CN102715902A - Emotion monitoring method for special people - Google Patents

Abstract

An emotional monitoring method for special populations: collecting neurophysiological information, collecting EEG signals and cerebral blood oxygen signals; performing signal processing, including signal preprocessing, feature extraction and emotion recognition; judging the type of intervention, and judging the recognized emotions , if the user’s emotion is in a calm state or a positive state, return to the initial stage to continue monitoring, otherwise enter the next stage; carry out system intervention, and determine whether to use the system’s emotional auxiliary regulation intervention method or remote intervention method according to different emotional states; store data, draw The emotional periodic table for a period of time records and stores the user's emotional fluctuations. Through the portable detection of EEG and cerebral blood oxygen signals, the present invention can carry out accurate emotional state identification, help users and the outside world detect bad emotions of users in time, and carry out humanized soothing measures, which can achieve the purpose of improving the emotional state of users, thereby avoiding The user's mental state deteriorates, and at the same time, it also reduces the burden of care and psychological pressure on the family members of patients with emotional disorders.

Description

Emotional monitoring methods for special populations

technical field

The invention relates to emotional state monitoring and regulation in daily life and medical health. In particular, it involves an emotional monitoring method for special populations that can provide doctors and patients with more comprehensive health information.

Background technique

With the development of society, people of all ages and fields have more and more mental distress, and the incidence of various emotional-related diseases is increasing, such as depression, mania, anxiety, obsessive-compulsive disorder, affective disorder, etc. . There are also some special occasions, such as astronauts in space for a long time in a claustrophobic environment, air crew, elderly people living alone, etc., some negative emotions will inevitably appear due to environmental factors. The continuous accumulation of negative emotions interferes with the work and life of this group of people, sometimes resulting in a decrease in work efficiency, an increase in operational errors, and even extreme behavior in severe cases. For example, some depressed patients are pessimistic and world-weary, leading to suicidal behavior. Therefore, when these people are alone, it is necessary to carry out emotional monitoring and early warning. When they have negative emotions, timely detection and psychological intervention or adjustment can help them soothe their emotions, dispel pessimistic thoughts, and reduce the possibility of tragedy. Sexuality, while reducing caregiving stress for family members of patients with emotional disorders.

Existing research shows that people's emotional state is measurable. There are many methods for emotion recognition research. The earliest research focused on recognizing human emotions through facial expressions, voice, and gesture signals. However, due to the ease of control and camouflage of the signals, the results cannot rule out the influence of subjective factors of the subjects. , sometimes the underlying, true emotional state cannot be observed. Subsequent emotion research introduced physiological signals of the autonomic nervous system such as electrocardiogram, blood volume pulsation, and breathing. Although these signals are objective in emotion recognition, they reflect more the level of emotional arousal. The accuracy of different types of emotion recognition still needs to be improved. Emotion is the result of the coordinated activity of the cerebral cortex and subcortical neural processes. Neurophysiological signals such as EEG and cerebral blood oxygen are closely related to people's emotional state and are spontaneous. Therefore, emotion recognition technology based on neurophysiological signals is safe and reliable. At the same time, it can also provide a theoretical basis for the neural feedback regulation of emotions, which has attracted extensive attention of researchers and made some progress in recent years. Current research has been able to distinguish between extreme negative emotions and calm states. Emotional feedback regulation based on neurophysiological information (such as EEG feedback regulation) has become an adjuvant treatment for clinical depression, anxiety and other emotional disorders.

In 2010, Yisi Liu, Olga Sourina and others in Singapore used the EEG characteristics of the subjects in different emotional states as input signals to drive 3D expression animations to express the subjects' emotional feelings in real time. The prefrontal cortex is a brain area closely related to emotional activities. Numerous studies on emotions and cerebral blood oxygen parameters have shown that under negative emotions, the bilateral cerebral blood flow in the prefrontal cortex of volunteers will decrease, while the content of oxyhemoglobin will increase. When the volunteers experienced a very strong unpleasant emotion, 3 to 4 seconds after the emotion started, there was a significant increase in blood flow to a specific area of the brain. Comparative experiments showed that pleasurable emotions can reduce blood flow to another part of the brain. Emotion detection technology based on cerebral blood oxygen signals has been used by NASA to monitor the emotions of astronauts.

In recent years, brainwave detection technology has developed rapidly. Emotiv systems in San Francisco, California, USA has produced the Epoc headset, a helmet that reads brainwaves. Brainwave scanning technology with 14 sensors can accurately read brainwave activity. Japan has launched a fashion toy: Necomimi Cat Ears. This product reads the brain waves through the TGAM EEG chip and analyzes them. It converts the EEG signals representing the emotional state of the person into digital signals that can be recognized by the "cat ears" to perform corresponding actions. commands to complete different actions to simply express the wearer's emotional response. For example, when the wearer is in a state of surprise, the cat ears will stand up high, and when the wearer is relaxed or depressed, they will shrug and pull down.

