Eye fatigue detection method and system for brain wave glasses

Abstract

The invention discloses an eyestrain detection method and system for brain wave glasses. The method comprises the following steps: collecting a biological original signal directly related to eyestrain; based on the physiological relevance of the biological original signal, performing joint denoising on the biological original signal to obtain an effective signal capable of reflecting the eyestrain state; extracting feature parameters strongly related to eyestrain from the effective signals, learning contribution degrees of features to fatigue through a feature weight distribution mechanism, and generating weighted feature vectors of focusing eyestrain; and analyzing the weighted feature vector by using a pre-trained multi-modal fusion model in combination with user individual physiological baseline calibration, and outputting an eyestrain level adapted to individual differences.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent devices, and relates to an eye fatigue detection method and system of electroencephalogram glasses. BACKGROUND

[0002] With the popularity of electronic devices, visual fatigue caused by long-time eye use has become an important problem affecting the vision health of the public. Visual fatigue not only causes discomfort symptoms such as dry eyes and eye swelling, but also may cause myopia deepening, dry eye and other eye diseases after long-term accumulation. Therefore, real-time and accurate monitoring of eye fatigue state is of great significance for prevention and intervention. As a daily wearing tool directly contacting the eyes, glasses are widely explored for use in life scenes. However, traditional smart glasses only have basic optical correction function or integrate simple sensors (such as accelerometers and ambient light sensors), and do not integrate a miniature camera in the frame to collect eye dynamic signals in real time. Existing eye fatigue detection methods rely on subjective questionnaire or scale evaluation, and the results are easily affected by user subjective perception bias, which is insufficient in accuracy. Or, only the length of time when the glasses are used is simply judged to determine whether the eyes are tired and need to rest. The above methods cannot more accurately detect eye fatigue. SUMMARY

[0003] The application provides an eye fatigue detection method and system of electroencephalogram glasses, which achieves the purpose of accurately detecting eye fatigue state by adapting to individual differences through synchronous acquisition and joint denoising of multi-modal biological signals, weighted extraction of fatigue strongly related features and analysis of personalized multi-modal fusion model.

[0004] In order to achieve the above purpose, the application adopts the following technical solutions: An eye fatigue detection method of electroencephalogram glasses, comprising the following steps: S1. Collecting biological raw signals directly related to eye fatigue, wherein the biological raw signals include electroencephalogram signals, eye movement and pupil signals, and heart rate variability signals; S2. Based on the physiological correlation of the biological raw signals, jointly denoising the biological raw signals to obtain effective signals reflecting the eye fatigue state; S3. Extracting feature parameters strongly related to eye fatigue from the effective signals, learning the contribution of each feature to fatigue through a feature weight distribution mechanism, and generating a weighted feature vector focusing on eye fatigue; S4. Using a pre-trained multi-modal fusion model, combining user individual physiological baseline calibration, analyzing the weighted feature vector, and outputting an eye fatigue level adapting to individual differences.

[0005] Preferably, the electroencephalogram signals are collected by a flexible dry electrode on the temple, covering the occipital lobe and temporal lobe regions; The eye movement and pupil signals are obtained by taking eye images with a miniature infrared camera in the frame; The heart rate variability signal is obtained by collecting the pulse wave behind the ear through the PPG sensor on the temple; The EEG signal, the eye movement and pupil signal, and the heart rate variability signal are synchronously sampled through the same clock source to ensure time stamp alignment.

[0006] Preferably, the method of performing joint denoising on the original biological signal based on the physiological relevance of the original biological signal to obtain an effective signal capable of reflecting the state of eye fatigue comprises the following steps: S21. Detect physiological events such as rapid blinking, gaze drift, and head movement using eye movement and pupil signals, and synchronously mark the corresponding time windows in the EEG and heart rate variability signals affected by these events. S22. De-noising is performed based on the interference characteristics of each modal signal, combined with a marked interference time window: For EEG signals, independent component analysis is used to separate and remove myoelectric artifacts, while also correcting or removing signal segments affected by oculoscopic artifacts. For eye movement and pupil signals, filtering is used to eliminate ambient light interference, and image registration algorithms are used to correct for image offsets caused by head movement. For heart rate variability signals, bandpass filtering is used to remove motion artifacts and smooth out abnormal heartbeat intervals. S23. Compare the correlation of the multimodal signals before and after denoising. If the deviation exceeds the preset threshold, re-execute steps S21 to S22 until the requirements are met, and finally obtain an effective signal that can accurately reflect the state of eye fatigue.

[0007] Preferably, extracting characteristic parameters strongly related to eye fatigue from the effective signal, learning the contribution of each feature to fatigue through a feature weight distribution mechanism, and generating a weighted characteristic vector of focusing eye fatigue includes the following steps: S31. Extracting characteristic parameters directly related to eye fatigue from the effective signals, wherein for EEG signals, extracting frequency domain features reflecting the degree of visual cortical fatigue; for eye movement and pupil signals, extracting time series features reflecting the fatigue state of eye muscles and pupil accommodation function; and for heart rate variability signals, extracting frequency domain features reflecting functional disorders of the autonomic nervous system caused by fatigue; S32. Learning the contribution of each feature to eye fatigue through a feature weight allocation mechanism, wherein the weight of each feature is dynamically adjusted according to the strength of the correlation between each feature and eye fatigue; S33. Multiply each feature by its corresponding weight and then concatenate them to generate a weighted feature vector of focusing eye fatigue.

[0008] Preferably, the characteristic parameters strongly related to eye fatigue include: The ratio of the power spectral density of the alpha wave to the beta wave of the brain electrical signal, and the ratio of the power spectral density of the theta wave to the beta wave; The blink frequency, pupil diameter change rate, and gaze point drift amplitude of the eye movement and pupil signal; The low-to-high frequency power ratio and total power of the heart rate variability signal.

[0009] Preferably, the pre-trained multi-modal fusion model is used in combination with individual physiological baseline calibration to analyze the weighted feature vector and output an eye fatigue level that adapts to individual differences, including the following steps: S41. The generated weighted feature vector of focused eye fatigue is input into the input layer of the multi-modal fusion model as initial data for model analysis; S42. Individual physiological baseline calibration includes initial calibration and online learning. During initial calibration, the user collects multi-modal signals in a normal eye use state for the first time, extracts features, and trains an initial model to establish the user's individual physiological baseline. During the online learning stage, multi-modal signals are continuously collected and features are extracted during the user's daily use, and model parameters are dynamically updated through incremental learning algorithms to adapt the model to changes in the user's individual eye use habits. S43. The calibrated multi-modal fusion model learns the correlation between multi-modal features through the interaction layer, performs deep analysis on the input weighted feature vector, and mines the potential correlation between features to evaluate the fatigue state; S44. The multi-modal fusion model classifies the analysis results through the output layer and finally outputs an eye fatigue level that adapts to individual differences.

