WO2023286313A1 - 信号処理装置および方法 - Google Patents

信号処理装置および方法 Download PDF

Info

Publication number
WO2023286313A1
WO2023286313A1 PCT/JP2022/007140 JP2022007140W WO2023286313A1 WO 2023286313 A1 WO2023286313 A1 WO 2023286313A1 JP 2022007140 W JP2022007140 W JP 2022007140W WO 2023286313 A1 WO2023286313 A1 WO 2023286313A1
Authority
WO
WIPO (PCT)
Prior art keywords
reliability
signal
predicted label
signal quality
processing device
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2022/007140
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
靖英 兵動
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sony Group Corp
Original Assignee
Sony Group Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sony Group Corp filed Critical Sony Group Corp
Priority to US18/576,473 priority Critical patent/US20240306967A1/en
Priority to KR1020237044736A priority patent/KR20240033217A/ko
Priority to EP22841663.2A priority patent/EP4371483A4/en
Priority to CN202280048229.5A priority patent/CN117615709A/zh
Priority to JP2023535093A priority patent/JP7794199B2/ja
Publication of WO2023286313A1 publication Critical patent/WO2023286313A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
    • A61B5/165—Evaluating the state of mind, e.g. depression, anxiety
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/02—Detecting, measuring or recording for evaluating the cardiovascular system, e.g. pulse, heart rate, blood pressure or blood flow
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
    • A61B5/316—Modalities, i.e. specific diagnostic methods
    • A61B5/369—Electroencephalography [EEG]
    • A61B5/372—Analysis of electroencephalograms
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/42—Detecting, measuring or recording for evaluating the gastrointestinal, the endocrine or the exocrine systems
    • A61B5/4261—Evaluating exocrine secretion production
    • A61B5/4266—Evaluating exocrine secretion production sweat secretion
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/68—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient
    • A61B5/6801—Arrangements of detecting, measuring or recording means, e.g. sensors, in relation to patient specially adapted to be attached to or worn on the body surface
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
    • A61B5/7207—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal of noise induced by motion artifacts
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7221—Determining signal validity, reliability or quality
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B5/00—Measuring for diagnostic purposes; Identification of persons
    • A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
    • A61B5/7235—Details of waveform analysis
    • A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems

