A mental state monitoring method, device and computer readable storage medium

Abstract

The application discloses a psychological state monitoring method and device and a computer readable storage medium, and belongs to the technical field of biological signal processing; the method comprises the following steps: synchronously acquiring an EEG signal, a artifact source signal used for representing a physical motion artifact, and a blood vessel activity signal which has a physiological coupling relationship with neural activity; calculating a data reliability coefficient based on the intensity of the artifact source signal in a calculation time window; extracting an EEG real-time feature from the EEG signal and extracting a blood vessel real-time feature from the blood vessel activity signal; cross verifying the EEG real-time feature and the blood vessel real-time feature based on a neural-vessel coupling physiological model to determine whether the EEG signal change is effective physiological activity; dynamically adjusting the weight between a real-time physiological feature fusion value and a historical data benchmark based on the data reliability coefficient, calculating a psychological state index, and evaluating the user's psychological state based on the psychological state index.

Description

Technical Field

[0001] This invention relates to the field of biosignal processing technology, specifically to a method, device, and computer-readable storage medium for monitoring psychological states. Background Technology

[0002] In recent years, with increasing attention to mental health, psychological state monitoring based on non-invasive brain-computer interface technology has been extensively studied. Among these, electroencephalography (EEG) is widely used due to its high temporal resolution and ability to directly reflect neural activity. By analyzing energy changes in specific frequency bands of EEG signals, an individual's psychological state, such as focus, relaxation, and fatigue, can be assessed.

[0003] To meet the needs of everyday applications, wearable EEG monitoring devices have emerged. However, existing technologies still face many challenges in practical applications. The first is the interference from motion artifacts. Wearable devices are typically used during users' daily activities (such as walking, talking, and even slight head movements). These physical movements cause relative displacement between the electrodes and the scalp or muscle electrical activity, generating artifact noise with intensity far exceeding that of weak EEG signals. This severely contaminates the raw data, leading to distorted monitoring results or even complete failure.

[0004] Secondly, existing technologies have limitations in artifact processing. Traditional filtering methods struggle to effectively separate artifacts that overlap with EEG signal frequency bands. Some methods attempt to use independent reference sensors (such as accelerometers) to identify motion artifacts, but after identification, they typically only discard contaminated data segments, leading to data discontinuity and hindering truly real-time, continuous monitoring. Quantifying the impact of artifacts and dynamically and smoothly correcting monitoring results accordingly remains a pressing problem.

[0005] Furthermore, psychological states are a complex physiological process, and a single-dimensional EEG signal is often insufficient to provide a comprehensive and accurate assessment. For example, certain external electromagnetic interferences may exhibit characteristics on the EEG spectrum similar to specific cognitive activities, making them difficult to distinguish based on EEG signals alone. The lack of information from other physiological dimensions for cross-validation limits the reliability and accuracy of the monitoring system.

[0006] Therefore, designing a technical solution that can effectively suppress various artifact interferences in dynamic and realistic daily environments and combine multi-dimensional physiological information for accurate and reliable continuous monitoring of psychological state is the current technical bottleneck in this field. Summary of the Invention

[0007] To address the technical problems in existing technologies, such as the susceptibility of wearable physiological signal monitoring devices to artifact interference in dynamic environments, leading to inaccurate monitoring results, and the difficulty in comprehensively assessing complex psychological states using a single signal dimension, this invention provides a psychological state monitoring method, device, and computer-readable storage medium.

[0008] This invention provides a method for monitoring psychological states, comprising the following steps: simultaneously acquiring at least one EEG signal characterizing neural activity, at least one artifact source signal characterizing physical motion artifacts, and at least one vascular activity signal characterizing a physiologically coupled relationship with the neural activity; and aligning the EEG signal, the artifact source signal, and the vascular activity signal according to the same computational time window; calculating a real-time data reliability coefficient based on the intensity of the artifact source signal within the computational time window, wherein the data reliability coefficient ranges from 0 to 1, and the value of the data reliability coefficient is negatively correlated with the intensity of the artifact source signal; extracting real-time EEG features characterizing cognitive activity states from the EEG signal, and extracting real-time features from the vascular activity signal... The system extracts real-time vascular features to characterize blood flow or blood oxygen metabolism. Based on a preset neuro-vascular coupling physiological model, cross-validation is performed on the real-time EEG features and the real-time vascular features to determine whether changes in the EEG signal constitute effective physiological activity. Based on the data reliability coefficient, the weight between the fusion value of the real-time physiological features formed by the real-time EEG features and the real-time vascular features and a preset historical data benchmark is dynamically adjusted to calculate a psychological state index. The user's psychological state is assessed based on the psychological state index. Furthermore, when the cross-validation result indicates that changes in the EEG signal are not effective physiological activity, the weight of the real-time EEG features in the fusion value of the real-time physiological features is reduced or the influence of the real-time EEG features on the psychological state index is suppressed.