[0010] Preferably, the following steps are also included: S5. Trigger intervention feedback according to the output fatigue level, wherein: Mild fatigue is prompted to rest through a frame vibration motor; Moderate fatigue is prompted through a bone conduction earphone; Severe fatigue synchronously pushes an alarm to a bound mobile phone APP and generates an eye fatigue trend report.

[0011] Preferably, the eye fatigue trend report includes: Fatigue time distribution records the specific time period and frequency of fatigue occurrence within a day / week; Fatigue intensity analysis calculates the duration ratio and change trend of mild, moderate, and severe fatigue; Correlation factor analysis identifies eye use behaviors that are strongly related to fatigue based on multi-modal signal features; Personalized recommendations are generated based on fatigue trends and correlation factors.

[0012] A system for detecting eye fatigue when using electroencephalogram glasses, for executing the method for detecting eye fatigue when using electroencephalogram glasses, comprising: Bio-signal acquisition module: used to collect original biological signals directly related to eye fatigue, including EEG signals, eye movement and pupil signals, and heart rate variability signals; A joint denoising module is used to perform joint denoising on the original biological signals based on their physiological relevance, thereby obtaining an effective signal that can reflect the state of eye fatigue. A feature extraction and weighting module is used to extract feature parameters that are strongly related to eye fatigue from the effective signal, learn the contribution of each feature to fatigue through a feature weight allocation mechanism, and generate a weighted feature vector for focusing eye fatigue; The fatigue assessment module is used to analyze the weighted feature vector using a pre-trained multimodal fusion model in combination with the user's individual physiological baseline calibration, and output an eye fatigue level adapted to individual differences.

[0013] Beneficial effects of the present invention: By collecting EEG, eye movement, pupil, and heart rate variability signals, the present invention breaks through the limitations of traditional single-modality detection that is susceptible to individual differences or environmental interference. It can capture collaborative fatigue patterns that a single signal cannot reflect, and significantly improves detection accuracy. The joint denoising mechanism based on physiological correlation effectively eliminates noise such as electromyographic artifacts, ambient stray light, and motion interference through cross-modal event marking and sub-modal denoising processing, ensuring signal quality. The feature extraction and dynamic weight allocation mechanism focuses on the core features of fatigue, and dynamically adjusts the weight according to the contribution of the features to fatigue, thereby enhancing the pertinence of the model input; personalized model training combined with initial calibration and online learning adapts to the physiological baselines and changes in eye habits of different users, avoiding the problem of general models not being universal. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is a schematic flow chart of the steps of the method of the present invention.

[0015] Figure 2 It is a schematic flow chart of the steps of the method for obtaining an effective signal of the present invention.

[0016] Figure 3 It is a schematic flow chart of the steps of the method for generating weighted feature vectors of the present invention.

[0017] Figure 4 It is a schematic flow chart of the steps of the method for outputting the eye fatigue level of the present invention. DETAILED DESCRIPTION

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meanings as commonly understood by those skilled in the art to which this invention belongs. The terms used in the specification of the application are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the accompanying drawings are intended to cover non-exclusive inclusions. The terms "first" and "second" in the specification and claims of the present invention and the accompanying drawings are used to distinguish different objects, not to describe a specific order.

[0019] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0021] The present invention provides Figures 1 to 4 In the embodiment of the present invention, a method and system for detecting eye fatigue using EEG glasses achieves the purpose of accurately and adaptively detecting eye fatigue status by synchronously collecting and jointly denoising multimodal biosignals, weighted extraction of fatigue-related features, and personalized multimodal fusion model analysis. Specific embodiment 1 A method for detecting eye fatigue when using electroencephalogram glasses comprises the following steps: S1. Collecting original biological signals directly related to eye fatigue, including EEG signals, eye movement and pupil signals, and heart rate variability signals; Specifically, when the user wears the EEG glasses, the device synchronously collects biological raw signals directly related to eye fatigue through multimodal sensors: the flexible dry electrodes built into the temples fit the temporal lobe, occipital lobe and other areas of the head to collect EEG signals reflecting visual cortical activity in real time; the miniature infrared camera at the front of the frame captures eye images, eyelid movements (such as blinking) and pupil dynamics (such as contraction / expansion) to obtain eye movement and pupil signals; the photoplethysmography (PPG) sensor at the end of the temple collects pulse waves behind the ears, calculates the variability of adjacent heartbeat intervals to obtain heart rate variability (HRV) signals; the three-modal signals are synchronously sampled through a unified clock source to ensure timestamp alignment, providing a time-related data basis for subsequent analysis.

[0022] Furthermore, the EEG glasses also include behavioral data collection and physiological signal collection. Behavioral data collection includes lighting environment (ambient light sensor), sitting habits (six-axis gyroscope), wearing time (timer), infrared distance (infrared sensor detects screen distance), outdoor activities (GPS positioning), and environmental noise (microphone sampling rate 16kHz). Physiological signal collection includes EEG signals. Four flexible dry electrodes (covering the occipital lobe and temporal lobe) are distributed on the temples of the glasses. Concentration, fatigue and emotional values, eye movement and pupil signals (mini infrared camera), and heart rate variability signals (PPG sensor) are collected at a certain frequency sampling rate. All data are uploaded to the cloud in real time via wireless network (WiFi / 4G). Behavioral data is compressed by minutes, and physiological signals are streamed.

[0023] S2. Based on the physiological relevance of the original biological signals, jointly denoise the original biological signals to obtain an effective signal that can reflect the state of eye fatigue; Specifically, based on the physiological relevance of multimodal signals, a joint denoising process is used to remove interference and retain effective information: first, physiological events such as rapid blinking and gaze drift are detected through eye movement and pupil signals, and the time windows in the EEG signals and HRV signals affected by the events are synchronously marked; then, based on the interference characteristics of each modal signal (such as myoelectric artifacts of EEG, ambient stray light of eye movement, motion interference of HRV), denoising processing is performed in combination with the marked time windows (such as independent component analysis of EEG, image filtering and alignment of eye movement, and bandpass filtering of HRV); finally, the correlation of multimodal signals before and after denoising is compared. If the deviation exceeds the preset threshold, the denoising parameters are iteratively optimized until an effective signal that can accurately reflect the state of eye fatigue is obtained.