Definitions

  • the present technology relates to a signal processing device and method, and more particularly to a signal processing device and method capable of improving the robustness of emotion estimation to noise.
  • Emotion estimation systems that estimate human emotions use sensor devices to read these physiological responses as biosignals, extract feature quantities such as physiological indicators that contribute to emotions through signal processing, and extract feature quantities using models obtained through machine learning. to estimate the user's emotion.
  • non-patent document 1 describes an emotion estimation technique that takes into account the influence of noise.
  • the technique described in Non-Patent Document 1 does not include an algorithm that considers the reliability of noise removal, and only evaluates a combination of noise removal and emotion estimation.
  • Patent Document 1 As an example of conventional signal quality determination technology, there is the technology described in Patent Document 1.
  • the bioavailability state is determined with consideration given to reliability, and is finally output.
  • the method for determining the biological valid state in Patent Document 1 is limited to the level of illuminance and face orientation, and applicable applications and situations are restricted.
  • body movement noise is generated by the user's movements. It's connected to.
  • This technology was developed in view of this situation, and is intended to improve the robustness of emotion estimation against noise.
  • a signal processing device includes a feature amount extraction unit that extracts a physiological index that contributes to emotion as a feature amount based on a measured biological signal, An emotional state time-series labeling unit that outputs time-series data of a predicted label of an emotional state according to the constructed identification model, and a result of weighting and adding the predicted label using the predicted label reliability, which is the reliability of the predicted label. and a stabilization processing unit that outputs an emotion estimation result based on.
  • a physiological index that contributes to an emotion is extracted as a feature value based on a measured biosignal, and the time-series data of the feature value is processed by a discrimination model constructed in advance to determine the emotion.
  • Time series data of predicted labels of states are output.
  • an emotion estimation result is output based on the result of weighted addition of the predicted label using the predicted label reliability, which is the reliability of the predicted label.
  • FIG. 1 is a block diagram showing a first configuration example of a biological information processing apparatus; FIG. FIG.
  • FIG. 4 is a diagram showing an image of performing time-series labeling of emotional states within a sliding window;
  • FIG. 4 is a diagram showing a calculation method for calculating a representative value of predictive labels of emotional states within a sliding window;
  • 9 is a flowchart for explaining processing of the biological information processing apparatus of FIG. 8;
  • FIG. 4 is a block diagram showing a second configuration example of the biological information processing apparatus;
  • FIG. 4 is a diagram showing an image of performing time-series labeling of emotional states within a sliding window;
  • FIG. 4 is a diagram showing a calculation method for calculating a representative value of predictive labels of emotional states within a sliding window;
  • FIG. 4 is a diagram showing examples of waveforms of electroencephalogram signals for each time; It is a figure which shows the example of quality stability determination using an electroencephalogram.
  • FIG. 10 is a diagram showing an example of calculating the weight of the signal quality score from each type and each ch; 13 is a flowchart for explaining processing of the biological information processing apparatus of FIG. 12; It is a block diagram which shows the structural example of a computer.
  • signals are transmitted from the brain via the autonomic nervous system, causing changes in each function such as breathing, skin temperature, sweating, heart, and vascular activity.
  • central nervous system activities such as electroencephalograms, which can be noninvasively measured by the human head
  • autonomic nervous system activities such as heartbeat and perspiration are known as physiological responses that express human emotions (for example, arousal).
  • EEG Electroencephalogram
  • Heart rate can be measured by measuring the electrical activity of the heart (electrocardiogram; ECG (Electrocardiogram)) or by optically measuring changes in volume of blood vessels that occur as the heart pumps blood (photoplethysmography; PPG (Photoplethysmography)). .
  • ECG Electrocardiogram
  • PPG Photoplethysmography
  • the commonly known heart rate is calculated by averaging the reciprocals of the beat-to-beat intervals (average heart rate).
  • the degree of variability in beat-by-beat intervals is measured as heart rate variability, and various physiological indices representing variability are defined (LF, HF, LF/HF, pNN (percentage of adjacent normal-to-normal intervals)50, RMSSD(root mean square successful difference), etc.).
  • Perspiration (mental perspiration) is expressed on the body surface as a change in the conductivity of the skin (Electrodermal Activity (EDA)), and can be electrically measured as a change in the conductance value of the skin. This is called Skin Conductance.
  • Physiological indices extracted from skin conductance signals are broadly divided into Skin Conductance Response (SCR), which represents instantaneous sweating activity, and Skin Conductance Level (SCL), which represents gradual changes in skin surface conditions. be. In this way, a plurality of feature quantities representing autonomic nerve activity are defined.
  • Emotion estimation systems that estimate the user's emotions read these physiological responses as biosignals using sensor devices, extract features such as physiological indicators that contribute to emotional responses through signal processing, and use models obtained by machine learning to extract features. Estimating the user's emotion from the quantity.
  • wearable devices For application in daily life and in the real world, wearable devices that can be worn on the wrist or ear and are easy to use in daily life are expected to realize emotion estimation by sensing autonomic nerve activity.
  • wearable devices such as VR (virtual reality) head-mounted displays that can be worn naturally on the head, and earphones that use technology that measures brain waves in and around the ears, such as In-ear EEG/Around-ear EEG, which has been researched in recent years. It is expected that emotion estimation based on brain wave sensing will be realized by wearable devices that naturally adapt to the user's experience, such as headphones and wearable devices.
  • the user's body movement affects applications in daily life and real environments.
  • the signal strength of biological signals is weak, and in order to perform accurate sensing in daily life and in real environments, it is necessary to improve robustness against the effects of noise accompanying user body movements.
  • Techniques for reducing the effects of noise caused by user body movements include (1) techniques for separating noise components using signal separation techniques such as adaptive filters and principal component analysis and independent component analysis, and (2) analysis of signal waveforms.
  • signal separation techniques such as adaptive filters and principal component analysis and independent component analysis
  • analysis of signal waveforms There are two main methods for identifying classes according to signal quality. These methods have been researched and developed mainly in the fields of medical engineering and signal processing, and are also applied to products such as general wearable heart rate sensors.
  • Non-Patent Document 1 A specific conventional technology is the technology described in Non-Patent Document 1.
  • the technique described in Non-Patent Document 1 introduces a noise reduction technique and evaluates emotion estimation performance in order to improve robustness against noise that occurs in a real environment.
  • Non-Patent Document 1 only introduces noise removal technology in the preprocessing stage of emotion estimation. Therefore, if the noise removal works well, the user's emotion can be estimated robustly. can occur.
  • noise can be reduced by obtaining a noise reference signal and utilizing statistical properties such as correlation analysis, principal component analysis, and independent component analysis. Removal is generally difficult.
  • Patent Document 1 A related technology is the technology described in Patent Document 1.
  • the bioavailability state is determined with consideration given to reliability, and is finally output.
  • the target of the technology described in Patent Document 1 is limited to vehicle technology and remote optical vital sensors, and the method of determining the biological effective state is limited to the level of illuminance and face orientation, and applicable applications and situations are limited. be.
  • the emotion estimation result is output based on the result of weighted addition of the predicted label using the predicted label reliability, which is the reliability of the predicted label.
  • FIG. 1 is a diagram showing a configuration example of an emotion estimation processing system according to an embodiment of the present technology.