[0009] Furthermore, the artifact source signals include motion signals acquired by a triaxial accelerometer to characterize head movement, and electromyographic signals acquired by an electromyography (EMG) sensing unit to characterize facial muscle activity. The vascular activity signals include blood flow pulse signals and blood oxygen saturation signals acquired by a photoplethysmography (PPG) module. By acquiring motion and EMG signals, the degree to which the current EEG signal is affected by head movements, facial expressions, blinking, or frowning can be quantified. By acquiring blood flow pulse and blood oxygen saturation signals, vascular metabolic dimension information that is physiologically coupled with neural activity can be provided, which can be used to verify the physiological consistency of EEG signal changes. By acquiring signals generated by specific motion and EMG activities, artifact sources generated by body movement or facial expressions can be identified and quantified more accurately.

[0010] In one implementation, the weights of the real-time physiological feature fusion value and the historical data benchmark in calculating the psychological state index are dynamically adjusted based on the data reliability coefficient. This includes: normalizing the real-time EEG features extracted from the EEG signal and the real-time vascular features extracted from the vascular activity signal; weighting and summing the normalized real-time EEG features and the real-time vascular features according to preset feature weights to obtain the real-time physiological feature fusion value; multiplying the real-time physiological feature fusion value by the data reliability coefficient to obtain a first calculation result; multiplying the historical data benchmark by 1 and the difference between the data reliability coefficient and the result to obtain a second calculation result; and adding the first calculation result and the second calculation result to obtain the psychological state index. This method provides a smooth transition mechanism: when the artifact source signal strength is low and the data reliability coefficient is high, the psychological state index is more biased towards the current real-time physiological features; when the artifact source signal strength is high and the data reliability coefficient is low, the psychological state index is more biased towards the historical data benchmark, thereby avoiding sudden changes in the psychological state index due to motion artifacts.

[0011] Optionally, the artifact source signal includes motion signals acquired by a triaxial accelerometer and electromyographic signals acquired by an electromyographic sensing unit. The data reliability coefficient is calculated using an exponential decay function, which is based on the standard deviation of the motion signal extracted from the artifact source signal and the envelope amplitude of the electromyographic signal. This calculation method can non-linearly map the intensity of physical motion to a coefficient between 0 and 1, conforming to the physical laws governing artifact effects.

[0012] Furthermore, based on a preset neuro-vascular coupling physiological model, cross-validation is performed on the real-time EEG features and the real-time vascular features, including: when the increment of the real-time EEG features relative to the historical EEG baseline is greater than a first threshold, a candidate cognitive activity enhancement event is determined to have occurred; within a preset delay time window after the occurrence of the candidate cognitive activity enhancement event, it is determined whether the increment of the real-time vascular features relative to the historical vascular baseline is greater than a second threshold; if the increment of the real-time vascular features relative to the historical vascular baseline is greater than the second threshold, the change in the EEG signal is determined to be a valid physiological activity; if the increment of the real-time vascular features relative to the historical vascular baseline is not greater than the second threshold, the change in the EEG signal is determined not to be a valid physiological activity, and the weight of the real-time EEG features is adjusted to λ times the original weight, where the value of λ ranges from 0 to 0.5, or the weight of the real-time EEG features within the current calculation time window is reset to zero.

[0013] The present invention also provides a psychological state monitoring device, comprising: a sensor group configured to synchronously acquire at least one EEG signal for characterizing neural activity, at least one artifact source signal for characterizing physical motion artifacts, and at least one vascular activity signal for characterizing vascular activity that has a physiological coupling relationship with the neural activity; and a processor electrically connected to the sensor group, the processor being configured to calculate a real-time data reliability coefficient based on the intensity of the artifact source signal, wherein the value of the data reliability coefficient is negatively correlated with the intensity of the artifact source signal.

[0014] In one embodiment, the device is a wearable device comprising a packaging carrier on which the sensor array is disposed. The packaging carrier employs a flexible circuit board structure adapted to conform to the curvature of the human body. This structural design improves the wearing comfort and signal contact stability of the device.

