KR20180041458A - Earable apparatus and method for measuring stress of human - Google Patents

Abstract

An earable device according to the present invention includes: a tool which can be worn on an ear of a human body; a sensor unit including an acceleration sensor unit connected to the tool and generating acceleration data according to the head movement of the human body by sensing the head movement of the human body; and a processor for calculating a degree of stress which the human body receives by performing signal processing for the acceleration data received from the acceleration sensor unit. Accordingly, the present invention can accurately monitor an emotional state and the stress at any time.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a method of measuring a stress of a human body,

The present invention relates to a method for measuring the stress of a human body and a succeeding device for the same.

Human beings instinctively make a regular vibration motion to balance the head three-dimensionally, and this micro-vibration is a biological signal that is reflected in the emotional response to the vestibular reflex, which is part of the central nervous system, It can be measured by digital camera or CCTV.

The security detection solution using the Vibra image has the following features. It is a technology that can monitor people with a high degree of suspicion about crime in a large number of crowd. It can recognize the microvibration of the human body as a psychophysiological parameter with a non-contact type device such as a digital camera or a CCTV, Visualization technology. And, you can use it by connecting a PC with viral image software to a camera or CCTV already installed.

However, there is a problem that the camera or the CCTV is severely noisy with respect to movement of surrounding objects, and can be measured only at a specific place where the camera is installed. In addition, continuous monitoring of a specific user is not easy, and there is a problem that the user may feel a sense of rejection against the camera.

As described above, there are various problems in the motion detection technology of an object using a vibra image, and in particular, there is a problem that accuracy is low because the user can not carry or wear it. However, no method has yet been proposed to solve this problem.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a continuous device.

It is another object of the present invention to provide a method for calculating the degree of stress of a human body.

According to another aspect of the present invention, there is provided a computer-readable recording medium storing a program for causing a computer to execute a method of calculating a stress level of a human body.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are not restrictive of the invention, unless further departing from the spirit and scope of the invention as defined by the appended claims. It will be possible.

According to an aspect of the present invention, there is provided a device for worn on the ear of a human body. And an acceleration sensor unit connected to the mechanism for detecting head movement of the human body and generating acceleration data according to head movement of the human body; And a processor for performing signal processing on the acceleration data received from the acceleration sensor unit to calculate a degree of stress of the human body.

Wherein the processor performs prefiltering to extract a frequency component corresponding to a predetermined frequency to be used for calculating the stress degree from the received acceleration data and uses the prefiltering data to calculate the stress degree from the preprocessed acceleration data Calculating at least one feature value, and calculating the degree of the stress by applying the calculated at least one feature value to a predetermined machine learning algorithm. The predetermined frequency to be used for calculating the degree of stress may correspond to a frequency of 1 Hz to 100 Hz.

Wherein the at least one feature value comprises an average signal strength of an acceleration signal corresponding to the preprocessed acceleration data, a standard deviation of the acceleration signal, a number of zero crossings when the acceleration signal is second-differentiated, A quadratic momentum of the signal is normalized to a square of standard deviation, a peak of the acceleration signal, a power spectral density generated by Fourier transform of the acceleration signal, an auto-correlation of the acceleration signal, The number of times the acceleration signal exceeds a certain threshold value in one epoch, the distribution of signal intensity as the third order momentum of the acceleration signal, the entropy of the acceleration signal, the energy in the frequency domain associated with the stress, And may include the time-dependent rate of change of the power spectrum during wavenumber analysis.

The processor may perform the prefiltering using a Short-Time Fourier Transform (STFT), a wavelet, or a Multivariate Empirical Mode Decomposition (MEMD).

The predetermined machine learning algorithm may include a random forest algorithm, a random subspace algorithm, or a bagging algorithm.

The sensor unit of the serial device may further include a body temperature sensor unit for sensing the body temperature of the human body, and the processor may calculate the degree of stress of the human body based on the data received from the body temperature sensor unit .