  • the emotion estimation processing system 1 of FIG. 1 includes a biological information processing device 11 .
  • the emotion estimation processing system 1 may include a server 12, a terminal device 13, and a network 14. In that case, in the emotion estimation processing system 1 , the biological information processing device 11 , the server 12 and the terminal device 13 are interconnected via the network 14 .
  • the emotion estimation processing system 1 is a system that detects a signal related to the state of a living body (hereinafter referred to as a biological signal) and estimates the emotion of the living body based on the detected biological signal.
  • a biological signal a signal related to the state of a living body
  • the biological information processing device 11 of the emotion estimation processing system 1 is directly attached to a living body in order to detect biological signals.
  • the biological information processing device 11 is used, for example, as shown in FIGS. 2 and 3 to estimate the emotion of the biological body.
  • FIGS. 2 and 3 are diagrams showing how the biological information processing device 11 is worn on a living body.
  • a wristband type biological information processing device 11 such as a wristwatch type is attached to the wrist of the user U1.
  • a headband type biological information processing device 11 such as a forehead contact type is wrapped around the head of the user U1.
  • the biological information processing apparatus 11 includes a biological sensor that detects a biological signal for estimating a biological emotion such as the perspiration state, pulse wave, myoelectric potential, blood pressure, blood flow, or body temperature of the user U1.
  • a biological emotion such as the perspiration state, pulse wave, myoelectric potential, blood pressure, blood flow, or body temperature of the user U1.
  • the emotion of the user U1 is estimated based on the obtained biosignal. Based on this emotion, the user's concentration state, wakefulness state, and the like can be confirmed.
  • FIGS. 2 and 3 show examples in which the biological information processing device 11 is worn on the arm or the head, the mounting position of the biological information processing device 11 is not limited to the examples in FIGS. Absent.
  • the biological information processing device 11 may be implemented in a form that can be attached to a part of the hand, such as a wristband, glove, smartwatch, or ring. Further, when the biological information processing device 11 contacts a part of the living body such as a hand, the biological information processing device 11 may be provided on an object that can come into contact with the user, for example.
  • the biological information processing device 11 can come into contact with the user through a mobile terminal, smart phone, tablet, mouse, keyboard, steering wheel, lever, camera, exercise equipment (golf club, tennis racket, archery bow, etc.), or writing equipment. It may be provided on the surface or inside the object.
  • the biological information processing apparatus 11 may be a part of the user's head, such as a head-mounted display (FIG. 4), headphones (FIG. 5), earphones (FIG. 6), a hat, accessories, goggles, or glasses (FIG. 7). It may be realized by a form that can be worn on the ear or on the ear.
  • FIG. 4 to 7 are diagrams showing other aspects of the biological information processing device 11.
  • FIG. 4 to 7 are diagrams showing other aspects of the biological information processing device 11.
  • FIG. 4 shows a head-mounted display type biological information processing device 11 .
  • the pad section 21, the band section 22, and the like are worn on the user's head.
  • FIG. 5 a headphone-type biological information processing device 11 is shown.
  • the band portion 31, the ear pads 32, and the like are worn on the user's head and ears.
  • FIG. 6 an earphone-type biological information processing device 11 is shown.
  • the earpiece 41 is attached to the user's ear.
  • FIG. 7 a spectacles-type biological information processing device 11 is shown.
  • the temples 51 are worn above the ears of the user.
  • the biological information processing device 11 may be provided in clothing such as sportswear, socks, underwear, armor, or shoes.
  • the mounting position and mounting method of the biological information processing device 11 are not particularly limited as long as the biological information processing device 11 can detect signals related to the state of the living body.
  • the biological information processing device 11 does not have to be in direct contact with the body surface of the living body.
  • the biological information processing device 11 may be in contact with the surface of the living body through clothing, a detection sensor protective film, or the like.
  • the above-described biological information processing device 11 does not necessarily need to perform processing by itself.
  • the biological information processing apparatus 11 includes a biological sensor that contacts a living body, transmits a biological signal detected by the biological sensor to another device such as the server 12 or the terminal device 13, and transmits the received biological signal to another device such as the server 12 or the terminal device 13.
  • the emotion of the living body may be estimated by performing information processing based on.
  • the biometric information processing device 11 transmits a biosignal acquired from the biosensor to the server 12 or a terminal device 13 such as a smartphone,
  • the server 12 or the terminal device 13 may perform information processing to estimate the emotion of the living body.
  • the biosensors provided in the biometric information processing apparatus 11 come into contact with the surface of the living body in various ways as described above and detect biosignals. Therefore, the measurement results of the biosensor are likely to be affected by fluctuations in the contact pressure between the biosensor and the living body caused by body movements of the living body. For example, a biosignal acquired from a biosensor contains noise caused by the body movement of the living body. It is desired to accurately estimate a biological emotion from such biological signals containing noise.
  • the body movement of the living body refers to the overall motion form when the living body operates. and other biological actions.
  • the contact pressure between the biosensor included in the biometrics information processing device 11 and the user U1 may fluctuate due to such a user's action.
  • the biological information processing apparatus 11 is equipped with a second sensor and a third sensor, which will be described below, in addition to the biological sensor described above. may
  • the second sensor is configured to detect changes in body motion of the living body.
  • the third sensor is configured to detect pressure changes in the living body in the sensing area of the biosensor.
  • the biological information processing apparatus 11 uses body movement signals and pressure signals detected by the second sensor and the third sensor to accurately reduce body movement noise from the biological signals detected by the biological sensors. can be done.
  • the biological signal corrected in this manner may be used to perform the emotion estimation processing of the present technology, which will be described below.
  • the server 12 is composed of a computer and the like.
  • the terminal device 13 is configured by a smart phone, a mobile terminal, a personal computer, or the like.
  • the server 12 and the terminal device 13 receive information and signals transmitted from the biological information processing device 11 and transmit information and signals to the biological information processing device 11 via the network 14 .
  • the server 12 and the terminal device 13 receive from the biological information processing apparatus 11 a biological signal obtained by a biological sensor included in the biological information processing apparatus 11, and perform signal processing on the received biological signal.
  • a biological signal obtained by a biological sensor included in the biological information processing apparatus 11, and perform signal processing on the received biological signal.
  • the network 14 is configured by the Internet, a wireless LAN (Local Area Network), and the like.
  • FIG. 8 is a block diagram showing a first configuration example of the biological information processing apparatus 11. As shown in FIG.
  • the biological information processing apparatus 11 is composed of a filter preprocessing unit 101, a feature amount extraction unit 102, an emotional state time-series labeling unit 103, and a stabilization processing unit 104.
  • the filter preprocessing unit 101 performs preprocessing such as a bandpass filter and noise removal on the measured biological signal.
  • the filter preprocessing unit 101 outputs the preprocessed biological signal to the feature quantity extraction unit 102 .
  • the biological signal is an electroencephalogram (EEG)
  • EEG electroencephalogram
  • the electroencephalogram is measured by attaching electrodes to the scalp and measuring the action potential of the brain that leaks through the scalp, skull, etc.
  • One of the characteristics of electroencephalograms measured in this way is known to be a very low SN ratio. Therefore, it is necessary to remove signals of unnecessary frequency components from EEG, which is a time-series signal, and a bandpass filter is applied (for example, a passband of 0.1 Hz to 40 Hz).