Physical and mental health is real health, and the monitoring of both physical and mental states deserves our attention. Portable physiological monitors have been used in clinical and home care for a long time, but portable emotional monitoring technology for special populations has not yet appeared on the market. The rapid development of theory and technology makes it possible to develop portable emotion monitoring technology.

Contents of the invention

The technical problem to be solved by the present invention is to provide a portable detection of EEG and cerebral blood oxygen signals, to carry out accurate emotional state recognition, to help the user and the outside world detect the user's bad mood in time, and to take humanized comfort measures, which can reach Emotional monitoring methods for special groups of people who improve the emotional state of users.

The technical solution adopted in the present invention is: a method for emotional monitoring of special groups of people, including the following stages:

(1) collecting neurophysiological information,

The neurophysiological information includes collecting EEG signals and cerebral blood oxygen signals;

(2) Perform signal processing, including: signal preprocessing, feature extraction and emotion recognition;

(3) Judging the type of intervention and judging the recognized emotion. If the user’s emotion is in a calm state or a positive state, return to the (1) stage to continue monitoring, otherwise enter the next stage;

(4) Carry out systematic intervention, and determine whether to use systematic emotional auxiliary regulation intervention or remote intervention according to different emotional states;

(5) Store data, draw a periodic table of emotions for a period of time, record and store the user's emotional fluctuations.

The signal preprocessing described in stage (2) is to filter noise and enhance the signal, which specifically includes the following steps:

(a) Adaptive filtering is used to suppress the power frequency interference mixed in the signal, and wavelet analysis or independent component analysis is used to filter out the interference of oculoelectricity and myoelectricity;

(b) On the basis of the amplification of the signal by the amplifier, some algorithms are used to enhance the EEG signal related to emotion, and the spatial distance between different types of signals is increased by using the co-space model filter according to the target state, thereby improving the signal-to-noise ratio .

The feature extraction described in stage (2) specifically includes the following steps:

(a) Analysis of time-frequency information: using short-time Fourier transform and Chirplet transform time-frequency transformation methods to observe the waveform characteristics and frequency domain energy characteristics of EEG signals and cerebral blood oxygen signals at different times and frequencies;

(b) Analysis of non-linear features, namely fractal dimension, correlation dimension and information entropy;

(c) From the perspective of phase analysis, the Hilbert transform is used to extract the phase information of the signal, and the synchronization characteristics of signals in different frequency bands are extracted.

The emotion recognition described in stage (2) specifically includes the following steps:

(a) The features extracted from the EEG signal and the cerebral blood oxygen signal are optimally selected and fused, and the feature optimization is to optimize the features by using a support vector machine-based greedy algorithm or a genetic algorithm or a recursive feature screening algorithm;

(b) Learn and identify the user's emotional state according to the optimal combination of features, and accurately classify the category and arousal of the emotion. Learning and identification uses artificial neural network or Bayesian linear classifier or support vector machine as integrated learning learner.

In stage (4), when the user’s negative emotion is in a state of low arousal, the system’s emotional auxiliary regulation intervention method is activated; when the user’s negative emotion is in a high arousal state, the system’s emotional auxiliary regulation intervention method and The remote intervention method, the remote intervention method is to use GPRS wireless communication to send the user's emotional state to the family's mobile phone and the psychiatrist monitoring center server in real time, and the family and the doctor will communicate with the user on-line according to the user's current state.

The emotional monitoring method for special groups of people in the present invention is to carry out accurate emotional state identification through portable detection of EEG and cerebral blood oxygen signals, help the user and the outside world to detect the user's bad mood in time, and carry out humanized comfort measures, which can reach The purpose of improving the user's emotional state is to avoid the deterioration of the user's mental state, and at the same time reduce the burden of care and psychological pressure on the family members of patients with emotional disorders. Secondly, the present invention can also be used for the spiritual comfort of victims after a major disaster, the spiritual guardianship of the elderly who are alone in an aging society, and the like. If this invention is combined with physiological monitoring technology, a "physical and mental health monitor" can be invented, which can provide doctors and patients with more comprehensive health information, help hospitals provide patients with humanized medical services, and reduce the complications of depression after serious illness Rate. The technology is objective, safe and credible, and has potential market and promotion value. Moreover, an emotional monitor can be developed on the basis of the present invention, which can properly intervene in the user's negative emotions and at the same time assist the user in emotional regulation, avoid the continuous accumulation of the user's negative state, and reduce various negative emotions caused by emotional factors. Influence.