Alternatively, the sensor unit of the peripheral device may further include a blood flow sensor unit for sensing blood flow of the human body, and the processor calculates a degree of stress of the human body based on the data received from the blood flow sensor unit can do. At this time, the processor may calculate at least one feature value from the data received from the blood flow sensor unit, and calculate the degree of stress the human body receives based on the calculated feature value.

Alternatively, the sensor unit of the peripheral device may further include an electromyogram sensor unit for verifying the electrical activity of the muscles of the human body. The processor may further include a stress sensor for receiving stress from the human body based on the data received from the electromyogram sensor unit, Can be calculated. At this time, the processor may calculate at least one feature value from the data received from the electromyogram sensor unit, and calculate the degree of stress that the human body receives based on the calculated feature value. Wherein at least one feature value calculated from the data received from the electromyogram sensor unit is an index obtained by quantifying how much the deviation is deviated toward the high frequency than the low frequency in the data or the area below the specific frequency is 50% ≪ / RTI > may include a value of the particular frequency at the time.

Meanwhile, the acceleration sensor unit generates acceleration data of three axes.

According to another aspect of the present invention, there is provided a method of calculating a degree of stress of a human body, the acceleration sensor connected to a mechanism wearable on the ear of the human body senses head movement of the human body, Generating acceleration data according to the acceleration data; And a step of calculating a degree of stress of the human body by performing signal processing on the generated acceleration data.

The step of calculating the degree of stressor may include: performing prefiltering to extract a frequency component corresponding to a predetermined frequency to be used for calculating the degree of stress in the generated acceleration data; Calculating at least one feature value to be used for calculating the degree of stress from the preprocessed acceleration data; And calculating the degree of stress by applying the calculated at least one feature value to a predetermined machine learning algorithm.

The method for calculating the degree of stress of the human body may further include generating data on a body temperature change by sensing the body temperature of the human body, wherein the step of calculating the degree of stress includes: And calculating a degree of stress of the human body based on the degree of stress.

Alternatively, the method for calculating the degree of stress of the human body may further include calculating data on a blood flow change by sensing the blood flow of the human body, wherein the step of calculating the degree of stress comprises: Calculating at least one feature value; And calculating a degree of stress of the human body based on the calculated characteristic value.

Alternatively, the method of calculating the degree of stress of the human body may further include generating data on the electrical activity of the muscles of the human body, wherein the step of calculating the degree of stress comprises: Calculating at least one feature value from the feature value; And calculating a degree of stress of the human body based on the calculated characteristic value.

The generated acceleration data may correspond to acceleration data of three axes.

The following device according to the present invention has an advantage of being able to monitor the emotional state and stress due to the surrounding situation from time to time,

The effects obtained by the present invention are not limited to the above-mentioned effects, and other effects not mentioned can be clearly understood by those skilled in the art from the following description will be.

BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of the specification, illustrate embodiments of the invention and, together with the description, serve to explain the principles of the invention.
1 is a view illustrating an acceleration sensor used in a succeeding device according to the present invention.
FIG. 2 is a diagram illustrating an example of a continuous device for measuring stress according to the present invention.
3 is a diagram illustrating a specific method of performing the sensing by the sensor unit 230 and a method of measuring or analyzing stress of the human body using the data acquired by the processor 240 from the sensor unit 240. [
Fig. 4 is a diagram illustrating an exemplary number of wavelets. Fig.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the accompanying drawings. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The following detailed description, together with the accompanying drawings, is intended to illustrate exemplary embodiments of the invention and is not intended to represent the only embodiments in which the invention may be practiced. The following detailed description includes specific details in order to provide a thorough understanding of the present invention. However, those skilled in the art will appreciate that the present invention may be practiced without these specific details.