[0024] S3 extracts characteristic parameters strongly related to eye fatigue from the effective signal, learns the contribution of each feature to fatigue through the feature weight distribution mechanism, and generates a weighted feature vector for focusing eye fatigue; Specifically, features that are strongly correlated with eye fatigue are extracted from the effective signal after denoising, and a weighted feature vector of focusing fatigue is generated through a weight distribution mechanism: the frequency domain features reflecting visual cortical fatigue are extracted from EEG signals (such as the power ratio of α waves to β waves, and theta waves to β waves); the temporal features reflecting eye muscle and pupil adjustment fatigue are extracted from eye movement and pupil signals (such as blinking frequency, pupil diameter change rate); the frequency domain features reflecting autonomic nervous system dysfunction are extracted from HRV signals (such as the power ratio of low frequency to high frequency); the contribution of each feature to fatigue is learned through a feature weight distribution mechanism (such as the weight of some features dynamically increases in the late stage of fatigue), and the weight of each feature is dynamically adjusted; finally, each feature is multiplied by its corresponding weight and spliced ​​to generate a weighted feature vector of focusing eye fatigue as the input for subsequent model analysis.

[0025] S4. Utilize the pre-trained multimodal fusion model, combined with the user's individual physiological baseline calibration, analyze the weighted feature vector, and output an eye fatigue level adapted to individual differences.

[0026] Specifically, the pre-trained multimodal fusion model is used in combination with the user's individual physiological baseline calibration to analyze the weighted feature vector and output the eye fatigue level that adapts to individual differences: the weighted feature vector is input into the input layer of the multimodal fusion model; when the user uses it for the first time, the multimodal signal of the normal eye state is collected, and the initial model is trained to establish the individual physiological baseline; during daily use, the model parameters are continuously updated through online learning to adapt to changes in the user's eye habits; the model learns the correlation between multimodal features through the interaction layer (such as the synergy between EEG and HRV features), and after classification processing, the fatigue level that adapts to individual differences (such as mild, moderate, and severe) is output.

[0027] This embodiment uses synchronous multimodal biosignal acquisition to cover multi-dimensional physiological information of the visual cortex, eye muscles, and autonomic nerves, thus solving the problem of single-modal detection being susceptible to interference. The joint denoising process is based on the correlation of physiological events, effectively eliminating environmental and motion interference and improving signal quality. The feature extraction and weighting mechanism focuses on the core characteristics of fatigue, and the dynamic weight allocation highlights key indicators, enhancing the pertinence of the model input. Individual physiological baseline calibration combines initial training with online learning to adapt to the physiological differences and habit changes of different users, avoiding the defect of "universal models are not universal." In summary, this embodiment achieves high-precision and personalized detection of eye fatigue, providing a reliable basis for real-time intervention. Specific embodiment 2 The EEG signals are collected through flexible dry electrodes on the temples, covering the occipital and temporal lobe areas; The eye movement and pupil signals are obtained by taking eye images with a miniature infrared camera in the frame; The heart rate variability signal is obtained by collecting the pulse wave behind the ear through the PPG sensor on the temple; The EEG signal, the eye movement and pupil signal, and the heart rate variability signal are synchronously sampled through the same clock source to ensure time stamp alignment.

[0029] Specifically, three flexible dry electrodes are positioned on the inner side of the temples. The electrode array is distributed along the temples, and when worn, they adhere to the user's bilateral temporal lobes (2 cm above the ears) and occipital lobes (1 cm in front of the posterior hairline). The electrodes are connected to the signal processing module via low-noise wires, which collect EEG signals (raw amplitude range 1-100 μV) at a 250 Hz sampling rate. The flexible dry electrodes require no conductive gel and adapt to the skin contours, ensuring comfort for extended wear.

[0030] Two miniature infrared cameras are embedded in the front of the frame (on either side of the nose bridge), aimed at the user's eyeballs. These cameras capture eye images using near-infrared reflection technology, capturing real-time eyelid movement (blinking) and pupil contraction / dilation. The output includes raw signals including blink frequency, pupil diameter, and gaze point coordinates.

[0031] A PPG sensor is integrated on the inner side of the temple end, which fits the skin behind the ear (the depression between the tragus and the mastoid process) when worn. The sensor collects the pulse wave signal behind the ear at a sampling rate of 100Hz, calculates the intervals between adjacent heart beats (RR intervals) through photoelectric capacitance analysis, and generates the original heart rate variability (HRV) signal.

[0032] The acquisition modules for EEG, eye movement and pupil signals, and HRV signals are all connected to the same clock source (a high-precision crystal oscillator with a frequency error of <0.01%). Each sensor acquires an initial timestamp synchronously upon startup, and the timestamp is calibrated using the clock source every 100ms during the sampling process, ensuring a time alignment error of ≤1ms for the three modal signals. Specific embodiment three The method of performing joint denoising on the biological original signal based on the physiological relevance of the biological original signal to obtain an effective signal that can reflect the state of eye fatigue includes the following steps: S21. Detect physiological events such as rapid blinking, gaze drift, and head movement using eye movement and pupil signals, and synchronously mark the corresponding time windows in the EEG and heart rate variability signals affected by these events. Specifically, the system detects physiological events such as rapid blinking, gaze drift, and head movement of the user through eye movement and pupil signals, and synchronously marks the time windows affected by these events in the electroencephalogram (EEG) and heart rate variability (HRV) signals. Rapid blink event detection: analyzing the blink duration of eye movement and pupil signals (calculating the time from eyelid closure to eyelid opening using eye images captured by an infrared camera). When a blink duration < 0.2s is detected, it is marked as a "rapid blink" event, and the start timestamp (t_start) and end timestamp (t_end) of the event are synchronously recorded. Gaze drift event detection: The gaze point coordinate offset (Δx, Δy) between two adjacent eye image frames is calculated using an image registration algorithm. When Δx or Δy exceeds 5 pixels (corresponding to an actual gaze offset > 2°), it is marked as a "gaze point drift" event. The time window is [t_start - 0.3s, t_end + 0.3s]; Head motion event detection: Combined with the built-in accelerometer in the temple (not the core of this claim but auxiliary verification), when the acceleration change rate is greater than 0.5g, it is marked as a "head motion" event, and the time window is [t_start-0.5s, t_end+0.5s]; Synchronous marking, mapping the time windows of the above events to the corresponding data segments of the EEG signal (sampling rate 250Hz) and HRV signal (sampling rate 100Hz). For example, the time window of the rapid eye blink event is [10.2s, 10.4s]. Then the corresponding sampling points 2550-2600 in the EEG signal (250Hz×10.2s=2550, 250Hz×10.4s=2600) and the corresponding sampling points 1020-1040 in the HRV signal (100Hz×10.2s=1020, 100Hz×10.4s=1040) are marked as interference segments.