Description of drawings

Figure 1 is a flow chart of the method of the present invention.

Detailed ways

The emotional monitoring method for special groups of people of the present invention will be described in detail below in conjunction with the embodiments and accompanying drawings.

The emotional monitoring method for special groups of the present invention combines brain wave detection and brain function imaging technology for emotional recognition, and on this basis, a portable emotional monitoring method that can be used for daily emotional state detection is invented. Initiate interventions. After the present invention is popularized and used, it can provide personalized care for special groups of people, help patients obtain psychological comfort from the outside world and relieve emotions, improve mental state, reduce the incidence of excessive behavior or work mistakes of special groups of people, and reduce emotional diseases (such as depression) incidence rate.

The emotional monitoring method for special populations of the present invention, as shown in Figure 1, includes the following stages:

(1) Collect neurophysiological information

The non-invasive dry electrode technology is used for EEG collection. The shape of the collection device is similar to a standard stereo wireless headset. The electrodes on the forearm of the headset are placed on the forehead to collect EEG signals. The headband of the earphone is embedded with a near-infrared spectrum emission and collection probe to wirelessly collect the cerebral blood oxygen signal on the forehead of the human brain, and the collected EEG signal and cerebral blood oxygen signal are sent to the information processing unit respectively.

(2) Perform signal processing, including signal preprocessing, feature extraction and emotion recognition, among which:

1) Signal preprocessing:

The collected two types of neurophysiological signals are purified after A/D conversion to improve the signal-to-noise ratio and facilitate feature extraction. The present invention intends to passivate the signal from the two aspects of signal noise filtering and signal enhancement, which specifically includes the following steps:

(a) Adaptive filtering is used to suppress deterministic noise signals such as power frequency interference mixed in the signal, and wavelet analysis (Wavelet Analysis, WA) or independent component analysis (Independent Component Analysis, ICA) is used to filter out electrooculogram, muscle electrical interference;

(b) On the basis of the amplification of the signal by the amplifier, some algorithms are used to enhance the EEG signals related to emotions. The Common Spatial Patterns (CSP) filter can increase the spatial distance between different types of signals according to the target state , thereby improving the signal-to-noise ratio. The present invention attempts to use this algorithm to enhance the target signal.

2) Feature extraction:

After the signal preprocessing, the present invention intends to comprehensively analyze the emotional signal from three different angles, extract features, and specifically include the following steps:

(a) Analysis of time-frequency information: Using short-time Fourier transform and Chirplet transform time-frequency transformation methods, respectively observe the waveform characteristics and frequency domain energy characteristics of EEG signals and cerebral blood oxygen signals at different times and frequencies. For example: the power spectrum energy ratio of each frequency band, the asymmetry of fast and slow waves, the quantitative and time-varying characteristics of oxygen and hemoglobin, deoxygenated hemoglobin and total blood flow, etc.;

(b) Analyzing nonlinear features, namely nonlinear parameters such as fractal dimension, correlation dimension and information entropy. These messages often contain a certain pattern, such as emotional types;

(c) From the perspective of phase analysis, using the Hilbert transform to extract the phase information of the signal, the synchronization characteristics of signals in different frequency bands can be extracted.

3) Emotion recognition

After optimizing and selecting the features extracted from various signals, they are fused, and then the user's emotional state is identified according to the optimal feature combination, and the emotional valence (positive, negative, and calm) and arousal (high, low) are accurately measured. Grading, such as whether the current state is high sadness or low disgust.

Emotion recognition of the present invention specifically comprises the following steps:

(a) Fusing the extracted features of EEG signals and cerebral blood oxygen signals after optimal selection, the feature optimization is to optimize the features by using SVM-based greedy algorithm or genetic algorithm or recursive feature selection;

(b) Learn and identify the user's emotional state according to the optimal combination of features, and accurately grade the category and intensity of the emotion. Learning and identification use artificial neural network or Bayesian linear classifier or support vector machine as integrated learning learner.

The information processing part can be made into a low-power ASIC intelligent integrated chip, which can simplify and speed up the signal processing and analysis process, realize real-time online analysis of the signal, reduce product research and application costs, and make this technology easy to promote and use. Information preprocessing and feature extraction and recognition technologies are guided by a large number of reliable cutting-edge research results, so as to be applicable to different groups of people and environments.