  • body motion components due to human body motion are superimposed on the biological signal as artifacts (noise other than the target signal, etc.).
  • signal processing techniques such as adaptive filters and independent component analysis are applied. Filter preprocessing is also performed in the same way when the biological signal is psychogenic perspiration (EDA), pulse wave (PPG), blood flow (LDF), or the like.
  • the feature amount extraction unit 102 converts signals from vital sensors (biological sensors) such as electroencephalogram (EEG), psychogenic perspiration (EDA), pulse wave (PPG), and blood flow (LDF) into time-series data. , and extract physiological indices that contribute to changes in emotion as features.
  • vital sensors biological sensors
  • EEG electroencephalogram
  • EDA psychogenic perspiration
  • PPG pulse wave
  • LDF blood flow
  • electroencephalogram can be measured by measuring the action potential of the brain that leaks through the scalp, skull, etc., using electrodes that are generally attached to the scalp, as described above.
  • Neuroscience letters 310.1 (2001): 57- 60 (hereinafter referred to as Cited Document 1) describes that the characteristics of the frequency components of signals such as ⁇ waves, ⁇ waves, and ⁇ waves are extracted as physiological indices that contribute to human emotion in electroencephalograms.
  • EDA psychogenic perspiration
  • SCR Skin Conductance Response
  • SCL Skin Conductance Level
  • heart rate and heart rate variability can be extracted from the pulse wave (PPG), but heart rate variability includes physiological indices such as mHR, LF, HF, LF/HF, pNN50, and RMSSD.
  • feature values are extracted using an analysis window (sliding window) of about 5 minutes.
  • an analysis window sliding window
  • Salahuddin, Lizawati, et al. “Ultrashort term analysis of heart rate variability for monitoring mental stress in mobile settings.” 2007 29th annual international conference of the ieee engineering in medicine and biology society. IEEE, 2007.
  • Reference 3 shows the validity of HRV with a short window of about several tens of seconds as Ultra-short-term HRV considering real-time properties.
  • the feature amount extraction unit 102 can also perform signal processing for extracting feature amounts that contribute to emotion in a data-driven manner, for example, by deep learning or autoencoder.
  • the method of time-series labeling of emotional states is generally assumed to be a discriminative model used in time-series data analysis and natural language processing.
  • SVM Support Vector Machine
  • k-NN k-Nearest Neighbor
  • LDA Linear Discriminant Analysis
  • HMM Hidden Markov Models
  • CRF Conditional Random Fields
  • SOSVM Structured Output Support Vector Machine
  • Bayesian Network Recurrent Neural Network
  • RNN Long Short Term Memory
  • LSTM Long Short Term Memory
  • the predicted label of the state is weighted and added according to the reliability of the predicted label of the emotional state within the sliding window. Output the confidence level r.
  • the reliability r of the representative value of the predicted label is the reliability when calculating the representative value of the predicted label.
  • the stabilization processing unit 104 calculates the reliability r of the representative value of the predicted label within the sliding window by weighting and adding the predicted label of the emotional state within the sliding window according to the reliability of the predicted label. do. Furthermore, the stabilization processing unit 104 performs threshold processing on the reliability r of the representative value of the predicted label, and outputs the representative value z of the predicted label as the emotion estimation result.
  • FIG. 9 is a diagram showing an image of how the stabilization processing unit 104 performs time-series labeling of emotional states within a sliding window.
  • FIG. 9 shows the inside of the sliding window in units of subsequences.
  • y is the predicted label of the emotional state
  • c is the reliability of the predicted label obtained from the discrimination model
  • ⁇ t i is the duration of the i-th event.
  • y which is the predicted label of the emotional state, is calculated for each event within the sliding window, along with the reliability c of the predicted label.
  • the predictive label y of the emotional state is identified by, for example, the degree of arousal
  • the low/high degree of arousal is defined by two classes of 0 or 1.
  • FIG. 9 for convenience of explanation, an example in which there is a gap between event durations is shown, but there is no need to leave a gap between event durations.
  • FIG. 10 is a diagram showing a calculation method for calculating the representative value of the predicted label according to the predicted label of the time-series emotional state within the sliding window and its reliability in the stabilization processing unit 104 .
  • the reliability r of the representative value of the predicted label is calculated by the following formula (1).
  • i represents the event number in multiple events detected within the sliding window.
  • y is the predicted label of the emotional state
  • c is the reliability of the predicted label obtained from the discriminative model
  • ⁇ t i is the duration of the i-th event.
  • w is a forgetting weight, and the weight is decreased as the time goes past.
  • the reliability of the representative value of the predicted label within the sliding window is continuous [-1 1] with respect to the reliability of the predicted label of the emotional state of multiple events detected within the sliding window. calculated as a value.
  • the representative value z of the predicted label is calculated as the emotion estimation result.
  • Equation (2) the numerical value of the representative value z of the predicted label of the emotion estimation result depends on the definition of the predicted label y of the user's emotional state.
  • the predictive label of the emotional state is to identify the degree of arousal
  • the predictive label of the emotional state is defined as two classes, 0 or 1, as low/high arousal.
  • the representative value z of the prediction label of the emotion estimation result is 0, the user's emotional state at the corresponding time is identified as being in a low arousal level (relaxed state).
  • the representative value z of the prediction label of the emotion estimation result is 1, the user's emotional state at the corresponding time is identified as being in a high arousal level (arousal and concentration state).
  • FIG. 11 is a flow chart for explaining the processing of the biological information processing apparatus 11 of FIG.
  • step S101 the filter preprocessing unit 101 preprocesses the biological signal measured by the biological sensor.
  • the filter preprocessing unit 101 outputs the preprocessed biological signal to the feature quantity extraction unit 102 .
  • step S ⁇ b>102 the feature amount extraction unit 102 extracts feature amounts based on the biological signal supplied from the filter preprocessing unit 101 .
  • step S103 the emotional state time-series labeling unit 103 receives as input the feature amount within the sliding window among the feature amounts supplied from the feature amount extraction unit 102, and performs time-series labeling of the emotional state.
  • the emotional state time-series labeling unit 103 outputs to the stabilization processing unit 104 the predicted label of the emotional state in the time series, which is the result of labeling the emotional state time-series.
  • step S104 the stabilization processing unit 104 receives the time-series predicted label of the emotional state supplied from the emotional state time-series labeling unit 103, and uses the above-described formula (1) to calculate the predicted label within the sliding window. Calculate the reliability r of the representative value of
  • step S105 the stabilization processing unit 104 performs threshold processing on the reliability r of the representative value of the predicted label using Equation (2) described above, and outputs the representative value z of the predicted label as the emotion estimation result.
  • the emotion estimation result is output based on the result of weighted addition of the predicted label using the predicted label reliability, which is the reliability of the predicted label. This improves the robustness of estimation accuracy of emotion estimation.
  • FIG. 12 is a block diagram showing a second configuration example of the biological information processing apparatus 11. As shown in FIG.
  • a signal quality determination unit 201 is added in order to further improve the robustness of the noise in emotion estimation when noise occurs due to body movement or the like in the real environment.
  • the biological information processing apparatus 11 of FIG. 12 differs from the biological information processing apparatus 11 of FIG. 8 in that the signal quality determination unit 201 is added and the stabilization processing unit 104 is replaced with the stabilization processing unit 202. different.
  • FIG. 12 parts corresponding to those in FIG. 8 are denoted by the same reference numerals.
  • the signal quality determination unit 201 analyzes the waveform of the biosignal measured by the biosensor and identifies the type of artifact. Signal quality determination section 201 determines signal quality based on the identification result, and calculates a signal quality score as the signal quality determination result.