[0034] S22. De-noising is performed based on the interference characteristics of each modal signal, combined with a marked interference time window: For EEG signals, independent component analysis is used to separate and remove myoelectric artifacts, while also correcting or removing signal segments affected by oculoscopic artifacts. For eye movement and pupil signals, filtering is used to eliminate ambient light interference, and image registration algorithms are used to correct for image offsets caused by head movement. For heart rate variability signals, bandpass filtering is used to remove motion artifacts and smooth out abnormal heartbeat intervals. Specifically, based on the interference characteristics of each modal signal, denoising is performed in combination with the marked time window: EEG signal denoising: Independent component analysis (ICA) is used to separate artifact components from EEG signals within the marked interference time window: 1. Input the original EEG signal into the ICA algorithm (based on FastICA implementation) and decompose it into multiple independent components (ICs); 2. Identify ICs corresponding to electromyographic artifacts (EMG, high-frequency components) and electrooculographic artifacts (EOG, strongly associated with blink events) by manual or automatic classification (e.g., based on a kurtosis threshold >3); 3. After removing the aforementioned artifact ICs, the EEG signal is reconstructed and the signal of the unlabeled interference segment is bandpass filtered (0.5-30 Hz) to remove DC drift and high-frequency noise; Eye movement and pupil signal denoising: 1. Ambient light interference processing: Perform median filtering (3×3 window) on the eye image collected by the infrared camera to eliminate random noise points; 2. Head motion offset correction: Using an image registration algorithm (based on SIFT feature matching), the current frame image is aligned with the previous frame reference image to correct image offset caused by head motion (maximum allowable offset is 10 pixels). 3. Eliminate abnormal data: If the pupil diameter change rate is >50% / s (outside the normal physiological range), mark the frame data as abnormal and interpolate to replace it (using the average of the previous and next frames); HRV signal denoising: 1. Motion artifact removal: Bandpass filter the PPG signal (0.5-3Hz) to retain the heartbeat-related frequency components; 2. Abnormal heartbeat interval correction: Calculate the difference between adjacent RR intervals (ΔRR). If ΔRR > 200 ms (exceeding the normal variability range), smooth the abnormal point by linear interpolation (taking the average of the two normal RR intervals before and after).

[0035] S23. Compare the correlation of the multimodal signals before and after denoising. If the deviation exceeds the preset threshold, re-execute steps S21 to S22 until the requirements are met, and finally obtain an effective signal that can accurately reflect the state of eye fatigue.

[0036] Specifically, the correlation of multimodal signals before and after denoising is compared to verify the denoising effect: The correlation calculation is to calculate the Pearson correlation coefficient (ρ) between the original signal before denoising and the effective signal after denoising. For EEG signals, the ρ value of the α wave power spectral density (PSD) is used, for eye movement signals, the ρ value of the blink frequency is used, and for HRV signals, the ρ value of the LF / HF ratio is used. For threshold judgment, the correlation threshold is preset to ρ ≥ 0.8 (retaining the main physiological information). If the ρ value of any modality signal is < 0.8, steps S21-S22 are re-executed (for example, adjusting the artifact recognition threshold of ICA or the number of feature matches for image registration); When the ρ values ​​of all modal signals are ≥0.8, the iteration is terminated and an effective signal that can accurately reflect the state of eye fatigue is output. Specific embodiment 4 The method of extracting feature parameters strongly related to eye fatigue from the effective signal, learning the contribution of each feature to fatigue through a feature weight distribution mechanism, and generating a weighted feature vector of focused eye fatigue includes the following steps: S31. Extracting characteristic parameters directly related to eye fatigue from the effective signals, wherein for EEG signals, extracting frequency domain features reflecting the degree of visual cortical fatigue; for eye movement and pupil signals, extracting time series features reflecting the fatigue state of eye muscles and pupil accommodation function; and for heart rate variability signals, extracting frequency domain features reflecting functional disorders of the autonomic nervous system caused by fatigue; Specifically, based on the effective signal after joint denoising, feature parameters directly related to eye fatigue are extracted from the three modal signals of EEG, eye movement and pupil, and heart rate variability (HRV). The specific steps include: EEG signal characteristics (reflecting visual cortex fatigue): Perform frequency domain analysis on the denoised EEG signal (sampling rate 250Hz, covering the occipital and temporal lobe areas): 1. Calculate the power spectral density (PSD) using the Welch algorithm (Hamming window, 2-second window length, 1-second overlap) and extract the PSD values ​​of α waves (8-13 Hz), θ waves (4-7 Hz), and β waves (13-30 Hz). 2. Calculate the α / β ratio (α wave PSD / β wave PSD) and the θ / β ratio (θ wave PSD / β wave PSD) as core frequency domain features reflecting the degree of visual cortical fatigue (fatigue is accompanied by an increase in theta waves and a decrease in the beta waves, resulting in an increase in the ratio).

[0038] Eye movement and pupil signal characteristics (reflecting eye muscle and pupil adjustment fatigue): Perform timing analysis on the denoised eye movement image sequence (30fps): 1. Blinking frequency: Count the number of blinks in 1 minute (blinks / minute). Blinking frequency decreases when fatigue occurs (normal: 15-20 blinks / minute, while fatigued: <10 blinks / minute). 2. Pupil diameter change rate: Calculate the change in pupil diameter over five consecutive frames (ΔD = D(t+1) - D(t)) and take the average of the absolute values. When the pupil is fatigued, its ability to accommodate decreases, and the change rate decreases. 3. Gaze point drift amplitude: Calculate the maximum offset (in pixels) of the gaze point coordinates (x, y) within 10 seconds. Eye fatigue weakens eye muscle control and increases the drift amplitude (normal <5 pixels, fatigue >10 pixels).

