A thinking force real-time evaluation intervention method based on electroencephalogram signals

Abstract

This invention discloses a real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals, relating to the field of brain-computer interface technology. The method includes the following steps: acquiring multi-channel raw EEG signals from the anterior cortex using a multi-channel EEG data acquisition device; segmenting the EEG signals by time-sliding windows and extracting the projections of each channel onto the θ, α, β, and γ frequency bands using multi-scale fluctuation analysis and wavelet packet transform as frequency band signals; calculating the time-varying sample entropy channel-by-channel and frequency band-by-frequency band to generate a single-channel multi-frequency band entropy vector. This invention introduces Mahalanobis distance to quantify the offset between the current cognitive feature vector and the fatigue manifold center, which can offset assessment biases caused by individual physiological baseline differences and intraday physiological fluctuations, significantly improving the robustness and anti-interference ability of the Thinking Ability Index (TFI) calculation. It can accurately distinguish between three types of transient thinking states: normal focus, inattention, and cognitive overload.

Description

Technical Field

[0001] This invention relates to the field of brain-computer interface technology, and more specifically, to a method for real-time assessment and intervention of thinking ability based on electroencephalogram (EEG) signals. Background Technology

[0002] In modern society, the increasing intensity of mental activity in work and study settings leads to a continuous depletion and fluctuation in individual concentration, short-term memory, and multi-task logical coordination abilities (collectively defined as thinking ability). Electroencephalography (EEG), a non-invasive neural signal acquisition technology with millisecond-level temporal resolution, can directly capture the firing activity of clusters of neurons in the cerebral cortex. It possesses natural technological advantages in the fields of cognitive function quantitative assessment and neural state monitoring, and is currently widely used in cognitive neuroscience, brain-computer interface-related research, and product development.

[0003] Existing cognitive assessment and intervention systems based on EEG signals have inherent and unavoidable flaws. On the one hand, the assessment indicators are singular, relying solely on simple linear indicators such as the theta / β power spectrum ratio for judgment, ignoring the dynamic collaborative network characteristics of different brain regions, and failing to fully characterize the true cognitive load and brain region collaboration level behind thought activities. On the other hand, the algorithms have weak dynamic adaptation capabilities, generally employing large-size time sliding window statistical operations, resulting in high signal processing latency and an inability to capture short-term fluctuations such as momentary inattention and cognitive overload in real time. Furthermore, there is no adaptive intervention system; conventional audio-visual stimulation and fixed-paradigm neurofeedback training cannot dynamically adjust the stimulation band and intensity according to real-time quantitative thinking indices, and long-term use can easily lead to brain fatigue or neural stimulation tolerance. Summary of the Invention

[0004] In response to the problems in related technologies, this invention proposes a real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals to overcome the aforementioned technical problems existing in the current related technologies.

[0005] The technical solution of this invention is implemented as follows:

[0006] A real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals includes the following steps:

[0007] Multi-channel raw EEG signals from the anterior cortex were acquired in advance using a multi-channel EEG data acquisition device.

[0008] The EEG signal was segmented by a time sliding window, and multi-scale wave analysis and wavelet packet transform were used to extract the projections of each channel onto the θ, α, β, and γ frequency bands as frequency band signals. The time-varying sample entropy was calculated channel by channel and frequency band by frequency band. Generate a single-channel multi-band entropy vector;

[0009] Based on the decomposed EEG signals of each frequency band, Hilbert transform was used to extract the instantaneous phase of each channel. Calculate the phase-locked value of any channel i and j in frequency band b for each pair. The corresponding frequency band brain connectivity topology matrix is ​​generated by pairwise PLV values ​​of all channels. ;

[0010] Constructing the integrated feature vector Z t And calculate Z t With respect to the standard fatigue manifold center Z fatigue Mahalanobis distance Get real-time thinking ability index ;

[0011] Based on the output With the preset target thinking ability value Calculate state deviation The variable parameter adaptive PID controller solves the intervention control output in real time.

[0012] The multi-channel raw EEG signal includes a clean reconstructed EEG sequence vector waveform obtained by removing electrooculography (EOG), electromyography (EMG), and power frequency noise through an adaptive artifact reduction algorithm. Where C represents the number of acquisition channels and t represents the real-time sampling time point.

[0013] Specifically, the calculation of the phase-locked values ​​of arbitrary channels i and j in frequency band b is described. The network used to characterize the collaborative response network of various functional areas of the cerebral cortex is represented as:

[0014] ;

[0015] In the formula, For the corresponding frequency band, Let N be the instantaneous phase of the Hilbert transform of signal i in frequency band b. win This represents the number of signal sampling points within the sliding window.