In some instances, well-known structures and devices may be omitted or may be shown in block diagram form, centering on the core functionality of each structure and device, to avoid obscuring the concepts of the present invention. In the following description, the same components are denoted by the same reference numerals throughout the specification.

Hereinafter, the following device for stress measurement will be described in detail. First, a brief description will be given of VER (Vestibular Emotional Reflex) used for stress measurement in a succeeding device according to the present invention.

The present invention measures indicators of the activity of the motor cells responding to stimulation of the autonomic nerves of the human body, establishes an index of evaluation of the autonomic nerves of the human body, and efficiently monitors the stresses that are nowadays present in modern society It is meaningful in that it can manage and receive effective treatment if necessary. Existing stress measuring products require a headgear with an EEG sensor directly on the head to reduce the portability. In order to solve this problem, a lightweight, For example, an earphone) can be worn to improve portability.

An earable device for stress measurement according to the present invention utilizes reflection of emotion by human electrophysiology. VER (Vestibular Emotional Reflex) is one of the vestibular-reflective abilities that are related to human psychology and emotion, reflecting the emotional reflection of human (or human) The vestibular system, which contributes to human balance and sense of spatial orientation, is a sensory organ that provides movement and equilibrioception. The vertical human head position is adjusted by the vestibular organ through the head-neck. For example, a two-month-old can not vertically mechanically balance a heavy head without this movement of the head on the reflex level.

Because the vestibular organ is connected to the sensory organs, neural organs and all parts of the body, the 3D trajectory of head movement is very complex, and vestibular reflex abilities are used to diagnose mental health in humans. Vestibulo-Ocular Reflex (VOR) is a reflex eye movement that stabilizes the image on the retina while the head is moving by moving the eye in the opposite direction to the head movement. This positions and maintains the image in the visual center.

Human senses can be classified into four stimuli: form, modality, intensity, location, duration of stimulation.

Receptors can be sensitive to certain types of modality stimuli. For example, receptors for other physical stimuli respond best to other forms of contact stimuli such as sharp or blunt objects. The receptor sends an impulse in some form to send information about the intensity of the stimulus. For example, the strength of sound. The location of the receptors to be stimulated provides information to the brain about the location of the stimulus. For example, stimulating the receptors of physical stimuli from the fingers is conveying information about the fingers to the brain. And the duration of the stimulus is transmitted by stimulating the form of the receptor.

Like the normal sensory organs, the vestibular system responds to stimuli. However, because gravity is constantly stimulating, the vertical head coordination of the vertical head works constantly and creates a reflex process. This corresponds to a major physiological difference between the cooperation of vertical head muscle movements and other sensory processes that sometimes work. This difference is due to the fact that the Vertical Head Coordination is used in conjunction with electrocardiogram (ECG), heart rate (HR) measured by blood pressure, brain activity measured by Electroencephalography (EEG) activity is transmitted to the normal physiological process by thermoregulation measured by galvanic-skin responding (GSR).

At the sensor level, depending on the human condition, the signals are sent from the vestibular receptors to the autonomic nervous system and the brain and muscles are delivered at different time intervals. This means that there is a dependency between emotional state and emotional reflection (VER) by vestibular head coordination or electrographic system of the electrophysiological head.

1 is a view illustrating an acceleration sensor used in a succeeding device according to the present invention.

An acceleration sensor is a generic term for sensors that measure changes in speed (acceleration). When an object with a mass receives an acceleration, a force is generated. Therefore, when the object is supported by a spring, the acceleration can be measured according to the degree of expansion. As shown in Fig. 1, when the sensor is tilted, the force applied to the spring in the sensor changes, so that the degree of gravitational acceleration with respect to each axis can be measured. In the following device according to the present invention, the acceleration can be measured according to the head movement of the human body using the acceleration sensor shown in FIG.

FIG. 2 is a diagram illustrating an example of a continuous device for measuring stress according to the present invention.