[0015] Preferably, the sensor array includes at least three EEG acquisition electrodes arranged in an isosceles triangle for acquiring the EEG signal. This arrangement optimizes the electrode spacing within a limited space, which helps to improve the differential gain and spatial resolution of signal acquisition.

[0016] Furthermore, the device also includes a ring-shaped shielding layer disposed around the EEG acquisition electrodes and electrically connected to system ground or a reference potential. This shielding layer can effectively suppress environmental electromagnetic noise from external space at the physical level, thereby improving the signal-to-noise ratio from a hardware perspective.

[0017] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the preceding claims.

[0018] The beneficial effects of this invention are as follows:

[0019] (1) Improved anti-interference and robustness of monitoring results. By introducing artifact source signals and calculating real-time data reliability coefficients, this invention can quantify the degree of contamination of the current signal by motion artifacts and dynamically adjust the weights of real-time data and historical benchmarks accordingly. This enables the system to smoothly transition to the reference historical state when faced with data distortion caused by violent motion, avoiding drastic jumps and erroneous judgments in the output results, and ensuring the continuity and stability of monitoring in dynamic scenarios.

[0020] (2) Improved accuracy and reliability of psychological state assessment. This invention breaks through the limitations of a single EEG dimension and introduces a cross-validation mechanism based on a neurovascular coupling physiological model. By comparing whether the changing trends of EEG signals and vascular activity signals conform to physiological laws, non-physiological artifacts (such as electromagnetic interference) can be effectively identified and eliminated, ensuring that the analyzed signals truly reflect neural activity, thereby improving the accuracy of psychological state assessment.

[0021] (3) It balances noise immunity at the hardware level with ease of use. By adopting an optimized design with an isosceles triangular electrode layout and a ring-shaped shielding layer, the noise suppression capability is enhanced from the physical source. Combined with the wearable design of the flexible circuit board, the device can ensure signal quality while also providing good wearing comfort and portability, expanding its application potential in real-world scenarios such as daily office work, study, and health management. Attached Figure Description

[0022] Figure 1A This is a schematic diagram of the application surface structure of a psychological state monitoring device according to an embodiment of the present invention.

[0023] Figure 1B This is a side cross-sectional view of a psychological state monitoring device according to an embodiment of the present invention.

[0024] Figure 1C This is a perspective view of a psychological state monitoring device according to an embodiment of the present invention.

[0025] Figure 2 This is a hardware system architecture block diagram of a psychological state monitoring device according to an embodiment of the present invention.

[0026] Figure 3 This is a high-level flowchart of a psychological state monitoring method according to an embodiment of the present invention.

[0027] Figure 4 This is a schematic diagram of the cross-validation logic based on the neurovascular coupling physiological model in an embodiment of the present invention.

[0028] Figure 5 This is a schematic diagram illustrating the wearing and use scenario of a psychological state monitoring device according to an embodiment of the present invention.

[0029] Explanation of reference numerals in the attached figures:

[0030] 100 – Flexible multimodal patch; 500 – Mobile analysis terminal; E1, E2, E3, E4, E5 – EEG acquisition electrodes; S1 – Multimodal signal acquisition step; S2 – Reliability coefficient calculation step; S3 – Mental state index calculation step; S4 – State assessment step; 401 – Raw signal acquisition layer; 402 – Adaptive artifact removal engine; 403 – Feature extraction matrix; 404 – Cross-validation decision logic; 405 – Mental state index. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0032] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0033] This invention provides a psychological state monitoring system, which includes a wearable psychological state monitoring device and an external mobile analysis terminal 500, such as a smartphone or tablet computer. Figure 5 As shown, the psychological state monitoring device can specifically be a flexible multimodal patch 100, which the user can wear on a specific part of the body, such as the forehead. This device is responsible for collecting the user's multimodal physiological and physical signals in real time, processing and analyzing the data internally, calculating an index characterizing the user's psychological state, and transmitting the results wirelessly to a mobile analysis terminal 500 for visualization and further application.

