first-order low-pass IIR filter. Such a filter can be characterized by a low-pass bandwidth and / or a time constant. The time constant of the filter depends on the sample frequency of the filter and the multiplication factor. Different tracking time constants are realized by different multipliers selected under the conditions of PostureChangedWindow, HR > BHR, and HR < BHR. The latter two conditions produce a "lower envelope tracking" behavior.

[0083] For an analog low-pass filter, the time constants for these different conditions can be 1 / (Fs * kfast), 1 / (Fs * krise), and 1 / (Fs * kfall). Generally, kfast >> kfall >> krise holds. The values of these constants can be selected by investigating the recorded signals of a number of test candidates. In other embodiments, per-person optimization can be done. In an exemplary implementation, Fs = 1 Hz, and kfast = 0.1, kfall = 0.01, and krise = 0.001. In other embodiments, other values can be used.

[0084] In a practical implementation of the disclosed stress detection system, the sensor module is a wrist-worn device incorporating a PPG and a motion sensor as well as a host processor device. Continuous data streams from both sensors are supplied to software operating on the host to extract a plurality of high-level metrics characterizing the physical state of the device's wearer. These metrics include AHR, an estimated value of the heart rate, and ACN, an activity count that holds a measurement of the amount of movement (~ = physical activity) applied by the user.

[0085] A new mental stress metric is added. The calculation of the mental stress metric requires AHR, ACN, and ACC as inputs. The output from the algorithm is recorded as the ARL metric. For example, there is a visual display on a 3-color LED.

[0086] Although the present invention has been illustrated and described in detail in the drawings and the above description, such illustrations and descriptions should be considered explanatory or illustrative and not limiting. That is, the present invention is not limited to the disclosed embodiments. Other variations of the disclosed embodiments can be understood and implemented by those skilled in the art in carrying out the claimed invention from a close examination of the drawings, disclosure and appended claims.

[0087] In the claims, the term “having (including)” does not exclude other elements or steps, and the singular form does not exclude the plural. A single element or other unit may perform the functions of several items described in the claims. The mere fact that certain means are described in different dependent claims does not imply that combinations of these means cannot be used advantageously.

[0088] Computer programs can be stored and distributed not only on appropriate non-temporary media such as optical storage media or solid-state media supplied together with or as part of other hardware, but also in other forms, such as via the Internet or other wired or wireless communication systems.

[0089] No reference numeral in a claim should be construed as limiting the scope.

Claims

1. A stress detection device for detecting mental stress in a person, An activity input unit that acquires activity information related to the activities of the aforementioned person, A heart rate input unit that acquires heart rate information that enables the display or calculation of the person's current heart rate, A processing unit, From the activity information acquired, a change in the person's posture is detected. Based on the acquired heart rate information, the base heart rate component is calculated considering the results of the detection of the change in posture. From the activity information obtained above, the activity heart rate component is calculated. The mental stress heart rate component is calculated by subtracting the calculated activity heart rate component and the calculated basal heart rate component from the person's current heart rate, and From the aforementioned mental stress heart rate component, mental stress information regarding the person's mental stress is calculated. Processing unit and An output unit that outputs the calculated mental stress information and A stress detection device having the following features.

2. The processing unit considers the results of the detection of the change in posture from the acquired heart rate information, The base heart rate component is calculated from the acquired heart rate component, and the calculated base heart rate component is applied when a change in posture is detected, and / or When a change in posture is detected, the calculation of the base heart rate component is adapted by employing a particularly different calculation method and / or by adapting one or more parameters used in the calculation. The stress detection device according to claim 1, wherein the basic heart rate component is calculated by doing so.

3. The stress detection device according to claim 1, wherein the processing unit calculates the base heart rate component as the lower envelope of the heart rate signal of the acquired heart rate information.

4. The stress detection device according to claim 3, wherein the processing unit, when a change in posture is detected, adapts the calculation of the base heart rate component by increasing the speed at which the base heart rate follows the heart rate signal during a time window after the detection of the change in posture.

5. The stress detection device according to claim 3, wherein the processing unit calculates the base heart rate component by taking the difference between the current value of the heart rate signal and the current value of the base heart rate, multiplying the difference by a multiplication constant, and obtaining a new value of the base heart rate by integrating the result.

6. The aforementioned processing unit is If the current value of the base heart rate is greater than the current value of the heart rate signal, the first multiplication constant is set. If the current value of the base heart rate is smaller than the current value of the heart rate signal, a second multiplication constant is applied, and If a change in posture is detected, a third multiplication constant is applied. Use, The second multiplication constant is smaller than the first multiplication constant, and the first multiplication constant is smaller than the third multiplication constant. The stress detection device according to claim 5.

7. The stress detection device according to claim 1, wherein the processing unit calculates the base heart rate component from the acquired heart rate component, and when a change in posture is detected, it adjusts the calculated base heart rate component by increasing it when the posture changes from sitting or lying down to standing, and by decreasing it when the posture changes from standing to sitting or lying down.

8. The stress detection device according to any one of claims 1 to 7, wherein the processing unit detects a change in the person's posture from the acquired activity information by detecting a change in the measured acceleration that exceeds an acceleration threshold and / or by detecting a change in the orientation of the sensor that acquires the activity information.

9. A stress detection device according to any one of claims 1 to 8, wherein the processing unit detects a change in the person's posture from the acquired activity information by detecting a change in an activity type scale indicating the type of activity of the person determined from the acquired activity information.

10. The stress detection device according to any one of claims 1 to 9, wherein the activity information includes one or more of activity count information, behavior type information, and accelerometer information.

11. A stress detection device according to any one of claims 1 to 10, wherein the processing unit calculates the activity heart rate component from the acquired activity information by minimizing the correlation between the activity heart rate component and the mental stress heart rate component.

12. A stress detection system for detecting mental stress in a person, An activity sensor that acquires activity information related to the activities of the aforementioned person, A heart rate sensor that acquires heart rate information that enables the display or calculation of the person's current heart rate, A stress detection device according to any one of claims 1 to 11, which detects a person's mental stress based on the acquired activity information and the acquired heart rate information. A stress detection system having the following features.

13. The activity sensor includes one or more of the following: an accelerometer, a body-worn camera, a remote camera, a skin electrical activity sensor, a gyrometer, and a temperature sensor, and / or The heart rate sensor includes one or more of the following: a photoelectric pulse wave sensor, a pulse oxygen concentration sensor, a body-worn camera, a remote camera, an ECG sensor, a wristband pressure sensor, and a wristband tension sensor. The stress detection system according to claim 12.

14. A method for operating a stress detection device for detecting a person's mental stress, The processor of the stress detection device acquires activity information relating to the person's activities. The processor obtains heart rate information that enables it to indicate or calculate the person's current heart rate, The processor includes the step of detecting a change in the person's posture from the acquired activity information, The processor performs the steps of calculating the base heart rate component from the acquired heart rate information, taking into account the results of detecting the change in posture, The processor performs the steps of calculating the activity heart rate component from the acquired activity information, The processor calculates the mental stress heart rate component by subtracting the calculated activity heart rate component and the calculated base heart rate component from the person's current heart rate. The processor performs the steps of calculating mental stress information relating to the person's mental stress from the mental stress heart rate component, The processor outputs the calculated mental stress information. A method for operating a stress detection device, comprising the following:

15. A computer program comprising, when executed on a computer, program code means for causing the computer to perform the steps of the method for operating the stress detection device described in claim 14.

Images