  • the stabilization processing unit 202 weights and adds the reliability of the predicted label of the emotional state and the signal quality score that is the determination result of the signal quality determination unit 201, and obtains the representative value z of the predicted label and the predicted label as the emotional estimation result. output the reliability r of the representative value of
  • FIG. 13 is a diagram showing an image of how the stabilization processing unit 202 performs time-series labeling of emotional states within a sliding window.
  • y is the predicted label of the emotional state
  • c is the reliability of the predicted label obtained from the discrimination model
  • ⁇ t i is the duration of the i-th event.
  • FIG. 13 shows the signal quality score s, which is the output of the signal quality determination section 201 at the same time as the sliding window.
  • y which is the predicted label of the emotional state, is calculated for each event within the sliding window, along with the reliability c of the predicted label, in the same way as in FIG. Furthermore, at the same time as the sliding window, signal quality determination section 201 outputs signal quality score s to stabilization processing section 202 .
  • FIG. 14 is a diagram showing a calculation method for calculating the representative value of the predicted label according to the predicted label of the emotional state within the sliding window, its reliability, and the signal quality score in the stabilization processing unit 202.
  • FIG. 14 is a diagram showing a calculation method for calculating the representative value of the predicted label according to the predicted label of the emotional state within the sliding window, its reliability, and the signal quality score in the stabilization processing unit 202.
  • Time-series data of the signal quality score s is calculated by the signal quality determination unit 201 and output to the stabilization processing unit 202 .
  • the stabilization processing unit 202 uses this signal quality score s and feeds back the signal quality as a weight for calculating the reliability of the representative value of the predicted label.
  • a method of calculating the reliability r of the representative value to which the signal quality is fed back can be defined as the following equation (3) based on the above equation (1).
  • the reliability of the representative value of the predicted label within the sliding window is continuous [-1 1] with respect to the reliability of the predicted label of the emotional state of multiple events detected within the sliding window. calculated as a value.
  • the representative value z of the predicted label is calculated as the emotion estimation result. be done.
  • Equation (3) has the property that the reliability r decreases in a sliding window with low signal quality.
  • Equation (4) can be normalized by including s i in the denominator. This makes it possible to perform unified emotion determination even between sliding windows with different signal qualities.
  • FIG. 15 is a diagram showing an example of waveforms of electroencephalogram signals over time.
  • Eye movement noise is, for example, noise that occurs on electroencephalogram signals placed particularly in the forehead when a person moves his or her line of sight.
  • Myoelectric noise is noise that is generated in electroencephalogram signals when, for example, a person changes facial expressions or moves muscles when conversing.
  • Blink noise is, for example, noise that occurs especially in brain wave signals of the frontal region when the frequency or intensity of blinking of a person changes.
  • a change derived from a person's heartbeat appears as a potential change, and the potential change can become electrocardiographic noise on an electroencephalogram signal.
  • the electroencephalogram signals on which various noises are superimposed have different waveforms. Therefore, by analyzing the waveforms, the signal waveform in the region where each artifact occurs can be identified, and the type of artifact can be identified. At that time, pattern matching, signal processing for waveform identification, machine learning technology, and the like are used for signal quality determination.
  • classification is performed using a classification model, which is a machine learning model, class classification is performed for each waveform of the biosignal, and signal quality determination is performed based on the class classification results.
  • a classification model which is a machine learning model
  • a signal quality score (SQE score) that is specialized for the processing (equation (3)) in the stabilization processing unit 202 is calculated. .
  • FIG. 16 is a diagram showing an example of quality stability determination using electroencephalograms.
  • a discrimination class is defined in advance when each noise occurs, and a discrimination model is constructed by supervised learning.
  • the discrimination model for quality judgment is referred to as the SQE discrimination model, which stands for Signal Quality Estimation
  • the predefined discrimination class is referred to as the SQE discrimination class.
  • the signal quality determination unit 201 identifies the waveform type using the SQE identification model. Then, signal quality determination section 201 calculates a signal quality score s specific to the signal processing method defined by Equation (3) above. A signal quality score s is calculated by the following equation (5).
  • m is the SQE discrimination class
  • ⁇ m is the class label corresponding to the SQE discrimination class (constant: preset [0, 1])
  • d m is the confidence level of the class label obtained from the SQE discrimination model ( [0,1] depending on the input signal)
  • f() is a function defined as a look-up table for adjustment (preset [0,1]).
  • ⁇ is an adjustment term that considers the performance difference of noise removal in the filter preprocessing unit 101 according to the type of noise identified by the SQE identification class.
  • Equation (3) when the SQE discrimination model determines that the electroencephalogram signal is clean, in Equation (3), the higher the reliability of the class label obtained from the SQE discrimination model, the greater the weight, and the positive To identify the classes, f() is a monotonically increasing look-up table.
  • a positive class is a class identified as having signal quality better than a predetermined threshold.
  • Negative classes are classes identified as noisy with signal quality worse than a predetermined threshold.
  • f() is set to It is assumed to be a monotonically decreasing look-up table.
  • ⁇ is adjusted according to the SQE discrimination class and the performance difference of the filter preprocessor 101 .
  • ⁇ m is positioned as an adjustment term, and its value is not restricted.
  • f(d m ) is monotonically increasing when m is the main signal and monotonically decreasing when m is the noise.
  • the signal quality score s[0.0 1.0] increases as the signal quality increases and decreases as the signal quality decreases. ) as a specialized signal processing method.
  • this technology also targets the process of estimating emotions from multiple types of biosignal modals (for example, multimodal signals such as electroencephalogram (EEG), pulse wave (PPG), and perspiration (EDA)). It is also assumed that an SQE identification model is constructed and the signal quality is determined.
  • EEG electroencephalogram
  • PPG pulse wave
  • EDA perspiration
  • FIG. 17 is a diagram showing an example of calculating the weight of the signal quality score (SQE score) from each type and each ch.
  • a graph is shown that shows the degree of contribution to the identification of the emotional state in the feature amount according to the signal of each type and each channel.
  • the graph shows features calculated from brain waves (theta, alpha, and beta waves), features calculated from pulse waves (average heart rate, RMSSD, and LF/HF), and features calculated from perspiration. Quantities (SCL and SCR) are indicated.
  • a comprehensive signal quality score s of estimation results is calculated by weighted addition of SQE scores for each of j types of signals. Specifically, the signal quality score s is calculated by the following equation (6). ... (6)
  • this technology can also be applied when calculating the SQE score weight from each type and each ch.
  • FIG. 18 is a flow chart for explaining the processing of the biological information processing apparatus 11 of FIG.
  • Steps S201 to S203 in FIG. 18 perform the same processing as steps S101 to S103 in FIG. 11, so description thereof will be omitted.
  • steps S204 and S205 are performed in parallel with the processes of steps S201 to S203.
  • step S204 the signal quality determination unit 201 analyzes the signal waveform of each biosensor and identifies the waveform type.
  • step S205 the signal quality determination unit 201 calculates a signal quality score according to the waveform type.
  • Signal quality determination section 201 outputs the calculated signal quality score to stabilization processing section 202 .
  • step S206 the stabilization processing unit 202 inputs the time-series emotional state label supplied from the emotional state time-series labeling unit 103 and the signal quality score supplied from the stabilization processing unit 202, and uses the above equation (3 ) to calculate the reliability r of the representative value of the predicted labels within the sliding window.