[0034] Please see Figure 2This invention demonstrates the hardware system architecture of a psychological state monitoring device according to an embodiment of the present invention. The core of the device is a processor, which is responsible for the control, data processing, and algorithm execution of the entire system. Electrically connected to the processor is a sensor group configured to simultaneously acquire multiple signals. Specifically, the sensor group includes an EEG acquisition circuit for acquiring EEG signals, and other sensors for acquiring artifact source signals and vascular activity signals, such as a triaxial accelerometer, an electromyography (EMG) sensing unit, and a photoplethysmography (PPG) monitoring module (hereinafter referred to as the PPG module). Data acquired by all sensors is sent to the processor. After executing a preset psychological state monitoring method, the processor transmits the processing results to an external mobile analysis terminal 500 via a wireless transmission module. Through this integrated hardware design, the entire process from multimodal data acquisition to processing and wireless transmission is completed within a miniaturized device, providing a hardware foundation for portable, real-time psychological state monitoring.

[0035] To achieve the above functions, this invention provides a specific structure for a psychological state monitoring device. Please refer to Figure 1, which shows a schematic diagram of the device's structure. In one specific embodiment, the device is designed as a miniaturized device, with its overall size controlled within the range of no more than 5cm x 3cm. For example, in a preferred embodiment, its size is 4.8cm x 2.8cm, and its thickness is no more than 5mm. This miniaturized design improves user comfort and concealment, reducing the burden on the user. To further improve wearing comfort and signal acquisition stability, the device's encapsulation carrier adopts a flexible circuit board structure suitable for conforming to the curvature of the human body. Specifically, the encapsulation carrier can adopt a structure combining a flexible circuit board (FPC) with medical-grade biocompatible adhesive. This flexible structure allows the device to adaptively conform to curved surfaces such as the forehead, ensuring close and stable contact between the sensor and the skin, thereby effectively reducing motion artifacts caused by insecure wearing and ensuring the quality of signal acquisition.

[0036] Please see Figure 1AThe figure shows a schematic diagram of the device's application surface structure. An electrode array for acquiring EEG signals is disposed on this application surface. In a preferred embodiment, the electrode array includes at least three EEG acquisition electrodes E1, E2, and E3 arranged in an isosceles triangle. EEG acquisition electrodes E1, E2, and E3 are dry electrodes. Two of the EEG acquisition electrodes, E1 and E2, are located at the two ends of the base of the isosceles triangle, with a center-to-center distance of 4 cm. When the device is applied to the forehead, these two electrodes correspond to the Fp1 and Fp2 regions on both sides of the forehead, respectively. The third EEG acquisition electrode, E3, is located at the vertex of the isosceles triangle, with a distance of 2.5 cm from the orthocenter of the base, corresponding to the Fpz region in the center of the forehead. Furthermore, to acquire EEG characteristics from more regions, the electrode array may also include EEG acquisition electrodes E4 and E5 corresponding to the AF3 and AF4 regions on the outer sides of the forehead, respectively. This isosceles triangular electrode layout increases the effective spacing between electrodes within a limited patch area, thereby helping to improve the differential gain and signal-to-noise ratio of the acquired EEG signal.

[0037] To effectively suppress electromagnetic interference from the external environment, the device also includes a ring-shaped shielding layer. For example... Figure 1A As shown, the annular shielding layer is disposed around the EEG acquisition electrodes (e.g., E1, E2, E3, E4, E5) and completely surrounds the electrode array. In one embodiment, the annular shielding layer may be a flexible conductive silver paste layer with a width of 2 mm, electrically connected to system ground or a reference potential. This design physically forms a Faraday cage-like structure, which can effectively shield and absorb electromagnetic noise from space, such as 50 / 60 Hz power frequency interference and radio frequency interference generated by surrounding electronic devices, thereby improving the anti-interference capability and purity of the EEG signal.

[0038] In addition to the EEG acquisition electrodes, other sensors are integrated into the device's application surface. For example, such as Figure 1A As shown, a PPG monitoring module window can be set up, with the PPG monitoring module integrated below it. This module may include a 660nm red LED, a 940nm infrared LED, and a high-sensitivity photodiode. It measures the volume changes of microvessels under the skin of the forehead using transmission or reflection photoplethysmography, thereby acquiring vascular activity signals. In addition, the device can also integrate a triaxial accelerometer unit and an infrared thermopile temperature sensor (TEMP). Figure 1B The diagram shows a side cross-sectional view of the device, illustrating its multi-layered stacked structure, which may include encapsulation layers, circuit layers, electrode layers, and adhesive layers. Figure 1C This showcases the device's three-dimensional appearance. Its outer surface can also be equipped with status indicator lights to display the device's current operating mode, as well as charging contacts for connecting to an external power source for power replenishment.