Referring to FIG. 2, the following apparatus includes a musical instrument 210, a power supply unit 220, a sensor unit 230, a (micro) processor 240 And a Bluetooth device 250. The Bluetooth device 250 may be a Bluetooth device.

The device 210 that can be worn on the ear of the human body is formed so that it can be worn by the human ear. The mechanism 210 of the earpiece can also be called an earphone main body.

The power supply unit 220 can supply power to the succeeding device through the charging circuit. For example, the power supply unit 220 may supply power to the sensor unit 230, the processor 240, and the Bluetooth device 250.

The sensor unit 230 performs a sensing operation for measuring the stress of the human body. The sensor unit 230 may further include an acceleration sensor unit 231 and may further include a temperature sensor unit 233, a blood flow sensor unit 235, and an electromyogram sensor unit 237.

The processor 240 performs a series of operations for measuring the stress of the human body using the data obtained through sensing at the sensor unit 230. This series of operations will be described below.

The Bluetooth device 250 enables the wireless device to communicate with another wireless communication device in a Bluetooth manner in close proximity.

3 is a diagram illustrating a specific method of performing the sensing by the sensor unit 230 and a method of measuring or analyzing stress of the human body using the data acquired by the processor 240 from the sensor unit 240. [

Referring to FIG. 3, the acceleration sensor unit 231 may acquire a three-axis acceleration signal by sensing the head movement of the human body. The acceleration sensor unit 231 measures the movement of the head of the human body and calculates three axes of raw data (hereinafter referred to as acceleration data) (S310). The acceleration sensor unit 231 can transmit the calculated acceleration data to the processor 240.

The processor 240 may perform prefiltering to extract a frequency component corresponding to a predetermined frequency to be used for calculating the degree of stress from the acceleration data received from the acceleration sensor unit 231 in operation S320. At this time, the frequency component corresponding to the predetermined frequency to be used for calculating the degree of stress may be a frequency component in the range of 1 Hz to 100 Hz, for example, and a sampling frequency of at least 22 Hz is used in this case. Most of the biological signals that can be acquired from the human body have nonlinear and nonstationary characteristics. In other words, the intensity and frequency of the bio-signal vary with time. In the case of EEG, the conventional Fourier-based frequency analysis method using fixed basis functions is not suitable. In order to analyze the frequency of nonlinear and abnormal signals, a time frequency analysis method is used. By using this method, the time axis is added to the existing frequency analysis method to observe the change of the frequency component with time, The frequency component corresponding to the frequency range (1-100 Hz) which can be used for the middle stress analysis is extracted.

Hereinafter, various methods of prefiltering the acceleration data will be briefly described.

Specifically, there are a short-time Fourier transform (STFT) method, a wavelet method, and a multivariate Empirical Mode Decomposition (MEMD) method as methods of performing preprocessing of acceleration data.

The STFT method is a Fourier analysis method that can observe the change of the frequency and phase information of the signal over time by applying the Fourier transform to the signal divided into several short sections. When a signal is divided into a short section and a Fourier transform is applied, a single Fourier spectrum is generated in each section. Generally, the Fourier spectra thus generated are represented by a spectrum change over time.

The wavelet represents a function whose start and end intensity of the signal is 0, and the wavelet transform performs the frequency analysis of the signal by using such a wavelet function as a basis. The wavelet function can be recursively defined from a basic function called a mother wavelet. The Mexican-hat function as shown in FIG. 4 is used as the number of the mother wavelet. However, assuming that the definition of the wavelet function described above is satisfied It is acceptable to use any number of masters.

 Wavelet transform has the advantage of using free number of moons as compared to Fourier analysis using sinusoidal wave basis. It is mainly used for analysis of signal frequency in audio and image processing fields. Comparing the time-frequency analysis results of the EEG signals using the STFT method and the wavelet transform method, the signal intensity of the STFT is widely dispersed throughout the Mu and beta regions, while the wavelet is relatively concentrated in each region And it is easier to analyze the frequency.