[0039] HRV signal characteristics (reflecting autonomic nervous system dysfunction): Perform frequency domain analysis on the denoised HRV signal (RR interval sequence): 1. Use fast Fourier transform (FFT) to calculate the power spectrum of low frequency (LF, 0.04-0.15Hz) and high frequency (HF, 0.15-0.4Hz); 2. Calculate the LF / HF ratio (low-frequency power / high-frequency power) and total power (LF + HF). Sympathetic nerve activation during fatigue increases the LF / HF ratio (normal: 1.5-2.5; >3.0 during fatigue).

[0040] S32. Learning the contribution of each feature to eye fatigue through a feature weight allocation mechanism, wherein the weight of each feature is dynamically adjusted according to the strength of the correlation between each feature and eye fatigue; Specifically, the contribution of each feature to eye fatigue is learned through a feature weight distribution mechanism, which is specifically implemented using a dynamic attention mechanism. The following steps are included: Among them, the initial weight setting is based on physiological knowledge (such as θ / β ratio weight 0.3, blink frequency weight 0.2, LF / HF ratio weight 0.2); Among them, the dynamic adjustment rule is: through the online learning algorithm (such as stochastic gradient descent, learning rate 0.01), the weight is dynamically updated according to the correlation between the current feature and the fatigue label: 1. Collect the Pearson correlation coefficient (ρ) between each feature and fatigue level in the historical data. For example, the θ / β ratio and severe fatigue have a ρ = 0.75, and the blink frequency and severe fatigue have a ρ = -0.68; 2. Weight update formula: W_i(t+1) = W_i(t) + η*(ρ_i - avg(ρ)), where η is the learning rate and avg(ρ) is the average correlation coefficient of all features; 3. Constrain the weight range (0≤W_i≤1) to avoid excessive weight of a single feature (for example, the maximum weight of the θ / β ratio does not exceed 0.4).

[0041] S33. Multiply each feature by its corresponding weight and then concatenate them to generate a weighted feature vector of focusing eye fatigue.

[0042] Specifically, each feature is multiplied by its corresponding dynamic weight and then concatenated to generate a weighted feature vector of focusing eye fatigue, which specifically includes the following steps: Feature-weight mapping: Assume that the current features and their weights are: θ / β ratio (0.35), α / β ratio (0.25), blink frequency (0.20), pupil diameter change rate (0.15), LF / HF ratio (0.25); Weighted calculation: multiply each eigenvalue by the corresponding weight. For example, if the measured value of the θ / β ratio is 1.8, the weighted value is 1.8×0.35=0.63; Vector concatenation: The weighted eigenvalues ​​are concatenated in a fixed order (e.g., [θ / β weighted value, α / β weighted value, blink frequency weighted value, pupil diameter change rate weighted value, LF / HF weighted value]) into a 5-dimensional weighted eigenvector [0.63, 0.50, 1.60, 0.30, 0.75], which serves as the input for the subsequent multimodal fusion model. Specific embodiment five The characteristic parameters strongly related to eye fatigue include: The power spectral density ratio of alpha waves to beta waves, and the power spectral density ratio of theta waves to beta waves of the EEG signal; Specifically, electroencephalogram (EEG) signals are collected via flexible dry electrodes on the temples (covering the occipital and temporal lobe regions). After joint denoising, the following frequency domain features are extracted to reflect the degree of visual cortical fatigue. The power spectral density ratio of alpha waves to beta waves (α / β ratio) is calculated by performing frequency domain analysis on the denoised 250Hz EEG signal and calculating the power spectral density (PSD) using the Welch algorithm with a Hamming window (window length 2s, overlap 1s). The alpha wave is 8-13Hz (corresponding to the rhythm of the visual cortex's relaxed state) and the beta wave is 13-30Hz (corresponding to high-frequency activity related to alertness). During normal eye use, the power of the alpha and beta waves is balanced (the α / β ratio is approximately 0.8-1. 2) When the eyes are tired, the excitability of the visual cortex decreases, resulting in a decrease in the power of beta waves, a relative increase in alpha waves, and an increase in the alpha / beta ratio (>1.5); the calculation method of the power spectral density ratio of theta waves to beta waves (theta / beta ratio) is the same as above, and the PSD value of theta waves (4-7Hz) is extracted. Theta waves are closely related to fatigue-related slow wave activity. When tired, the power of theta waves increases significantly (reflecting the inhibition of neural activity), while the power of beta waves continues to weaken, resulting in an increase in the theta / beta ratio (normal <0.5, fatigue >0.8).

[0044] Eye movement and pupil signals include blink frequency, pupil diameter change rate, and gaze point drift amplitude; Specifically, eye movement and pupil signals are collected by a miniature infrared camera (30fps) on the frame and after denoising, the following time series features are extracted to reflect the fatigue state of the eye muscles and pupil accommodation function: Blinking frequency (times / minute) is calculated by counting the number of complete cycles of eyelid closure and opening recognized by infrared images within 1 minute. It is 15-20 times / minute during normal eye use (to maintain corneal moisture). When fatigued, it is reduced to <10 times / minute due to eye muscle weakness, which can easily cause dry eyes and sore eyes; Pupil diameter change rate (% / s) is the absolute change in pupil diameter within 5 consecutive frames (about 0.17s) compared with the initial diameter. The average value of the diameter ratio. When the eyes are used normally, the pupil automatically adjusts with the light or the target of gaze (the rate of change is about 5-10% / s). When fatigued, the adjustment ability is reduced due to the weakening of the pupillary sphincter and dilator muscles, and the rate of change is reduced to <3% / s; the gaze point drift amplitude (pixel) is determined by the image registration algorithm after the gaze point position in each frame, and the maximum offset of the gaze point coordinates (x, y) within 10 seconds is calculated. When the eyes are used normally, the eye muscles stably control the gaze point (drift amplitude <5 pixels). When fatigued, the eye muscles' control ability is reduced, and the gaze point is prone to involuntary drift (amplitude >10 pixels), which manifests as "seeing double images".

[0045] The ratio of low-frequency to high-frequency power and the total power of the heart rate variability signal.