[0039] Based on the aforementioned psychological state monitoring device, this invention further provides a method for monitoring psychological states. Please refer to [link / reference]. Figure 3 This figure is a high-level flowchart of a psychological state monitoring method according to an embodiment of the present invention. This method aims to solve the problems of traditional EEG monitoring being susceptible to artifact interference in dynamic scenarios and having a single assessment dimension, thereby providing a more accurate and robust psychological state assessment scheme.

[0040] The method first performs a multimodal signal acquisition step S1. In this step, the device's processor synchronously acquires multiple signals from a sensor array at a preset sampling rate (e.g., 250 Hz). These signals include at least: at least one EEG signal characterizing neural activity; at least one artifact source signal, such as a signal characterizing head movement acquired by a triaxial accelerometer and a signal characterizing facial muscle activity acquired by an electromyography (EMG) sensing unit; and at least one vascular activity signal, such as a signal acquired by a PPG module. Synchronous acquisition is fundamental for subsequent multimodal data fusion and cross-validation.

[0041] Next, the method proceeds to step S2, which calculates the reliability coefficient. To address the issue of EEG signals being easily contaminated by motion artifacts in dynamic scenes, this invention introduces a quantitative assessment of data reliability. The processor calculates a data reliability coefficient in real time based on the intensity of the artifact source signal obtained in step S1. This data reliability coefficient is negatively correlated with the intensity of the artifact source signal. That is, when vigorous body movement or facial muscle activity is detected, the artifact source signal intensity is high, resulting in a low data reliability coefficient, and vice versa. This coefficient provides prior information about the current data quality for subsequent data processing.

[0042] After obtaining the data reliability coefficient, the method proceeds to step S3, which calculates the psychological state index. In this step, the processor dynamically adjusts the weights of real-time features extracted from the real-time signal and historical data benchmarks in calculating the psychological state index based on the data reliability coefficient. Simultaneously, the method includes a cross-validation logic: based on a preset neuro-vascular coupling physiological model, the EEG signal and the vascular activity signal are cross-validated to determine whether changes in the EEG signal represent valid physiological activity. This step is used to assess the psychological state; it not only considers the unreliability of data due to artifacts but also utilizes the inherent correlations between different physiological signals to distinguish between genuine and false signals.

[0043] Finally, the method proceeds to state assessment step S4. Based on the mental state index calculated in step S3, the processor assesses the user's current mental state (e.g., focus, relaxation, or tension). Specifically, when cross-validation indicates that the changes in the EEG signal are not valid physiological activity, the processor adjusts the influence of the EEG signal in assessing the user's mental state accordingly, for example, by reducing its weight or directly ignoring the characteristics of that signal segment. Through this series of steps, the present invention can output a more reliable and accurate mental state assessment result that has been validated by artifact suppression and physiological mechanisms.

[0044] The core algorithm details in the above method flow will be explained in more detail below.

[0045] First, regarding the calculation of the data reliability coefficient, this invention provides a specific calculation method. The data reliability coefficient can be calculated using an exponential decay function, which is based on the standard deviation of the motion signal extracted from the artifact source signal and the envelope amplitude of the electromyographic signal. For example, the data reliability coefficient can be calculated using the formula... The calculation yielded the result. In this calculation method, This represents the standard deviation of the resultant vibration measured by the triaxial accelerometer within a certain time window. This value can effectively quantify the degree of head shaking or body vibration of the user. This represents the envelope amplitude obtained after filtering and envelope extraction of the electromyography (EMG) signal within the same time window. This value mainly reflects the activity intensity of the user's facial muscles (such as blinking and frowning). k1 and k2 are preset non-negative attenuation coefficients, which control the degree of influence of motion signals and EMG signals on the data reliability coefficient, respectively. In a specific embodiment, k1 can be set to 0.5 and k2 can be set to 0.8. The physical meaning of this calculation method is that when the user is in a static state, and When the values ​​are all close to 0, the data reliability coefficient approaches 1, indicating that the data is highly reliable; however, when users engage in strenuous exercise or frequently make facial expressions, or An increase in the reliability coefficient will cause the data reliability coefficient to decay rapidly to near 0 in an exponential manner, indicating that the data is severely contaminated and has low reliability.