(3) Determine the type of intervention

Determine whether the user’s emotional state needs to start intervention. If the user’s emotional state is calm or positive, no intervention is required, and return to stage (1) to continue monitoring; if it is in a negative state, it will enter the next stage for different degrees of intervention ;

(4) System intervention

The systemic intervention described includes the following two intervention methods:

(a) Systematic Emotion Assistance Regulatory Intervention: This part is achieved by activating the regulatory music library and "psychological tips" that match various negative emotional states. The "psychological tips" can intelligently play professional and friendly emotional adjustment prompts, such as "sit down quietly and breathe deeply with the music..." Gently remind the user of the current state, and help the user adjust or talk to vent .

(b) Remote intervention method: The GPRS wireless module is used to send the user's emotional state to the family's mobile phone and the psychiatrist's monitoring center server in real time. The family and the doctor can talk with the user on the line according to the user's current state, listen to their feelings and carry out interventions. Spiritual comfort, help users empathize, and block their bad emotions in time.

When the user's negative emotion is in a state of low arousal, start the first intervention method above, that is, the system's emotional auxiliary regulation intervention method; when the user's negative emotion is in a state of high arousal, start the first intervention method and the first intervention method 2 Intervention methods, that is, systematic emotional auxiliary regulation intervention methods and remote intervention methods.

(5) Store data

Draw an emotional cycle table for a period of time (short-term such as one day, long-term such as a week), record and store the user's emotional fluctuations, help users understand their own emotional changes, and provide a basis for the user's future psychological counseling or evaluation of the efficacy of psychological rehabilitation. The storage unit is equipped with an industrial standard serial UART input and output interface for reading and clearing information.

Claims (5)

1. a kind of emotional monitoring method of special crowd, is characterized in that, comprises following stage: (1) collecting neurophysiological information, The neurophysiological information includes collecting EEG signals and cerebral blood oxygen signals; (2) Perform signal processing, including: signal preprocessing, feature extraction and emotion recognition; (3) Judging the type of intervention and judging the recognized emotion. If the user’s emotion is in a calm state or a positive state, return to the (1) stage to continue monitoring, otherwise enter the next stage; (4) Carry out systematic intervention, and determine whether to use systematic emotional auxiliary regulation intervention or remote intervention according to different emotional states; (5) Store data, draw a periodic table of emotions for a period of time, record and store the user's emotional fluctuations. 2. The emotional monitoring method for special groups of people according to claim 1, characterized in that the signal preprocessing in stage (2) is to filter and enhance the signal, which specifically includes the following steps: (a) Adaptive filtering is used to suppress the power frequency interference mixed in the signal, and wavelet analysis or independent component analysis is used to filter out the interference of oculoelectricity and myoelectricity; (b) On the basis of the amplification of the signal by the amplifier, some algorithms are used to enhance the EEG signal related to emotion, and the spatial distance between different types of signals is increased by using the co-space model filter according to the target state, thereby improving the signal-to-noise ratio . 3. The emotional monitoring method for special groups of people according to claim 1, characterized in that the feature extraction described in stage (2) specifically includes the following steps: (a) Analysis of time-frequency information: using short-time Fourier transform and Chirplet transform time-frequency transformation methods to observe the waveform characteristics and frequency domain energy characteristics of EEG signals and cerebral blood oxygen signals at different times and frequencies; (b) Analysis of non-linear features, namely fractal dimension, correlation dimension and information entropy; (c) From the perspective of phase analysis, the Hilbert transform is used to extract the phase information of the signal, and the synchronization characteristics of signals in different frequency bands are extracted. 4. The emotional monitoring method for special groups of people according to claim 1, characterized in that the emotional recognition in stage (2) specifically includes the following steps: (a) The features extracted from the EEG signal and the cerebral blood oxygen signal are optimally selected and fused, and the feature optimization is to optimize the features by using a support vector machine-based greedy algorithm or a genetic algorithm or a recursive feature screening algorithm; (b) Learn and identify the user's emotional state according to the optimal combination of features, and accurately classify the category and arousal of the emotion. Learning and identification uses artificial neural network or Bayesian linear classifier or support vector machine as integrated learning learner. 5. The emotional monitoring method for special groups of people according to claim 1, characterized in that, in stage (4), when the user's negative emotion is in a state of low arousal, the system's emotional auxiliary regulation intervention mode is started; when the user When the negative emotions are in a state of high arousal, the system’s emotional auxiliary regulation intervention mode and remote intervention mode are started at the same time. The remote intervention mode is to use GPRS wireless communication to send the user’s emotional state to the family’s mobile phone and the psychologist in real time. The monitoring center server, family members and doctors communicate with the user on-line according to the user's current status.

Images