A physiological and electrocardio comprehensive detection method and system for evaluating psycho-neural activity

Abstract

The present application relates to the technical field of intelligent physiological information detection, in particular to a physiological and electrocardio comprehensive detection method and system for evaluating psychological neural activity; the method comprises the following steps: collecting physiological electrical signals and interface impedance sequences of frontal and temporal regions, normalizing the interface impedance sequences, and extracting total energy of the frontal region and total energy of the temporal region; constructing frontal region compression proxy, temporal region compression proxy, frontal-temporal contact pressure migration index and frontal-temporal reverse linkage coefficient based on the normalized impedance sequences; constructing original psychological neural activity load based on the total energy of the frontal region and the total energy of the temporal region, and combining the above quantities to form a feature vector for state recognition to obtain a stress migration state sequence; constructing a migration interference component, correcting the original psychological neural activity load, and combining a reference baseline to output a detection result. The present application can separate the pseudo changes caused by contact pressure migration, and improve the authenticity, stability and interpretability of psychological neural activity evaluation.

Description

Technical Field

[0001] This invention relates to the field of intelligent physiological information detection technology, specifically to a comprehensive physiological and electrocardiogram detection method and system for assessing psychoneural activity. Background Technology

[0002] With the continuous development of immersive training, virtual reality interaction, brain-computer interface-assisted assessment, and AI-driven human-machine state analysis technologies, the real-time detection of subjects' psychoneural activities during training using head-mounted physiological acquisition devices has become an important technical direction in cognitive load assessment, alertness level analysis, psychological stress monitoring, and training quality optimization. Existing solutions typically collect physiological electrical signals from the frontal and temporal regions and combine this with pattern recognition to quantify changes in subjects' psychoneural activities during task execution, thus providing a data foundation for training feedback, adaptive task adjustment, and intelligent assessment. However, in immersive training scenarios, the head-mounted acquisition device is attached to the forehead and bilateral temporal regions for extended periods. Subjects under tension, focus, or stress are prone to low-amplitude, sustained changes in facial muscle tension, causing a redistribution of contact pressure between the frontal support position and the temporal tightening position. This leads to changes in interfacial impedance and the apparent distribution of regional physiological electrical activity, resulting in additional perturbations in the detection results caused by both actual psychoneural activity changes and contact state changes, affecting the authenticity and stability of the assessment results.

[0003] Existing technologies, when addressing the aforementioned issues, typically focus more on denoising, classifying, or improving the accuracy of artificial intelligence modeling of physiological electrical signals. However, they lack specific processing for the contact pressure migration caused by stress between the forehead and temporal regions of head-mounted devices. They often simply treat related signal fluctuations as random noise, motion interference, or individual differences and filter them out uniformly. This type of processing makes it difficult to identify the continuous relationship between the interfacial impedance changes caused by contact pressure redistribution and the regional signal shift. Furthermore, it is difficult to distinguish between real changes in psychoneural activity and spurious changes caused by changes in wearing status. This can easily lead to problems such as biased detection results, inaccurate classification, and distortion of artificial intelligence model input, thereby affecting the reliability, interpretability, and practical application effect of psychoneural activity assessment in immersive training scenarios. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies, such as the difficulty in identifying additional disturbances caused by the migration of contact pressure between the frontal and temporal regions, which makes it difficult to separate real changes in psychoneural activity from spurious changes caused by changes in contact state, resulting in insufficient authenticity and stability of detection results. This invention proposes a comprehensive physiological and electrocardiographic detection method and system for assessing psychoneural activity.

[0005] The technical solution of this invention: A comprehensive physiological and electrocardiographic detection method for assessing psychoneural activity, comprising: S1. Collect physiological electrical signals and interface impedance sequences of the frontal and temporal regions, normalize the interface impedance sequences to obtain normalized impedance sequences, and extract the total energy of the frontal and temporal regions according to the time window. S2. Construct the frontal and temporal compression surrogate quantities based on the normalized impedance sequence, and calculate the frontotemporal contact pressure migration index based on the frontal and temporal compression surrogate quantities. S3. Calculate the frontotemporal reverse linkage coefficient based on the normalized impedance sequence, construct the original psychoneural activity load based on the total energy of the frontal region and the total energy of the temporal region, form a feature vector based on the combination of the original psychoneural activity load, the frontotemporal contact pressure migration index and the frontotemporal reverse linkage coefficient, and perform state recognition on the feature vector to obtain the pressure migration state sequence. S4. Construct migration interference components based on the stress migration state sequence, and correct the original psychoneural activity load based on the migration interference components to obtain the corrected psychoneural activity load. S5. Compare the corrected psychoneural activity level with the reference baseline to obtain the relative offset, determine the psychoneural activity detection result based on the relative offset, generate auxiliary physiological prompts based on ECG-related auxiliary physiological parameters, and use the psychoneural activity detection result and auxiliary physiological prompts together as a comprehensive detection output.