[0046] Secondly, regarding the calculation of the psychological state index, this invention proposes a dynamic weighted fusion strategy. The dynamic adjustment of the weights of real-time features and historical data benchmarks based on the data reliability coefficient specifically includes the following calculation process: First, the real-time features extracted from the real-time signal are multiplied by the data reliability coefficient to obtain a first calculation result; then, the historical data benchmark is multiplied by the difference between 1 and the data reliability coefficient to obtain a second calculation result; finally, the first calculation result and the second calculation result are added to obtain the final psychological state index. This calculation process can be summarized as: Psychological State Index = (Data Reliability Coefficient × Real-time Physiological Feature Fusion Value) + ((1 − Data Reliability Coefficient) × Historical Data Benchmark). In this calculation, the real-time physiological feature fusion value refers to the sum of multiple feature values ​​extracted from and normalized by each modality of real-time signal (e.g., the energy ratio of β waves to α waves in the EEG spectrum, the rate of change of blood oxygen saturation extracted from vascular activity signals, or the body surface temperature fluctuation value measured by a temperature sensor, etc.) multiplied by their respective weighting factors. Historical data baselines are relatively smooth and stable state reference values, such as the weighted average state baseline value over a user's past period. This calculation method uses a data reliability coefficient as a harmonizing factor to achieve a dynamic balance between real-time response and stable output.

[0047] Secondly, regarding the calculation of real-time physiological feature fusion values, this invention provides a specific weighted fusion method. The processing unit first normalizes the real-time EEG features and real-time vascular features to obtain normalized real-time EEG features. Real-time characteristics of blood vessels after normalization Then, the real-time physiological feature fusion value is calculated according to the following formula. :

[0048]

[0049] in, Represents the EEG feature weights. Represents the weights of vascular features, and satisfies:

[0050]

[0051] In a preferred embodiment, The value range is from 0.5 to 0.8. The value range is 0.2 to 0.5. Since EEG signals directly reflect neural activity, and vascular activity signals are used to verify the physiological coupling of EEG feature changes, the weight of EEG features can be set higher than the weight of vascular features. Preferably, , The preset feature weights can be pre-stored in the processor or determined through calibration. Specifically, when the device leaves the factory or is worn by the user for the first time, EEG signals and vascular activity signals can be collected in a resting state and a standard cognitive task state, respectively. The weights are determined based on the correlation between real-time EEG features and reference heart rate status indicators. Determined based on the correlation between real-time vascular characteristics and reference heart rate status markers And normalize the two to make them satisfy When no individualized calibration is performed, the processor uses the default weights. , .

[0052] Furthermore, regarding the cross-validation logic based on the neurovascular coupling physiological model, this invention utilizes the physiological basis of brain activity. Please refer to [link / reference needed]. Figure 4 The diagram illustrates the logical flow of cross-validation. The signal enters from the raw signal acquisition layer 401, undergoes preliminary processing by the adaptive artifact removal engine 402, then enters the feature extraction matrix 403 to extract features for each modality, subsequently entering the cross-validation judgment logic 404, and finally outputting the validated psychological state index 405. This validation logic is based on a recognized physiological model, namely, a positive correlation between neural activity and local blood flow and blood oxygen metabolism. Specifically, when the increment of the real-time EEG feature relative to the historical EEG baseline exceeds a first threshold, a candidate cognitive activity enhancement event is identified (e.g., a surge in β-wave energy related to deep focus is detected). The system does not immediately determine that the user has entered a focused state, but instead initiates cross-validation. It checks within a preset delay time window after the occurrence of the candidate cognitive activity enhancement event whether the increment of the real-time vascular feature relative to the historical vascular baseline exceeds a second threshold. If the increment of the real-time vascular features relative to the historical vascular baseline is greater than the second threshold (i.e., the blood flow pulse intensity or blood oxygen saturation in the vascular activity signal also show a corresponding upward trend), it proves that the metabolic activity in the frontal lobe region of the brain has indeed increased, and the system determines that this change in the EEG signal is a real cognitive state change caused by effective physiological activity. Conversely, if the increment of the real-time vascular features relative to the historical vascular baseline is not greater than the second threshold (e.g., it does not rise synchronously, or even decreases or remains unchanged), the system will determine that the change in the EEG signal is likely not due to real neural activity, but more likely to be an illusion caused by external high-frequency electromagnetic interference (such as nearby mobile phone signals) or other artifacts. In this case, the system will perform an adjustment, that is, adjust the weight of the real-time EEG features to λ times the original weight, where the value of λ ranges from 0 to 0.5, or reset the weight of the real-time EEG features within the current calculation time window to zero, thereby avoiding incorrect judgment.