Next, the MEMD method will be described.

The EMD (Empirical Mode Decomposition) method analyzes data-driven signals using data base derived from data without using fixed base functions such as those used in Fourier and wavelets for the effective analysis of nonlinear and abnormal biosignals. Technology. By performing frequency analysis using data without using a fixed base such as a sine wave, it is possible to prevent the signal intensity of the frequency spectrum from being dispersed widely so that the signal intensity of the intrinsic mode function (IMF) The effect can be obtained.

The MEMD method is a frequency analysis method used to apply EMD to a bio signal using a plurality of channels, and the Intrinsic Mode Function (IMF) generated in each channel is designed to share the same frequency domain. In the results of frequency analysis using MEMD, it is difficult to confirm the exact frequency of each component because the frequency components of the Mu and Beta regions are dispersed in STFT and wavelet-based frequency analysis methods. In MEMD, however, The frequency analysis can be performed more precisely and it can be easily used for instantaneous frequency analysis such as Hilbert transform.

As such, the processor 240 may perform preprocessing on the acceleration data using a Short-Time Fourier Transform (STFT) method, a Wavelet method, and a Multivariate Empirical Mode Decomposition (MEMD) method.

Thereafter, the processor 240 may calculate at least one feature value to be used for calculating the degree of stress from the preprocessed acceleration data (S330). Twelve feature values used for stress analysis are extracted or calculated from the preprocessed acceleration data using STFT, wavelet or MEMD method, and each feature value can be defined as follows.

1. Mean: Indicates the average signal strength of the acceleration signal.

2. Standard deviation: The standard deviation of the acceleration signal intensity.

3. Zero Crossing: indicates the number of zero crossings when the acceleration signal is second-differentiated.

4. Kurtosis: The value obtained by normalizing the fourth order momentum of the signal to the square of the standard deviation, which indicates the peakedness of the probability distribution of the signal intensity.

5. Crest factor: The ratio of the peak value of the signal. It is similar to Kurtosis and is a feature value that describes the signal's impulse response.

6. Power Spectrum Density: Represents the power spectrum density generated by the Fourier transform of the acceleration signal.

7. Correlation: It indicates the auto-correlation of the acceleration signal.

8. Threshold Crossing: It shows the number of filtered raw data over a certain threshold in one Epoch.

9. Skewness: The third order momentum of the acceleration signal, indicating the degree to which the distribution of signal strength is biased.

10. Entropy: Indicates the predictability of the signal.

11. Band energy: Represents energy in the frequency domain associated with stress.

12. Spectral Flux: PSD (t) / PSD (t-1) means the rate of change of the power spectrum over time when analyzing the signal frequency.

The processor 240 may calculate the degree of stress by applying at least one of the twelve calculated feature values to a predetermined machine learning algorithm (S340). Here, as an example, a predetermined machine learning algorithm includes a random forest algorithm, a random subspace algorithm, or a bagging algorithm. The processor 240 can classify feature values extracted from the acceleration data through a machine learning (machine learning) algorithm to detect or measure the degree of stress. Hereinafter, the machine learning algorithm will be briefly described.

The random subspace method randomly divides the subspace of the feature value space and uses a plurality of classifiers trained from each divided subspace. The result generated through the plurality of classifiers is used for voting or using the maximum value Generate final classification results. The random forest method is a decision tree-based algorithm. Forests are generated through a plurality of decision trees trained using randomly reconstructed extracted feature value samples. The final classification result is determined in a similar manner as in the random subspace method. The result of classification of the tree is collected and generated.

Bagging is an abbreviation for bootstrap aggregating, which means that several weak classifiers are generated and the final decision is a sorting method that combines the classification results of many weak classifiers into voting. By combining the classification results in this manner, the bagging algorithm can obtain the stability of classification results, prevent overfitting, and improve accuracy.