[0006] Preferably, the interface impedance sequence is normalized to obtain a normalized impedance sequence, and the total energy of the frontal region and the total energy of the temporal region are extracted according to a time window, including: Physiological electrical signals and interfacial impedance sequences were collected at corresponding locations on the left frontal, right frontal, left temporal, and right temporal sides, respectively. Calculate the median and absolute median difference of the interface impedance sequence corresponding to each acquisition location; Calculate the difference between the current time series value of the interface impedance sequence at each acquisition location and the corresponding median, and divide by the corresponding absolute median difference to obtain the normalized impedance sequence at each acquisition location. The signal energy of the physiological electrical signals corresponding to each collection location in the frontal region is extracted according to the time window, and the total energy of the frontal region is obtained by pooling them together. The signal energy of the physiological electrical signals corresponding to each acquisition location in the temporal region is extracted according to the time window, and the total energy of the temporal region is obtained by summing them.

[0007] Preferably, the frontal and temporal compression surrogate quantities are constructed based on the normalized impedance sequence, and the frontotemporal contact pressure migration index is calculated based on the frontal and temporal compression surrogate quantities, including: Calculate the mean value of the normalized impedance sequence corresponding to each acquisition position in the quota area within the current time window, and construct the quota area compression proxy based on the mean value; The mean value of the normalized impedance sequence corresponding to each acquisition location in the temporal region within the current time window is calculated, and the temporal region compression proxy is constructed based on the mean value. Based on the difference and amplitude relationship between the temporal compression surrogate and the frontal compression surrogate, the frontotemporal contact pressure migration index, which characterizes the relative migration direction and degree of pressure between the frontal and temporal regions, is calculated.

[0008] Preferably, the calculation of the frontotemporal inverse linkage coefficient based on the normalized impedance sequence includes: A local analysis window is constructed based on the current time window and its adjacent time windows; Within the local analysis window, the average time window value of the normalized impedance sequence corresponding to each acquisition location in the frontal region is averaged to obtain the average impedance sequence of the frontal region. Within the local analysis window, the average time window value of the normalized impedance sequence corresponding to each acquisition location in the temporal region is averaged to obtain the temporal region impedance mean sequence. The correlation coefficient between the mean impedance sequence of the frontal region and the negative sequence of the mean impedance sequence of the temporal region is calculated, and the correlation coefficient is used as the frontotemporal inverse linkage coefficient to characterize the inverse relationship between the changes in impedance of the frontal region and the temporal region.

[0009] Preferably, the original psychoneural activity load is constructed based on the total energy of the frontal region and the total energy of the temporal region. A feature vector is formed by combining the original psychoneural activity load, the frontotemporal contact pressure migration index, and the frontotemporal reverse linkage coefficient, including: The energy comparison data for the corresponding time window is constructed based on the ratio of total energy in the frontal region to total energy in the temporal region. Logarithmic transformation of the energy comparison quantity yields the original psychoneural activity load; The changes in the frontotemporal contact pressure migration index and the changes in the original psychoneural activity load between adjacent time windows are extracted. The original psychoneural activity load, the frontotemporal contact pressure migration index, the frontotemporal reverse linkage coefficient, the changes in the frontotemporal contact pressure migration index, and the changes in the original psychoneural activity load are combined to form a feature vector.

[0010] Preferably, state recognition is performed on the feature vector, including: The feature vectors are input into the Hidden Markov Model for state classification to obtain the stress migration state sequence corresponding to each time window. The pressure migration state sequence includes a stable fit state, a frontal-to-temporal migration state, a temporal-to-frontal migration state, and a migration transition state.