[0053] To further illustrate the beneficial effects of the technical solution of this invention, a comparative description is provided below through two specific application scenarios.

[0054] Scenario 1: The user is in a static work or study state.

[0055] In this scenario, the user's body and head movements are minimal, and facial expressions are relatively few. Therefore, the artifact source signals acquired by the triaxial accelerometer and electromyography (EMG) sensing unit are extremely weak, with characteristic values ​​(such as standard deviation and envelope amplitude) approaching 0. Based on the calculation method for the data reliability coefficient, the calculated data reliability coefficient will be very close to 1. At this point, in the formula for calculating the psychological state index, the term (1 - data reliability coefficient) approaches 0, the weight of historical data benchmarks is almost negligible, and the value of the psychological state index is mainly determined by real-time characteristic terms. This means that the system completely trusts the currently acquired real-time physiological signals and can accurately and sensitively reflect the user's true psychological state changes, such as an increase in focus or a decrease due to fatigue.

[0056] Scenario 2: The user is in a dynamic state, such as walking in an office, talking to someone, or engaging in light activity. In this scenario, the user's head inevitably shakes, and facial muscles also move more, leading to a significant increase in the intensity of the artifact source signal, and its feature value becomes very large. According to the formula, the data reliability coefficient will rapidly decay exponentially, approaching 0. At this time, in the formula for calculating the psychological state index, the weight of the data reliability coefficient term becomes extremely small, while the weight of the (1 - data reliability coefficient) term approaches 1. Therefore, the value of the psychological state index will be mainly determined by historical data benchmarks, while the influence of contaminated real-time features on the psychological state index is greatly weakened. In this way, the present invention effectively avoids drastic and false jumps in the psychological state index caused by motion artifacts, ensuring the smoothness and stability of the output curve, so that the monitoring results still have reference value and credibility in dynamic scenarios.

[0057] The application value of this invention is also reflected in its interaction with user terminals. The final psychological state index obtained after processing by the above algorithm, along with optional raw signal or feature data, can be transmitted in real time to the user's mobile analysis terminal 500, such as an application (APP) on a smartphone, via a wireless transmission module (e.g., Bluetooth Low Energy protocol). This APP can transform abstract numerical data into an intuitive graphical interface; for example, it can display the user's status in multiple dimensions such as "focus," "relaxation," and "fatigue warning" in real time using a radar chart. Furthermore, the APP can execute warning logic based on the received data. For example, when the system detects that the user's psychological state index has been continuously below a preset threshold for a certain period (e.g., 30 seconds), and simultaneously the temperature sensor shows abnormal fluctuations in body surface temperature, the APP can proactively trigger a "fatigue rest prompt," reminding the user to rest in a timely manner, thus playing a practical role in health monitoring and human-computer interaction.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. The present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the method described. Since the above embodiments have been described in detail, those skilled in the art can clearly understand that the content of the above method embodiments can be implemented in the form of a computer program and stored in a computer-readable storage medium, such as ROM, RAM, disk, or optical disk, etc., which will not be repeated here.

Claims

1. A method for monitoring psychological states, characterized in that, Includes the following steps: At least one EEG signal for characterizing neural activity, at least one artifact source signal for characterizing physical motion artifacts, and at least one vascular activity signal for characterizing vascular activity that has a physiological coupling relationship with the neural activity are acquired simultaneously, and the EEG signal, the artifact source signal and the vascular activity signal are time-aligned according to the same calculation time window; Based on the intensity of the artifact source signal within the calculation time window, a real-time data reliability coefficient is calculated, wherein the value of the data reliability coefficient ranges from 0 to 1, and the value of the data reliability coefficient is negatively correlated with the intensity of the artifact source signal. Real-time EEG features for characterizing cognitive activity status are extracted from the EEG signal, and real-time vascular features for characterizing blood flow or blood oxygen metabolism status are extracted from the vascular activity signal. Based on a pre-defined neuro-vascular coupling physiological model, the real-time EEG features and the real-time vascular features are cross-validated to determine whether the changes in the EEG signal are valid physiological activities. Based on the data reliability coefficient, the weight between the real-time physiological feature fusion value formed by the real-time EEG features and the real-time vascular features and the preset historical data benchmark is dynamically adjusted to calculate the psychological state index. The user's psychological state is assessed based on the psychological state index, and when the cross-validation result indicates that the change in the EEG signal is not a valid physiological activity, the weight of the real-time EEG feature in the real-time physiological feature fusion value is reduced or the influence of the real-time EEG feature on the psychological state index is suppressed.