The n-fold crossover verification method is used to evaluate the classification performance and to avoid over sum. The performance evaluation criteria are the accuracy, recall, precision, and receiver operating characteristic (ROC) AUC (area under ROC curve) can be used.

The sensor unit 230 may further include a body temperature sensor 233, a blood flow sensor 235 and an electromyogram sensor 237 in addition to the acceleration sensor 231.

When the sensor unit 230 further includes a body temperature sensor 233 for sensing the body temperature of the human body, the processor 240 calculates the degree of stress the human body receives based on the data received from the body temperature sensor 233 can do. Here, since the electrical characteristics of ordinary materials and electronic devices vary according to the temperature, they can be all the temperature sensors or the temperature sensor unit 233. However, the detection temperature range, the detection accuracy, the temperature characteristics, the mass productivity, The temperature sensor is suitable for the purpose of use. (NTC, PTC, CTR) thermally sensitive ferrite and metal type thermometer are widely used as sensors for industrial measurement, and thermistor, temperature measuring resistor, thermistor (NTC) Nuclear quadrupole), ultrasonic sensors, and sensors using optical fibers. For the reference temperature, gas thermometer, glass double tube thermometer, quartz thermometer, platinum temperature measuring resistor and platinum-platinum rhodium thermocouple are used. The body temperature sensor 233 may be any one of the various temperature sensors.

When the sensor unit 230 further includes a blood flow sensor unit 235 for sensing the blood flow rate, the processor 240 determines the feature values based on the data received from the blood flow sensor unit 235 And the calculated degree of stress can be calculated by applying the calculated at least one characteristic value to a predetermined machine learning algorithm. The blood flow detection sensor unit 235 is a sensor capable of measuring a blood flow amount flowing in a blood vessel of a human body. The blood flow detection sensor unit 235 has a PPC sensor unit as an example.

The PPG sensor is called a photoplethysmograph sensor. Photoplethysmograph (PPG) is a pulse wave measurement method that can detect heart rate activity by measuring blood flow through blood vessels using the optical characteristics of living tissue. Pulse is a pulsating wave of blood that the blood swells in the heart. It can be measured by changing the blood flow, which is caused by the relaxation and contraction of the heart, that is, the volume change of the blood vessel. Optical pulse measurement is a method of measuring pulse waves using light. The optical sensor senses changes in optical characteristics such as reflection, absorption and transmission ratios of a living tissue when the volume is changed. . This method has advantages such as non-invasive pulse measurement, miniaturization, and ease of use, which are widely used and are advantageous in that it is easy to develop a wearable life signal detection sensor.

At least one characteristic value calculated from the data received from the PPG sensor by the processor 240 may be as follows.

1. Peak value of measured PPG signal (Peak of PPG and R waveform of ECG have a deep correlation)

2. A signal obtained by dividing peak value data of the measured PPG signal by a predetermined window

3. Obtain the following characteristic values from the peak values of the PPG signal (NN interval is the interval between the peak point of the PPG signal and the next peak point)

(1) meanNN: expected value of NN Interval

(2) SDNN: standard deviation of NN interval

(3) SDSD: Standard deviation of the difference of NN Interval

(4) NNx_Count: Tested 10, 20, 30, 40, 50 as the value of x, which is derived from intervals (NN)

(5) pNNx: Percentage of absolute differences in successive NN values> x (ms) Test 10, 20, 30, 40,

(6) TF: power of the TF (total frequency) band (0.14-0.4 Hz) at the NN interval PSD

(7) VLF: Power of VLF (very low frequency) band (0-0.04Hz)

(8) LF: Power of LF (low frequency) band (0.04-0.15Hz)

(9) HF: Power of HF (high frequency) band (0.15-0.4Hz)

(10) LFn: the ratio of the LF band to the LF + HF band (0.04 to 0.4 Hz)

(11) HFn: ratio of HF band to LF + HF band (0.04-0.4 Hz)

(12) LFHF: ratio of LF band to HF band

4. The processor 240 may calculate the degree of stress by substituting the values of 1, 2, and 3 corresponding to the feature values calculated from the data received from the PPG sensor into a predetermined machine learning algorithm.