  • step S207 the stabilization processing unit 202 performs threshold processing on the reliability r of the representative value of the predicted label using Equation (2) described above, and outputs the representative value z of the predicted label as the emotion estimation result.
  • the result of emotion estimation is output based on the result of weighted addition of the reliability of the predicted label and the result of signal quality determination. Therefore, the estimation accuracy of emotion estimation is further improved in robustness compared to the first embodiment.
  • a pulse wave has a strong periodicity corresponding to pulsation when it can be measured normally.
  • noise such as body movement noise occurs, the periodicity of the signal becomes low.
  • Cited Document 6 International Publication No. 2017/199597 (hereinafter referred to as Cited Document 6), the periodicity of the pulse wave signal is analyzed by analyzing the autocorrelation of the signal (the relationship between the shift amount of the signal itself and the correlation value). stated to be evaluated.
  • the autocorrelation value is low, it is recognized that the periodicity is low, and signal quality can be determined.
  • the signal quality determination unit 201 may output a signal quality score according to the strength of signal periodicity without using machine learning.
  • wearable devices such as wristbands, headbands, and earphones, which can be used in daily life environments, reduce the burden on users and measure neural activity that contributes to emotional changes. In particular, it is expected to easily measure autonomic nerve activity (pulse wave, perspiration, etc.).
  • wearable devices that can be worn naturally on the head, such as VR head-mounted displays, and earphones and headphones that use technology that measures brain waves in and around the ears, such as In-ear EEG/Around-ear EEG, which has been researched in recent years.
  • Wearable devices such as wearable devices, are expected to realize emotion estimation by brain wave sensing.
  • the user's body movement affects applications in daily life and in the real environment.
  • the signal strength of biosignals is weak, and in order to perform accurate sensing in daily life and in real environments, it is necessary to improve robustness against the effects of noise caused by user's body movements.
  • the emotion estimation result is output based on the result of weighted addition of the predicted label using the predicted label reliability, which is the reliability of the predicted label.
  • the emotion estimation result is output based on the result of weighted addition of the reliability of the predicted label of the emotional state and the result of the signal quality determination.
  • FIG. 19 is a block diagram showing a hardware configuration example of a computer that executes the series of processes described above by a program.
  • a CPU (Central Processing Unit) 301 , a ROM (Read Only Memory) 302 and a RAM (Random Access Memory) 303 are interconnected by a bus 304 .
  • An input/output interface 305 is further connected to the bus 304 .
  • the input/output interface 305 is connected to an input unit 306 such as a keyboard and a mouse, and an output unit 307 such as a display and a speaker.
  • the input/output interface 305 is also connected to a storage unit 308 such as a hard disk or nonvolatile memory, a communication unit 309 such as a network interface, and a drive 310 that drives a removable medium 311 .
  • the CPU 301 loads a program stored in the storage unit 308 into the RAM 303 via the input/output interface 305 and the bus 304 and executes the above-described series of processes. is done.
  • the program executed by the CPU 301 is recorded on the removable media 311, or provided via a wired or wireless transmission medium such as a local area network, the Internet, or digital broadcasting, and installed in the storage unit 308.
  • the program executed by the computer may be a program in which processing is performed in chronological order according to the order described in this specification, or in parallel or at a necessary timing such as when a call is made. It may be a program in which processing is performed.
  • a system means a set of multiple components (devices, modules (parts), etc.), and it does not matter whether all the components are in the same housing. Therefore, a plurality of devices housed in separate housings and connected via a network, and a single device housing a plurality of modules in one housing, are both systems. .
  • Embodiments of the present technology are not limited to the above-described embodiments, and various modifications are possible without departing from the gist of the present technology.
  • this technology can take the configuration of cloud computing in which one function is shared by multiple devices via a network and processed jointly.
  • each step described in the flowchart above can be executed by a single device, or can be shared and executed by a plurality of devices.
  • the multiple processes included in the one step can be executed by one device or shared by multiple devices.
  • This technique can also take the following configurations.
  • a feature amount extraction unit that extracts a physiological index that contributes to emotion as a feature amount based on the measured biological signal;
  • an emotional state time-series labeling unit for outputting time-series data of predictive labels of emotional states based on a discrimination model constructed in advance for the time-series data of the feature amount;
  • a signal processing device comprising: a stabilization processing unit that outputs an emotion estimation result based on a result of weighted addition of the predicted label using predicted label reliability, which is the reliability of the predicted label.
  • the emotional state time-series labeling unit outputs the time-series data of the predicted label of the emotional state within the sliding window using the identification model for the time-series data of the feature amount within the sliding window.
  • the stabilization processing unit A representative value reliability, which is the reliability of the representative value of the predicted label, is calculated by weighted addition of the predicted label and the predicted label reliability, and threshold processing of the representative value reliability is performed as the emotion estimation result. , which outputs a representative value of the predicted label.
  • the signal processing device according to (1) or (2).
  • the stabilization processing unit sets the reliability of the representative value to Calculated by The signal processing device according to (3), wherein y is the predicted label, c is the predicted label reliability, and ⁇ t i is the duration of the i-th segment. (5) Further comprising a signal quality determination unit that determines the signal quality of the biosignal, The stabilization processing unit outputs the emotion estimation result based on the result of weighted addition of the predicted label using the predicted label reliability and the signal quality determination result. (1) or (2) The signal processing device according to . (6) The stabilization processing unit calculates a representative value reliability, which is the reliability of the representative value of the predicted label, by performing weighted addition on the predicted label using the predicted label reliability and the determination result of the signal quality.
  • the stabilization processing unit sets the reliability of the representative value to Calculated by y is the predicted label, c is the predicted label reliability, s is the signal quality score that is the signal quality determination result, and ⁇ t i is the duration of the i-th event. signal processor.
  • the signal quality determination unit uses two or more class labels a identified by a quality determination identification model, the reliability d of the class label, and a function f () for adjusting the reliability d,
  • the function f() is monotonically increasing for classes identified as having better signal quality than a predetermined threshold in each of one or more classes, and identified as containing noise with a signal quality worse than a predetermined threshold.
  • the signal processing device according to (8) above which is monotonically decreasing with respect to the class.
  • the stabilization processing unit sets the reliability of the representative value of the predicted label within the sliding window to:
  • the weight Wj of the weighted addition of the signal quality score is the model contribution wjk of the feature k belonging to j types of signals, is represented by The signal processing device according to (11) above.
  • the stabilization processing unit sets the reliability of the representative value to (y is the predicted label, c is the reliability of the predicted label, s is the signal quality score, and ⁇ t i is the duration of the i-th event)
  • the housing is configured to be wearable.
  • a signal processing device Based on the measured biosignals, physiological indicators that contribute to emotion are extracted as feature quantities, outputting time-series data of predictive labels of emotional states based on a pre-constructed discrimination model for the time-series data of the feature quantity; A signal processing method for outputting an emotion estimation result based on a result of weighted addition of the predicted label using predicted label reliability, which is reliability of the predicted label.