[0011] Preferably, a migration interference component is constructed based on the stress migration state sequence, and the original psychoneural activity load is corrected according to the migration interference component, including: For the frontotemporal migration state, temporal-frontal migration state, or migration transition state corresponding to the current time window, the migration interference component is constructed based on the state coefficients obtained by fitting the data obtained by the current subject in the calibration data acquisition segment after the reference baseline stage. The migration interference component includes a first component based on the frontotemporal contact pressure migration index and a second component based on the change of the frontotemporal contact pressure migration index between adjacent time windows. The original psychoneural activity load is differentially corrected based on the migration interference component to obtain the corrected psychoneural activity load. When the current time window corresponds to a stable fit state, the original psychoneural activity load is directly used as the corrected psychoneural activity load.

[0012] Preferably, the corrected psychoneural activity level is compared with a reference baseline to obtain a relative offset, and the psychoneural activity detection result is determined based on the relative offset, including: Calculate the statistical center value of the corrected psychoneural activity for each time window within the reference baseline period, and use it as the baseline reference value; The offset of the corrected psychoneural activity level within the current time window relative to the baseline reference value is used as the relative offset. The upper and lower decision limits are determined based on the offset distribution range of the corrected psychoneural activity amount relative to the baseline reference value for each time window within the reference baseline phase. When the relative offset is greater than the upper judgment threshold, a detection result representing enhanced psychoneural activity is obtained; When the relative offset is less than the lower decision threshold, a detection result representing a weakening of psychoneural activity is obtained; When the relative offset is between the upper and lower decision limits, a detection result representing stable psychoneural activity is obtained.

[0013] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: This invention constructs frontal compression surrogate quantity, temporal compression surrogate quantity, frontotemporal contact pressure migration index, and frontotemporal reverse linkage coefficient by simultaneously acquiring physiological electrical signals and interface impedance sequences of the frontal and temporal regions. Furthermore, it combines the total energy of the frontal and temporal regions to form the original psychoneural activity load. This enables the identification of the contact pressure redistribution process between the frontal and temporal regions caused by stress during psychoneural activity detection. By elevating contact state changes from a simple interference phenomenon to a quantifiable and analyzable migration characteristic, the detection process no longer relies solely on changes in single physiological electrical activity for judgment. Instead, it can simultaneously grasp the relationship between changes in the contact interface and changes in regional physiological activity, improving the ability to identify the source of psychoneural activity changes. This invention further obtains a stress migration state sequence by performing state recognition on the feature vector, and constructs a migration interference component in the migration-related state to perform directional correction on the original psychoneural activity load. This separates the spurious changes caused by frontotemporal contact stress migration from the real psychoneural activity changes. By adopting a relative offset determination method based on an individual reference baseline, adaptive detection of enhanced, weakened, and stable states of psychoneural activity can be achieved, reducing the impact of individual baseline differences and wearing state fluctuations on the detection results, and improving the authenticity, stability, and interpretability of psychoneural activity assessment in immersive training scenarios. Attached Figure Description

[0014] Figure 1 This is a flowchart of a comprehensive physiological and electrocardiogram detection method for assessing psychoneural activity proposed in this invention; Figure 2 This is a block diagram of a physiological and electrocardiogram integrated detection system for assessing psychoneural activity proposed in this invention. Detailed Implementation

[0015] Example 1, as Figure 1 As shown, the present invention proposes a comprehensive physiological and electrocardiographic detection method for assessing psychoneural activity, comprising: S1. Collect physiological electrical signals and interface impedance sequences of the frontal and temporal regions, normalize the interface impedance sequences to obtain normalized impedance sequences, and extract the total energy of the frontal and temporal regions according to the time window. S2. Construct the frontal and temporal compression surrogate quantities based on the normalized impedance sequence, and calculate the frontotemporal contact pressure migration index based on the frontal and temporal compression surrogate quantities. S3. Calculate the frontotemporal reverse linkage coefficient based on the normalized impedance sequence, construct the original psychoneural activity load based on the total energy of the frontal region and the total energy of the temporal region, form a feature vector based on the combination of the original psychoneural activity load, the frontotemporal contact pressure migration index and the frontotemporal reverse linkage coefficient, and perform state recognition on the feature vector to obtain the pressure migration state sequence. S4. Construct migration interference components based on the stress migration state sequence, and correct the original psychoneural activity load based on the migration interference components to obtain the corrected psychoneural activity load. S5. Compare the corrected psychoneural activity level with the reference baseline to obtain the relative offset, determine the psychoneural activity detection result based on the relative offset, generate auxiliary physiological prompts based on ECG-related auxiliary physiological parameters, and use the psychoneural activity detection result and auxiliary physiological prompts together as a comprehensive detection output.