When the sensor unit 230 further includes an electromyogram sensor unit 237 for confirming the electrical activity of the muscles of the human body, the processor 240 determines whether or not the blood pressure of the human body is greater than the data received from the blood flow rate sensor unit 237 The feature values may be calculated and the calculated degree of stress may be calculated by applying the calculated at least one feature value to a predetermined machine learning algorithm. The EMG sensor unit 237 may include an EMG sensor as an example.

The processor 240 receives raw data from the electromyogram sensor unit 237 and divides the received raw data into a predetermined time window to extract data from the data divided by a certain window per window The following feature values can be calculated

1. SEF50 (Spectral edge frequency): This index is used to quantify how much bias is applied to the high frequency as compared with the low frequency. It is expressed by the area ratio of the whole frequency region of the bio signal in the spectrum graph.

2. SEF50: The value indicated by the value of the specific frequency when the area under the specific frequency in the spectrum graph is 50% of the area for the whole frequency.

Generally, higher EMF in EMG leads to less high frequency and lower frequency components, and therefore, as muscle fatigue increases, SEF50 also decreases. The processor 240 may calculate or measure the degree of stress by applying the feature values 1 and 2 to a predetermined machine learning algorithm. The processor 240 may determine whether the current human body or the user is receiving the stress based on the degree of stress thus calculated.

INDUSTRIAL APPLICABILITY As described above, the following device of the present invention is advantageous in that the wearer is easy to wear and can monitor the emotional state and stress according to the surrounding situation from time to time in daily life.

Further, the following device according to the present invention is capable of monitoring the fine movement of the head using an acceleration sensor by incorporating or connecting an acceleration sensor to a mechanism that can be worn on the ear in a small size, Can be measured.

The embodiments described above are those in which the elements and features of the present invention are combined in a predetermined form. Each component or feature shall be considered optional unless otherwise expressly stated. Each component or feature may be implemented in a form that is not combined with other components or features. It is also possible to construct embodiments of the present invention by combining some of the elements and / or features. The order of the operations described in the embodiments of the present invention may be changed. Some configurations or features of certain embodiments may be included in other embodiments, or may be replaced with corresponding configurations or features of other embodiments. It is clear that the claims that are not expressly cited in the claims may be combined to form an embodiment or be included in a new claim by an amendment after the application.

It will be apparent to those skilled in the art that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. Accordingly, the above description should not be construed in a limiting sense in all respects and should be considered illustrative. The scope of the present invention should be determined by rational interpretation of the appended claims, and all changes within the scope of equivalents of the present invention are included in the scope of the present invention.

Claims (20)