[0046] Specifically, the heart rate variability (HRV) signal was collected by the temple PPG sensor (100 Hz) and after denoising, the following frequency domain features were extracted to reflect the functional disorder of the autonomic nervous system caused by fatigue: the low-frequency to high-frequency power ratio (LF / HF ratio) was calculated by performing fast Fourier transform (FFT) on the RR interval series to calculate the low-frequency (LF, 0.04-0.15 Hz, reflecting sympathetic nerve activity) and high-frequency (HF, The LF / HF ratio is 1.5-2.5 (sympathetic-parasympathetic balance) during normal eye use. During fatigue, the sympathetic nerves are continuously activated, resulting in an increase in LF power and a decrease in HF power, leading to an increase in the LF / HF ratio (>3.0). The total power (LF+HF) is the sum of the LF and HF powers, which normally reflects the overall level of heart rate variability (500-1500ms²). During fatigue, the autonomic nervous system regulation function is disturbed, resulting in a decrease in total power (<300ms²), indicating an accumulation of physiological stress. Specific embodiment six The method utilizes a pre-trained multimodal fusion model, combines it with the user's individual physiological baseline calibration, analyzes the weighted feature vector, and outputs an eye fatigue level adapted to individual differences, including the following steps: S41. The generated weighted feature vector of focusing eye fatigue is input into the input layer of the multimodal fusion model as the initial data for model analysis; Specifically, the weighted feature vectors for focusing eye fatigue generated in Example 4 (e.g., [θ / β weighting, blink frequency weighting, LF / HF weighting]) are input into the input layer of the multimodal fusion model. The input layer normalizes the feature values ​​(mapping them to the [0, 1] interval) to eliminate dimensional differences. For example, the θ / β weighting is 0.63 (original value 1.8 × weighting 0.35), which after normalization is 0.63 / 1.5 (maximum theoretical value) = 0.42, ensuring the stability of the model input data.

[0048] S42. Individual physiological baseline calibration includes initial calibration and online learning. During initial calibration, multimodal signals from the user's first use of the device are collected, features are extracted, and an initial model is trained to establish the user's individual physiological baseline. During the online learning phase, multimodal signals are continuously collected and features are extracted during daily use. Model parameters are dynamically updated using an incremental learning algorithm to adapt the model to changes in the user's individual eye habits. Specifically, the initial calibration phase (when the user wears the EEG glasses for the first time) requires a 25-minute calibration process: in the first 10 minutes, the user reads a paper book in a natural light environment (to avoid interference from blue light from electronic screens), and the device synchronously collects EEG, eye movement and HRV signals, extracting features such as the θ / β ratio (average 0.4), blinking frequency (18 times / minute), and LF / HF ratio (2.0) as "fatigue-free" baseline data; in the next 15 minutes, the user continuously uses a mobile phone (screen brightness 50%) to browse short videos, and the device collects features in the fatigue state (θ / β ratio rises to 0.9, blinking frequency drops to 10 times / minute, and LF / HF ratio rises to 3.2) as "fatigue" label data; then the fatigue-free (label 0) and fatigue (labels 1-3) data are input into a multimodal fusion model with randomly initialized initial parameters, and the model weights are optimized using the cross-entropy loss function (learning rate 0.001, 50 rounds of iterations), and finally an individual physiological baseline model of the user is established.

[0049] During the online learning phase (during daily user use), the model updates its parameters once an hour using an incremental learning algorithm suitable for small-batch data updates (such as FTRL). First, multimodal signals within the current 30 minutes are collected and features are extracted, marking the actual fatigue state (manually confirmed by the user through the app or automatically predicted by the model). Next, the deviation between the current feature and the baseline feature is calculated (for example, user A's baseline θ / β ratio is 0.4, the current measured value is 0.6, and the deviation is +0.2). Finally, the weights of the attention interaction layer are adjusted through gradient descent (for example, reducing the weight of the θ / β ratio by 0.05 and increasing the weight of blink frequency by 0.03), so that the model adapts to the individual user's physiological fluctuations (for example, when user B's baseline LF / HF ratio is high, the model automatically reduces the contribution of this feature to fatigue assessment).

[0050] S43. The calibrated multimodal fusion model learns the correlation between multimodal features through the interaction layer, performs in-depth analysis on the input weighted feature vector, and mines the potential correlation between features to assess fatigue status. Specifically, the calibrated multimodal fusion model learns the association between multimodal features through an interaction layer based on the Transformer self-attention mechanism: the interaction layer assigns query (Q), key (K), and value (V) vectors to each feature and calculates the attention score between the features (for example, the attention score of the θ / β ratio and the LF / HF ratio is 0.7, indicating a strong correlation between the two). The attention score is then weighted and summed with the feature value to generate a fusion feature (for example, fusion value = 0.7×θ / β weighted value + 0.2×blink frequency weighted value + 0.1×LF / HF weighted value), thereby mining collaborative fatigue patterns that cannot be reflected by single modal features (for example, when increased θ / β and increased LF / HF occur simultaneously, the fatigue risk increases by 2 times).

[0051] S44. The multimodal fusion model classifies the analysis results through the output layer and finally outputs the eye fatigue level that adapts to individual differences.

[0052] Specifically, after dimensionality reduction of the fused features by a fully connected layer (256 neurons), the softmax activation function outputs a four-category probability distribution (Level 0: No fatigue, Level 1: Mild, Level 2: Moderate, Level 3: Severe). The category with the highest probability is taken as the final fatigue level. For example, if a user's current fused feature value is 0.85 (threshold: Level 0 <0.3, Level 1 0.3-0.5, Level 2 0.5-0.8, Level 3 >0.8), the model output probabilities are [0.05, 0.15, 0.6, 0.2], and the final judgment is Level 2 (Moderate fatigue). Specific embodiment seven The following steps are also included: S5. Triggering intervention feedback based on the output fatigue level, wherein: In case of mild fatigue, the frame vibration motor will vibrate to remind you to take a rest; Moderate fatigue, voice prompts played through bone conduction headphones; In case of severe fatigue, an alert will be sent to the bound mobile phone APP and an eye fatigue trend report will be generated.

[0054] The eye fatigue trend report includes: Fatigue time distribution, recording the specific time period and frequency of fatigue occurrence within a day / week; Fatigue intensity analysis, statistics on the duration and changing trends of mild, moderate and severe fatigue; Correlation factor analysis, combining multimodal signal features, identifies eye behaviors that are strongly correlated with fatigue; Personalized suggestions generate targeted eye guidance based on fatigue trends and related factors.

[0055] Specifically, after the multimodal fusion model outputs the eye fatigue level (level 0: no fatigue, level 1: mild, level 2: moderate, level 3: severe), the system automatically triggers corresponding feedback according to the level threshold (intervention is started when the level is ≥1).