[0016] Wherein, the calculation of Z t With respect to the standard fatigue manifold center Z fatigue Mahalanobis distance , is represented as:

[0017] .

[0018] Among them, the acquisition of real-time thinking ability index , is represented as:

[0019] ;

[0020] In the formula, Calibrate weighting coefficients for offline cognitive experiments; This indicates finding the largest eigenvalue of a matrix.

[0021] Wherein, the calculation of state deviation , is represented as:

[0022] .

[0023] The solution for the intervention control output quantity is expressed as:

[0024] .

[0025] This also includes adjusting the corresponding sound, light, and electrical intervention devices according to the current instantaneous U(t).

[0026] The beneficial effects of this invention are:

[0027] 1. Based on the DMNI-FMD dynamic multi-source neural information fusion and manifold transformation algorithm, this invention achieves multi-dimensional deep fusion of extracted temporal nonlinear sample entropy features and constructed brain network phase-locked spatial topological features, overcoming the limitations of traditional schemes that rely solely on single-band power ratios. Simultaneously, by introducing Mahalanobis distance to quantify the offset between the current cognitive feature vector and the fatigue manifold center, it can offset the assessment bias caused by individual physiological baseline differences and intraday physiological fluctuations, significantly improving the robustness and anti-interference ability of the thinking ability index (TFI) calculation, and accurately distinguishing between three types of instantaneous thinking states: normal focus, inattention, and cognitive overload.

[0028] 2. This invention uses a variable-parameter PID controller to dynamically output intervention control quantities based on the real-time thinking ability index. The intensity and rhythm of sound and light stimulation and transcranial electrical stimulation can be adjusted as needed. When a decline in thinking ability is detected, targeted neural stimulation with matching frequency bands is automatically applied. After the thinking state returns to the optimal range, the stimulation output is automatically reduced, avoiding brain fatigue and nerve tolerance caused by fixed stimulation patterns. This achieves real-time adaptive cognitive intervention with no manual intervention and low latency, effectively improving the regulation effect of thinking arousal and sustained focus. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 This is a flowchart illustrating a real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the scope of protection of the present invention.

[0032] According to an embodiment of the present invention, a real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals is provided.

[0033] like Figure 1 As shown, the real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to an embodiment of the present invention includes the following steps:

[0034] Step S1: First, use a multi-channel EEG data acquisition device to acquire multi-channel raw EEG signals from the anterior cortex at a sampling rate fs. Then, use an adaptive artifact reduction algorithm to remove electrooculography (EOG), electromyography (EMG), and power line noise to obtain a clean reconstructed EEG sequence vector waveform. Where C represents the number of acquisition channels and t represents the real-time sampling time point.

[0035] Specifically, in the implementation of this technical solution, the subject wears a dry electrode head-mounted EEG helmet with 8 EEG electrode channels. The electrode positions are arranged according to the international 10-20 system, placed at Fp1, Fp2, F3, F4, C3, C4, O1, and O2 respectively to acquire multi-channel raw EEG signals from the anterior cortex. At the same time, the baseline drift caused by the subject's blinking or blinking is eliminated by the Artificial Artifact Subspace Reconstruction (ASR) algorithm.

[0036] Step S2: The filtered EEG signal is segmented by a time sliding window. Multi-scale fluctuation analysis and wavelet packet transform are used to extract the projections of each channel in the frequency bands θ (4–8Hz), α (8–12Hz), β (12–30Hz), and γ (30–45Hz) as frequency band signals. The time-varying sample entropy is calculated channel by channel and frequency band by frequency band. Generate a single-channel multi-band entropy vector, that is, the entropy value of each band of any channel c is represented as... ;

[0037] Specifically, a fixed-time sliding window is set for the reconstructed EEG sequence output in step S1, and the window width W is calibrated. a The possible values ​​are 2s and the sliding step size is W. s The value is 200ms, and the four frequency bands θ, α, β and γ are obtained by wavelet packet transform and multi-scale wave analysis decomposition.

[0038] Step S3: Based on the EEG signals after frequency band decomposition in Step S2, Hilbert transform is used to extract the instantaneous phase of each channel. Calculate the phase-locked value of any channel i and j in frequency band b for each pair. The corresponding frequency band brain connectivity topology matrix is ​​generated by pairwise PLV values ​​of all channels. ;

[0039] Among them, the phase-locking values ​​of any two channels i and j in each specific frequency band are calculated in real time. The network used to characterize the collaborative response network of various functional areas of the cerebral cortex is represented as:

[0040] ;

[0041] In the formula, For the corresponding frequency band, Let N be the instantaneous phase of the Hilbert transform of signal i in frequency band b. win This represents the number of signal sampling points within the sliding window.