2. The method according to claim 1, characterized in that, The artifact source signals include motion signals acquired by a triaxial accelerometer to characterize head movement, and electromyographic signals acquired by an electromyography sensing unit to characterize facial muscle activity; the vascular activity signals include blood flow pulse signals and blood oxygen saturation signals acquired by a photoplethysmography (PPG) module.

3. The method according to claim 1, characterized in that, Based on the data reliability coefficient, the weights between the real-time physiological feature fusion value formed by the real-time EEG features and the real-time vascular features and the preset historical data benchmark are dynamically adjusted to calculate the psychological state index, including: The real-time EEG features and the real-time blood vessel features are normalized. The normalized real-time EEG features and the real-time vascular features are weighted and summed according to preset feature weights to obtain the real-time physiological feature fusion value. Multiply the real-time physiological feature fusion value by the data reliability coefficient to obtain the first calculation result; The second calculation result is obtained by multiplying the historical data benchmark by 1 and the difference between the data reliability coefficient and the benchmark. The psychological state index is obtained by adding the first calculation result to the second calculation result.

4. The method according to claim 3, characterized in that, The artifact source signal includes motion signals acquired by a triaxial accelerometer and electromyographic signals acquired by an electromyographic sensing unit; the data reliability coefficient is calculated using an exponential decay function based on the standard deviation extracted from the motion signal and the envelope amplitude of the electromyographic signal.

5. The method according to claim 1, characterized in that, Based on a pre-defined neuro-vascular coupling physiological model, cross-validation is performed on the real-time EEG features and the real-time vascular features, including: when the increment of the real-time EEG features relative to the historical EEG baseline is greater than a first threshold, a candidate cognitive activity enhancement event is identified; within a pre-defined delay time window after the occurrence of the candidate cognitive activity enhancement event, it is determined whether the increment of the real-time vascular features relative to the historical vascular baseline is greater than a second threshold; if the increment of the real-time vascular features relative to the historical vascular baseline is greater than the second threshold, the change in the EEG signal is determined to be a valid physiological activity; if the increment of the real-time vascular features relative to the historical vascular baseline is not greater than the second threshold, the change in the EEG signal is determined not to be a valid physiological activity, and the weight of the real-time EEG features is adjusted to λ times the original weight, where the value of λ ranges from 0 to 0.5, or the weight of the real-time EEG features within the current calculation time window is reset to zero.

6. A psychological state monitoring device, characterized in that, The device includes: a sensor array configured to simultaneously acquire at least one EEG signal characterizing neural activity, at least one artifact source signal characterizing physical motion artifacts, and at least one vascular activity signal characterizing vascular activity that is physiologically coupled to the neural activity; and A processor, electrically connected to the sensor group, is configured to: Based on the intensity of the artifact source signal, a real-time data reliability coefficient is calculated, wherein the value of the data reliability coefficient is negatively correlated with the intensity of the artifact source signal. Real-time EEG features for characterizing cognitive activity status are extracted from the EEG signal, and real-time vascular features for characterizing blood flow or blood oxygen metabolism status are extracted from the vascular activity signal. Based on a pre-defined neuro-vascular coupling physiological model, the real-time EEG features and the real-time vascular features are cross-validated to determine whether the changes in the EEG signal are valid physiological activities. Based on the data reliability coefficient, the weight between the real-time physiological feature fusion value formed by the real-time EEG features and the real-time vascular features and the preset historical data benchmark is dynamically adjusted to calculate the psychological state index. The user's psychological state is assessed based on the psychological state index, and when the cross-validation result indicates that the change in the EEG signal is not a valid physiological activity, the weight of the real-time EEG feature in the real-time physiological feature fusion value is reduced or the influence of the real-time EEG feature on the psychological state index is suppressed.

7. The apparatus according to claim 6, characterized in that, The device is a wearable device, which includes a packaging carrier, on which the sensor group is disposed. The packaging carrier adopts a flexible circuit board structure suitable for conforming to the curvature of the human body.

8. The apparatus according to claim 6, characterized in that, The sensor group includes at least three EEG acquisition electrodes arranged in an isosceles triangle for acquiring the EEG signal.

9. The apparatus according to claim 8, characterized in that, The device also includes an annular shielding layer disposed around the EEG acquisition electrode and electrically connected to system ground or a reference potential.

10. A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method of any one of claims 1 to 5.

Images