[0056] The triggering condition for mild fatigue feedback is a fatigue level of 1 (mild), which is manifested by a θ / β ratio of 0.5-0.8, a blinking frequency of 10-15 times / minute, and an LF / HF ratio of 2.5-3.0 (based on the value after individual baseline calibration of the user). The feedback method is that the built-in vibration motor of the frame (located at the connection between the temple and the frame, with an amplitude of 0.1mm) vibrates at a frequency of 50Hz for 1 second, and repeats every 30 seconds (to avoid excessive interference). For example, after a student has used the computer for 40 minutes continuously, the model outputs a fatigue level of 1, and the temple vibrates to prompt "Currently mild fatigue, it is recommended to rest after 5 minutes."

[0057] The triggering condition for moderate fatigue feedback is a fatigue level of Level 2 (moderate), which is manifested by a θ / β ratio of 0.8-1.2, a blinking frequency of 8-10 times / minute, and an LF / HF ratio of 3.0-3.5. The feedback method is that the bone conduction headphones integrated into the temples (frequency response range 200-8000Hz, output sound pressure level 60dB) play a preset voice prompt, which reads "Currently moderate fatigue, it is recommended to close your eyes and rest for 5 minutes, and massage the area around the eyes to relieve it." It is repeated every 2 minutes (to ensure user attention). If the user continuously browses the phone for 1 hour, the model determines that it is Level 2 fatigue, and the bone conduction headphones play a voice prompt. At the same time, the vibration motor synchronizes a short vibration (20Hz, 0.5 seconds) to enhance the reminder.

[0058] The triggering conditions for severe fatigue feedback are fatigue level 3 (severe), manifested as a θ / β ratio > 1.2, a blink frequency < 8 times / minute, an LF / HF ratio > 3.5, and a duration ≥ 10 minutes (to avoid occasional interference and misjudgment). The feedback method is to push notifications to the bound mobile phone APP via Bluetooth (BLE 5.0 protocol). The notification bar displays a red warning icon and the text "Warning: Currently severely fatigued, stop using your eyes immediately!" At the same time, the APP triggers a ringtone (maximum volume, lasting 10 seconds).

[0059] Severe fatigue feedback also includes the generation of eye fatigue trend reports. The system automatically summarizes fatigue data from the past 7 days and generates a report containing the following content, including: Fatigue time distribution: Mark the daily fatigue peak period (e.g. 20:00-22:00 is the peak); Fatigue intensity analysis: Count the duration of mild (30%), moderate (50%), and severe (20%) fatigue, as well as weekly trends (e.g., the duration of severe fatigue this week increased by 15% compared to last week); Correlation factor analysis: Combining multimodal features (e.g., the correlation coefficient between the θ / β ratio and continuous screen time was 0.82), we identified "continuous screen time >1 hour" as the main cause of fatigue; Personalized recommendations: Based on analysis results, specific guidance is generated, such as "take a 10-minute break every 40 minutes of using electronic devices" and "turn on warm light mode when studying at night."

[0060] Furthermore, the cloud backend receives behavioral data (including lighting environment, sitting habits, wearing time, infrared distance, outdoor activity trajectory and environmental noise) and physiological signals (EEG concentration / fatigue / emotional values, eye movement and pupil signals, heart rate variability signals) uploaded by the glasses, associates multi-source information through a timestamp alignment algorithm (for example, synchronously analyzing the correlation between sitting tilt period and EEG fatigue fluctuations), and generates a structured node report (output once every minute) based on the pre-trained multimodal fusion model. The report content includes: Fatigue level classification (mild / moderate / severe); concentration score (0-100) and emotional state (positive / neutral / negative); Behavioral-physiological correlation analysis conclusions (e.g., "Staying at a screen distance of less than 30cm for more than 10 minutes causes fatigue levels to rise to moderate levels"). Users can view real-time historical fatigue trend heat maps and concentration change curves through the mobile app or web backend, and receive personalized intervention recommendations (e.g., "Ambient noise > 65dB, recommend wearing noise-canceling headphones"). Specific embodiment eight A system for detecting eye fatigue when using electroencephalogram glasses, for executing the method for detecting eye fatigue when using electroencephalogram glasses, comprising: Bio-signal acquisition module: used to collect original biological signals directly related to eye fatigue, including EEG signals, eye movement and pupil signals, and heart rate variability signals; A joint denoising module is used to perform joint denoising on the original biological signals based on their physiological relevance, thereby obtaining an effective signal that can reflect the state of eye fatigue. A feature extraction and weighting module is used to extract feature parameters that are strongly related to eye fatigue from the effective signal, learn the contribution of each feature to fatigue through a feature weight allocation mechanism, and generate a weighted feature vector for focusing eye fatigue; The fatigue assessment module is used to analyze the weighted feature vector using a pre-trained multimodal fusion model in combination with the user's individual physiological baseline calibration, and output an eye fatigue level adapted to individual differences.

[0062] The invention is operational with numerous general purpose or special purpose computer system environments or configurations.

[0063] For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics devices, network PCs, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, etc.

[0064] The invention may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer.

[0065] Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The invention may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network.

[0066] In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

[0067] Specifically, those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing related hardware via computer-readable instructions. The computer-readable instructions can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes in the above-described method embodiments. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0068] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps and they may be performed in other orders.

[0069] Moreover, at least part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. Their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0070] Obviously, the embodiments described above are only some of the embodiments of the present invention, rather than all of them. The accompanying drawings provide preferred embodiments of the present invention, but do not limit the scope of the present invention. The present invention can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to facilitate a more thorough and comprehensive understanding of the disclosure of the present invention.

[0071] Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art may still modify the technical solutions described in the aforementioned specific embodiments or replace some of the technical features therein with equivalents. Any equivalent structure made using the contents of the present invention's description and drawings, directly or indirectly applied to other related technical fields, shall also fall within the scope of protection of the present invention.

Claims

1. A method for detecting eye fatigue when using electroencephalogram glasses, characterized in that: The following steps are involved: S1. Collecting original biological signals directly related to eye fatigue, including EEG signals, eye movement and pupil signals, and heart rate variability signals; S2. Based on the physiological relevance of the original biological signals, jointly denoise the original biological signals to obtain an effective signal that can reflect the state of eye fatigue; S3 extracts characteristic parameters strongly related to eye fatigue from the effective signal, learns the contribution of each feature to fatigue through the feature weight distribution mechanism, and generates a weighted feature vector for focusing eye fatigue; S4. Utilize the pre-trained multimodal fusion model, combined with the user's individual physiological baseline calibration, analyze the weighted feature vector, and output an eye fatigue level adapted to individual differences.