[0042] Step S4: Based on the multi-channel frequency band sample entropy mean characteristics output in step S2 and the maximum eigenvalue of the high-frequency β / γ brain network topology matrix output in step S3... Construct the integrated feature vector Z t ; Calculate Z t With respect to the standard fatigue manifold center Z fatigue Mahalanobis distance Calculate the real-time thinking ability index ;

[0043] Among them, the computational thinking ability index, which integrates temporal information entropy and spatial correlation matrix, maps multi-dimensional and high-dimensional features to a cognitive topological manifold, and is expressed as:

[0044] ;

[0045] In the formula, Calibrate weighting coefficients for offline cognitive experiments; This indicates finding the largest eigenvalue of a matrix; Z represents the current state feature vector t Distance from the standard fatigue manifold center Z fatigue The Mahalanobis distance is expressed as:

[0046] ;

[0047] Step S5, based on the output of step S4 With the preset target thinking ability value Calculate state deviation The intervention control output is solved in real time by a variable parameter adaptive PID controller, and is expressed as:

[0048] ;

[0049] Specifically, based on the current instantaneous U(t), the corresponding sound, light, and electrical intervention devices are adjusted, such as binaural difference frequency sound wave rhythm, spatial modulation visual flicker, or weak transcranial electrical stimulation parameters.

[0050] Among them, the state deviation is calculated. Represented as:

[0051] ;

[0052] Specifically, in the implementation of this technical solution, the intervention device is a binaural difference-frequency sound wave generating device. Determine if attention is scattered, set goals It outputs dual-channel audio at 250Hz and 265Hz to form a 15Hz β band difference frequency stimulus; when When the sound pressure level rises back to the optimal range of 70-80, the PID controller automatically reduces the intensity of the sound pressure level stimulation. The intervention device is replaced with a spatially modulated visual flickering device or a weak transcranial electrical stimulation device, with the control output U(t) modulating the duty cycle of the light flickering and the amplitude of the transcranial electrical stimulation current, respectively.

[0053] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Those skilled in the art, upon considering the disclosure in the specification and embodiments, will readily conceive of other embodiments of this disclosure. This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and embodiments are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the claims.

[0054] It should be understood that this disclosure is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this disclosure is limited only by the appended claims.

Claims

1. A real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals, characterized in that, Includes the following steps: Multi-channel raw EEG signals from the anterior cortex were acquired in advance using a multi-channel EEG data acquisition device. The EEG signal was segmented by a time sliding window, and multi-scale wave analysis and wavelet packet transform were used to extract the projections of each channel onto the θ, α, β, and γ frequency bands as frequency band signals. The time-varying sample entropy was calculated channel by channel and frequency band by frequency band. Generate a single-channel multi-band entropy vector; Based on the decomposed EEG signals of each frequency band, Hilbert transform was used to extract the instantaneous phase of each channel. Calculate the phase-locked value of any channel i and j in frequency band b for each pair. The corresponding frequency band brain connectivity topology matrix is ​​generated by pairwise PLV values ​​of all channels. ; Constructing the integrated feature vector Z t And calculate Z t With respect to the standard fatigue manifold center Z fatigue Mahalanobis distance Get real-time thinking ability index ; Based on the output With the preset target thinking ability value Calculate state deviation The variable parameter adaptive PID controller solves the intervention control output in real time.

2. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The multi-channel raw EEG signals, including those processed by an adaptive artifact reduction algorithm to remove electrooculography (EOG), electromyography (EMG), and power line noise, yield a clean reconstructed EEG sequence vector waveform. Where C represents the number of acquisition channels and t represents the real-time sampling time point.

3. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The calculation of the phase-locked value of arbitrary channels i and j in frequency band b. The network used to characterize the collaborative response of various functional areas of the cerebral cortex is represented as: ； In the formula, For the corresponding frequency band, Let N be the instantaneous phase of the Hilbert transform of signal i in frequency band b. win This represents the number of signal sampling points within the sliding window.

4. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The calculation of Z t With respect to the standard fatigue manifold center Z fatigue Mahalanobis distance , is represented as: 。 5. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 4, characterized in that, The acquisition of real-time thinking ability index , is represented as: ； In the formula, Calibrate weighting coefficients for offline cognitive experiments; This indicates finding the largest eigenvalue of a matrix.

6. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 1, characterized in that, The calculation of state deviation , is represented as: 。 7. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 6, characterized in that, The solution for the intervention control output is expressed as: 。 8. The real-time assessment and intervention method for thinking ability based on electroencephalogram (EEG) signals according to claim 1, characterized in that, Also includes: Adjust the corresponding sound, light, and electrical intervention devices according to the current instantaneous U(t).

Images