A device that can be worn on the human ear;
And an acceleration sensor unit connected to the mechanism for detecting head movement of the human body and generating acceleration data according to head movement of the human body; And
And a processor for performing signal processing on the acceleration data received from the acceleration sensor unit to calculate a degree of stress of the human body. The method according to claim 1,
The processor comprising:
Performing prefiltering to extract a frequency component corresponding to a predetermined frequency to be used for calculating the degree of stress from the received acceleration data,
Calculating at least one feature value to be used for calculating the degree of stress from the preprocessed acceleration data,
And applying the calculated at least one feature value to a predetermined machine learning algorithm to calculate the degree of stress. 3. The method of claim 2,
Wherein the predetermined frequency to be used for calculating the degree of stress corresponds to a frequency of 1 Hz to 100 Hz. 3. The method of claim 2,
Wherein the at least one feature value comprises an average signal strength of an acceleration signal corresponding to the preprocessed acceleration data, a standard deviation of the acceleration signal, a number of zero crossings when the acceleration signal is second-differentiated, A quadratic momentum of the signal is normalized to a square of standard deviation, a peak of the acceleration signal, a power spectral density generated by Fourier transform of the acceleration signal, an auto-correlation of the acceleration signal, The number of times the acceleration signal exceeds a certain threshold value in one epoch, the distribution of signal intensity as the third order momentum of the acceleration signal, the entropy of the acceleration signal, the energy in the frequency domain associated with the stress, And a rate of change of the power spectrum over time in a wavenumber analysis. 3. The method of claim 2,
Wherein the predetermined machine learning algorithm comprises a random forest algorithm, a random subspace algorithm, or a bagging algorithm. 3. The method of claim 2,
Wherein the processor performs the prefiltering using a Short-Time Fourier Transform (STFT), a wavelet, or a Multivariate Empirical Mode Decomposition (MEMD). The method according to claim 1,
The sensor unit may further include a body temperature sensor unit for sensing body temperature of the human body,
Wherein the processor calculates a degree of stress that the human body receives based on the data received from the body temperature sensor unit. The method according to claim 1,
The sensor unit may further include a blood flow sensor unit for sensing blood flow of the human body,
Wherein the processor calculates a degree of stress that the human body receives based on the data received from the blood flow sensor unit. The method according to claim 1,
Wherein the sensor unit further includes an electromyogram sensor unit for confirming an electrical activity of the muscles of the human body,
Wherein the processor calculates the degree of stress that the human body receives based on the data received from the electromyogram sensor unit. 9. The method of claim 8,
Wherein the processor calculates at least one feature value from the data received from the blood flow amount sensor unit and calculates a degree of stress the human body receives based on the calculated feature value. 10. The method of claim 9,
Wherein the processor calculates at least one feature value from the data received from the electromyographic sensor unit and calculates a degree of stress the human body receives based on the calculated feature value. 12. The method of claim 11,
Wherein the at least one feature value calculated from the data received from the electromyogram sensor unit is an indicator indicating how much the deviation is deviated toward the high frequency than the low frequency in the data or when the area below the specific frequency is 50% And the value of the specific frequency. The method according to claim 1,
Wherein the acceleration sensor unit generates acceleration data of three axes. A method for calculating the degree of stress of a human body in a following device,
Generating acceleration data according to the head movement of the human body by sensing the head movement of the human body by an acceleration sensor unit connected to a device wearable on the ear of the human body; And
Calculating a degree of stress of the human body by performing signal processing on the generated acceleration data. 15. The method of claim 14,
The step of calculating the degree of stressor includes:
Performing prefiltering on the generated acceleration data to extract a frequency component corresponding to a predetermined frequency to be used for calculating the stress degree;
Calculating at least one feature value to be used for calculating the degree of stress from the preprocessed acceleration data; And
And calculating the degree of stress by applying the calculated at least one characteristic value to a predetermined machine learning algorithm. 15. The method of claim 14,
Sensing the body temperature of the human body and generating data on the body temperature change,
Wherein the step of calculating the degree of stress further comprises the step of calculating a degree of stress that the human body receives based on the data on the generated body temperature change. 15. The method of claim 14,
Sensing the blood flow of the human body and calculating data on the change in the blood flow,
The step of calculating the degree of stress includes:
Calculating at least one characteristic value from the data on the blood flow change; And
And calculating the degree of stress the human body receives based on the calculated characteristic value. 15. The method of claim 14,
Further comprising generating data on an electrical activity of the muscles of the human body,
The step of calculating the degree of stress includes:
Calculating at least one characteristic value from data on the electrical activity of the muscle; And
And calculating the degree of stress the human body receives based on the calculated characteristic value. 15. The method of claim 14,
Wherein the generated acceleration data corresponds to acceleration data of three axes. A computer-readable recording medium having recorded thereon a program for causing a computer to execute a method of calculating a degree of stress of a human body according to any one of claims 14 to 19.

Images