2. The method for detecting eye fatigue using electroencephalogram glasses according to claim 1, characterized in that: The EEG signals are collected through flexible dry electrodes on the temples, covering the occipital and temporal lobe areas; The eye movement and pupil signals are obtained by taking eye images with a miniature infrared camera in the frame; The heart rate variability signal is obtained by collecting the pulse wave behind the ear through the PPG sensor on the temple; The EEG signal, the eye movement and pupil signal, and the heart rate variability signal are synchronously sampled through the same clock source to ensure time stamp alignment.

3. The method for detecting eye fatigue using electroencephalogram glasses according to claim 2, characterized in that: The method of performing joint denoising on the biological original signal based on the physiological relevance of the biological original signal to obtain an effective signal that can reflect the state of eye fatigue includes the following steps: S21. Detect physiological events such as rapid blinking, gaze drift, and head movement using eye movement and pupil signals, and synchronously mark the corresponding time windows in the EEG and heart rate variability signals affected by these events. S22. De-noising is performed based on the interference characteristics of each modal signal, combined with a marked interference time window: For EEG signals, independent component analysis is used to separate and remove myoelectric artifacts, while also correcting or removing signal segments affected by oculoscopic artifacts. For eye movement and pupil signals, filtering is used to eliminate ambient light interference, and image registration algorithms are used to correct for image offsets caused by head movement. For heart rate variability signals, bandpass filtering is used to remove motion artifacts and smooth out abnormal heartbeat intervals. S23. Compare the correlation of the multimodal signals before and after denoising. If the deviation exceeds the preset threshold, re-execute steps S21 to S22 until the requirements are met, and finally obtain an effective signal that can accurately reflect the state of eye fatigue.

4. The method for detecting eye fatigue using electroencephalogram glasses according to claim 1, characterized in that: The method of extracting feature parameters strongly related to eye fatigue from the effective signal, learning the contribution of each feature to fatigue through a feature weight distribution mechanism, and generating a weighted feature vector of focused eye fatigue includes the following steps: S31. Extracting characteristic parameters directly related to eye fatigue from the effective signals, wherein for EEG signals, extracting frequency domain features reflecting the degree of visual cortical fatigue; for eye movement and pupil signals, extracting time series features reflecting the fatigue state of eye muscles and pupil accommodation function; and for heart rate variability signals, extracting frequency domain features reflecting functional disorders of the autonomic nervous system caused by fatigue; S32. Learning the contribution of each feature to eye fatigue through a feature weight allocation mechanism, wherein the weight of each feature is dynamically adjusted according to the strength of the correlation between each feature and eye fatigue; S33. Multiply each feature by its corresponding weight and then concatenate them to generate a weighted feature vector of focusing eye fatigue.

5. The method for detecting eye fatigue using electroencephalogram glasses according to claim 3, characterized in that: The characteristic parameters strongly related to eye fatigue include: The power spectral density ratio of alpha waves to beta waves, and the power spectral density ratio of theta waves to beta waves of the EEG signal; Eye movement and pupil signals include blink frequency, pupil diameter change rate, and gaze point drift amplitude; The ratio of low-frequency to high-frequency power and the total power of the heart rate variability signal.

6. The method for detecting eye fatigue using electroencephalogram glasses according to claim 5, characterized in that: The method utilizes a pre-trained multimodal fusion model, combines it with the user's individual physiological baseline calibration, analyzes the weighted feature vector, and outputs an eye fatigue level adapted to individual differences, including the following steps: S41. The generated weighted feature vector of focusing eye fatigue is input into the input layer of the multimodal fusion model as the initial data for model analysis; S42. Individual physiological baseline calibration includes initial calibration and online learning. During initial calibration, multimodal signals from the user's first use of the device are collected, features are extracted, and an initial model is trained to establish the user's individual physiological baseline. During the online learning phase, multimodal signals are continuously collected and features are extracted during daily use. Model parameters are dynamically updated using an incremental learning algorithm to adapt the model to changes in the user's individual eye habits. S43. The calibrated multimodal fusion model learns the correlation between multimodal features through the interaction layer, performs in-depth analysis on the input weighted feature vector, and mines the potential correlation between features to assess fatigue status. S44. The multimodal fusion model classifies the analysis results through the output layer and finally outputs the eye fatigue level that adapts to individual differences.

7. The method for detecting eye fatigue using electroencephalogram glasses according to claim 6, characterized in that: The following steps are also included: S5. Triggering intervention feedback based on the output fatigue level, wherein: In case of mild fatigue, the frame vibration motor will vibrate to remind you to take a rest; Moderate fatigue, voice prompts played through bone conduction headphones; In case of severe fatigue, an alert will be sent to the bound mobile phone APP and an eye fatigue trend report will be generated.

8. The method for detecting eye fatigue using electroencephalogram glasses according to claim 7, characterized in that: The eye fatigue trend report includes: Fatigue time distribution, recording the specific time period and frequency of fatigue occurrence within a day / week; Fatigue intensity analysis, statistics on the duration and changing trends of mild, moderate and severe fatigue; Correlation factor analysis, combining multimodal signal features, identifies eye behaviors that are strongly correlated with fatigue; Personalized suggestions generate targeted eye guidance based on fatigue trends and related factors.

9. A system for detecting eye fatigue when using electroencephalogram glasses, used to execute the method for detecting eye fatigue when using electroencephalogram glasses according to claims 1 to 8, characterized in that: include: Bio-signal acquisition module: used to collect original biological signals directly related to eye fatigue, including EEG signals, eye movement and pupil signals, and heart rate variability signals; A joint denoising module is used to perform joint denoising on the original biological signals based on their physiological relevance, thereby obtaining an effective signal that can reflect the state of eye fatigue. A feature extraction and weighting module is used to extract feature parameters that are strongly related to eye fatigue from the effective signal, learn the contribution of each feature to fatigue through a feature weight allocation mechanism, and generate a weighted feature vector for focusing eye fatigue; The fatigue assessment module is used to analyze the weighted feature vector using a pre-trained multimodal fusion model in combination with the user's individual physiological baseline calibration, and output an eye fatigue level adapted to individual differences.

Images