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Medical Informatics (AREA)
  • Surgery (AREA)
  • Public Health (AREA)
  • Molecular Biology (AREA)
  • Veterinary Medicine (AREA)
  • Biophysics (AREA)
  • Pathology (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Psychiatry (AREA)
  • Physiology (AREA)
  • Signal Processing (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Psychology (AREA)
  • Social Psychology (AREA)
  • Child & Adolescent Psychology (AREA)
  • Developmental Disabilities (AREA)
  • Educational Technology (AREA)
  • Hospice & Palliative Care (AREA)
  • Cardiology (AREA)
  • Endocrinology (AREA)
  • Gastroenterology & Hepatology (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
PCT/JP2022/007140 2021-07-15 2022-02-22 信号処理装置および方法 Ceased WO2023286313A1 (ja)

Priority Applications (5)

Application Number Priority Date Filing Date Title
US18/576,473 US20240306967A1 (en) 2021-07-15 2022-02-22 Signal processing apparatus and method
KR1020237044736A KR20240033217A (ko) 2021-07-15 2022-02-22 신호 처리 장치 및 방법
EP22841663.2A EP4371483A4 (en) 2021-07-15 2022-02-22 SIGNAL PROCESSING DEVICE AND METHOD
CN202280048229.5A CN117615709A (zh) 2021-07-15 2022-02-22 信号处理设备和方法
JP2023535093A JP7794199B2 (ja) 2021-07-15 2022-02-22 信号処理装置および方法

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2021-117097 2021-07-15
JP2021117097 2021-07-15

Publications (1)

Publication Number Publication Date
WO2023286313A1 true WO2023286313A1 (ja) 2023-01-19

Family

ID=84919220

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/007140 Ceased WO2023286313A1 (ja) 2021-07-15 2022-02-22 信号処理装置および方法

Country Status (6)

Country Link
US (1) US20240306967A1 (https=)
EP (1) EP4371483A4 (https=)
JP (1) JP7794199B2 (https=)
KR (1) KR20240033217A (https=)
CN (1) CN117615709A (https=)
WO (1) WO2023286313A1 (https=)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
NL2037018A (en) * 2023-02-16 2024-08-29 Milbotix Ltd Monitoring a person using machine learning
WO2025211253A1 (ja) * 2024-04-05 2025-10-09 ソニーグループ株式会社 情報処理方法、情報処理装置、および学習方法

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN118490233B (zh) * 2024-07-18 2024-12-03 南昌航空大学 一种基于脑电信号和眼动信号的动态情绪识别方法及系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017199597A1 (ja) 2016-05-20 2017-11-23 ソニー株式会社 生体情報処理装置、生体情報処理方法、及び情報処理装置
WO2018218286A1 (en) * 2017-05-29 2018-12-06 Saltor Pty Ltd Method and system for abnormality detection
JP2019170967A (ja) 2018-03-29 2019-10-10 パナソニックＩｐマネジメント株式会社 生体信号検知装置及び生体信号検知方法
JP2020048622A (ja) * 2018-09-24 2020-04-02 トヨタ自動車株式会社 生体状態推定装置
JP2020203051A (ja) * 2019-06-19 2020-12-24 株式会社プロアシスト コンピュータプログラム、情報処理装置、情報処理方法、学習済みモデルの生成方法及び学習済みモデル
JP2021053261A (ja) * 2019-10-01 2021-04-08 Ｎｅｃソリューションイノベータ株式会社 解析方法、解析装置、プログラム

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2009282686A (ja) 2008-05-21 2009-12-03 Toshiba Corp 分類モデル学習装置および分類モデル学習方法
US8463595B1 (en) * 2012-03-06 2013-06-11 Reputation.Com, Inc. Detailed sentiment analysis
US10512407B2 (en) * 2013-06-24 2019-12-24 Fitbit, Inc. Heart rate data collection
WO2017136938A1 (en) * 2016-02-10 2017-08-17 Tandemlaunch Inc. A quality adaptive multimodal affect recognition system for user-centric multimedia indexing
EP3238611B1 (en) * 2016-04-29 2021-11-17 Stichting IMEC Nederland A method and device for estimating a condition of a person
CN108042145A (zh) * 2017-11-28 2018-05-18 广州视源电子科技股份有限公司 情绪状态识别方法和系统、情绪状态识别设备
JP2020068973A (ja) 2018-10-30 2020-05-07 クラリオン株式会社 感情推定統合装置、感情推定統合方法およびプログラム
KR102737426B1 (ko) * 2019-10-22 2024-12-02 현대자동차주식회사 오류 모니터링을 이용한 운전자 숙련용 주행 모델 생성 장치 및 방법
JP7635779B2 (ja) 2020-03-31 2025-02-26 ソニーグループ株式会社 学習システム及びデータ収集装置

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017199597A1 (ja) 2016-05-20 2017-11-23 ソニー株式会社 生体情報処理装置、生体情報処理方法、及び情報処理装置
WO2018218286A1 (en) * 2017-05-29 2018-12-06 Saltor Pty Ltd Method and system for abnormality detection
JP2019170967A (ja) 2018-03-29 2019-10-10 パナソニックＩｐマネジメント株式会社 生体信号検知装置及び生体信号検知方法
JP2020048622A (ja) * 2018-09-24 2020-04-02 トヨタ自動車株式会社 生体状態推定装置
JP2020203051A (ja) * 2019-06-19 2020-12-24 株式会社プロアシスト コンピュータプログラム、情報処理装置、情報処理方法、学習済みモデルの生成方法及び学習済みモデル
JP2021053261A (ja) * 2019-10-01 2021-04-08 Ｎｅｃソリューションイノベータ株式会社 解析方法、解析装置、プログラム

Non-Patent Citations (7)

* Cited by examiner, † Cited by third party
Title
AFTANAS, L. I., S. A. GOLOCHEIKINE: "Human anterior and frontal midline theta and lower alpha reflect emotionally positive state and internalized attention: high-resolution EEG investigation of meditation.", NEUROSCIENCE LETTERS, vol. 310, no. 1, 2001, pages 57 - 60
BENEDEK, M.KAERNBACH, C.: "A continuous measure of phasic electrodermal activity", JOURNAL OF NEUROSCIENCE METHODS, vol. 190, 2010, pages 80 - 91, XP027106841
KHATWANIMOHIT ET AL.: "Energy efficient convolutional neural networks for eeg artifact detection", IEEE BIOMEDICAL CIRCUITS AND SYSTEMS CONFERENCE (BIOCAS, 2018
LAWHERNVERNONW. DAVID HAIRSTONKAY ROBBINS: "DETECT: A MATLAB toolbox for event detection and identification in time series, with applications to artifact detection in the EEG signals", PLOS ONE, vol. 8.4, 2013, pages 62944
SALAHUDDINLIZAWATI ET AL., ULTRA SHORT TERM ANALYSIS OF HEART RATE VARIABILITY FOR MONITORING MENTAL STRESS IN MOBILE SETTINGS, 2007
See also references of EP4371483A4
VAL-CALVOMIKEL ET AL.: "Optimization of real-time EEG artifact removal and emotion estimation for human-robot interaction applications", FRONTIERS IN COMPUTATIONAL NEUROSCIENCE, vol. 13, 2019, Retrieved from the Internet <URL:https://www.frontiersin.org/articles/10.3389/fncom.201>

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
NL2037018A (en) * 2023-02-16 2024-08-29 Milbotix Ltd Monitoring a person using machine learning
WO2025211253A1 (ja) * 2024-04-05 2025-10-09 ソニーグループ株式会社 情報処理方法、情報処理装置、および学習方法

Also Published As

Publication number Publication date
KR20240033217A (ko) 2024-03-12
EP4371483A1 (en) 2024-05-22
CN117615709A (zh) 2024-02-27
JPWO2023286313A1 (https=) 2023-01-19
JP7794199B2 (ja) 2026-01-06
EP4371483A4 (en) 2024-10-30
US20240306967A1 (en) 2024-09-19

Similar Documents

Publication Publication Date Title
CN112955973B (zh) 基于深度神经网络来辨别精神行为属性的电子设备
Bernal et al. Galea: A physiological sensing system for behavioral research in Virtual Environments
Sharma et al. Objective measures, sensors and computational techniques for stress recognition and classification: A survey
Gravina et al. Automatic methods for the detection of accelerative cardiac defense response
US20150282768A1 (en) Physiological signal determination of bioimpedance signals
RU2602797C2 (ru) Способ и устройство измерения стресса
Tamura et al. Seamless healthcare monitoring
US20150230756A1 (en) Determining physiological characteristics from sensor signals including motion artifacts
US20260058007A1 (en) Real-time artifact processing and feature extraction method and system for electroencephalogram signal
JP7794199B2 (ja) 信号処理装置および方法
Supratak et al. Survey on feature extraction and applications of biosignals
Montanari et al. Earset: A multi-modal dataset for studying the impact of head and facial movements on in-ear ppg signals
Prabakaran et al. Review on the wearable health-care monitoring system with robust motion artifacts reduction techniques
WO2021084488A1 (en) Smartglasses for detecting physiological parameters
Yue et al. Heart rate and heart rate variability as classification features for mental fatigue using short-term PPG signals via smartphones instead of ECG recordings
Traunmuller et al. Wearable healthcare devices for monitoring stress and attention level in workplace environments
Singh et al. Physiological Signal Processing and Analysis for Mental Stress Assessment: A Critical Review
Bousefsaf et al. Remote assessment of physiological parameters by non-contact technologies to quantify and detect mental stress states
Kraft et al. Reliability factor for accurate remote PPG systems
Davis-Stewart Stress detection: Stress detection framework for mission-critical application: Addressing cybersecurity analysts using facial expression recognition
Gupta et al. Overview of the effect of physiological stress on different biological signals
Chen et al. A wearable physiological detection system to monitor blink from faint motion artifacts by machine learning method
von Rosenberg et al. Smart helmet: monitoring brain, cardiac and respiratory activity
Karmuse et al. A review on real time heart rate monitoring system using usb camera
Poli Measurement and processing of multimodal physiological signals in response to external stimuli by wearable devices and evaluation of parameters influencing data acquisition

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22841663

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 2023535093

Country of ref document: JP

WWE Wipo information: entry into national phase

Ref document number: 202280048229.5

Country of ref document: CN

WWE Wipo information: entry into national phase

Ref document number: 2022841663

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE

ENP Entry into the national phase

Ref document number: 2022841663

Country of ref document: EP

Effective date: 20240215

WWW Wipo information: withdrawn in national office

Ref document number: 1020237044736

Country of ref document: KR

WWW Wipo information: withdrawn in national office

Ref document number: 2022841663

Country of ref document: EP