[0016] In one embodiment of the present invention, physiological electrical signals and interface impedance sequences of the frontal and temporal regions are collected, the interface impedance sequences are normalized to obtain normalized impedance sequences, and the total energy of the frontal and temporal regions is extracted according to time windows, including: This embodiment applies to real-time assessment of head-mounted physiological data acquisition and psychoneural activity in immersive training scenarios. The entire process of this method is executed using a head-mounted physiological data acquisition device. The device is equipped with four acquisition electrodes, fixed to the subject's left frontal, right frontal, left temporal, and right temporal positions, respectively. The left and right frontal electrodes correspond to the Fp1 and Fp2 positions of the International EEG Electrode Positioning 10-20 system, while the left and right temporal electrodes correspond to the T7 and T8 positions. The four acquisition electrodes simultaneously acquire physiological electrical signals and interface impedance sequences at their corresponding positions. The physiological electrical signals are obtained using EEG signals at a sampling rate of 250 Hz. The interface impedance sequences are acquired synchronously with the physiological electrical signals using timestamps. The interface impedance is acquired using a two-electrode method, focusing on the contact impedance between the acquisition electrodes and a reference electrode uniformly set on the head-mounted device. The excitation signal for impedance acquisition was a 100 Hz, 10 μA constant current sinusoidal signal. The sampling rate of the interface impedance sequence was set to 1 Hz. The total acquisition time for a single session was set to 360 seconds, with the first 60 seconds being the reference baseline phase. During the reference baseline phase, the subject was required to remain in a resting state with eyes open, without cognitive task input, limb movement, or facial expression changes, and to maintain a stable head posture throughout to ensure that the head-mounted acquisition device was in stable contact with the skin of the forehead and temporal region. After the reference baseline phase, a 90-second calibration data acquisition segment was set, followed by a 210-second immersive task execution phase. A non-overlapping sliding time window of 1 second was used to window the acquired signal. Each time window corresponded to 250 sampling points for physiological electrical signals and 1 sampling point for the interface impedance sequence. If there was an incomplete time window of less than 1 second at the end of the total acquisition time, that part of the data could be discarded. In one optional embodiment, the head-mounted physiological data acquisition device can also establish a time synchronization relationship with an electrocardiogram (ECG) acquisition module, a blood pressure acquisition module, or an external wearable physiological monitoring device to synchronously acquire the subject's ECG signals and / or blood pressure parameters; the ECG signals are used to extract heart rate, ECG rhythm parameters, or heart rate variability parameters, and the blood pressure parameters may include systolic blood pressure, diastolic blood pressure, or mean arterial pressure; the ECG signals and / or blood pressure parameters serve as auxiliary physiological state reference quantities and are not involved in the calculation of frontal compression surrogate quantity, temporal compression surrogate quantity, frontotemporal contact pressure migration index, frontotemporal reverse linkage coefficient, and migration interference component; When the electrocardiogram (ECG) signal is a continuously acquired signal, it is divided into corresponding time windows according to the sampling timestamp, and heart rate, ECG rhythm parameters, or heart rate variability parameters are extracted within the corresponding time window. When the blood pressure parameter is a continuous blood pressure parameter, it is divided into corresponding time windows according to the sampling timestamp. When the blood pressure parameter is an intermittent measurement result, the blood pressure measurement result most recent to the current time window and not exceeding the preset valid duration is used as the blood pressure parameter corresponding to the current time window, or two adjacent blood pressure measurement results are kept in chronological order until the next update. Through the above timestamp alignment method, ECG signals and / or blood pressure parameters can be correlated with frontal total energy, temporal total energy, frontotemporal contact pressure migration index, and corrected psychoneural activity under the same time window number.

[0017] It should be noted that, to ensure stable acquisition of individual calibration samples corresponding to the frontotemporal migration state, temporal-frontal migration state, and migration transition state within the calibration data acquisition segment, subjects were guided to complete three sets of standardized pressure migration induction tasks sequentially during the 90-second calibration data acquisition segment. The first task was a frontotemporal migration induction task, which involved guiding subjects to continuously perform slight squinting and slight tension of the bilateral temporal masseter muscles while maintaining an overall unchanged head posture, and preventing excessive contraction of the glabella area. This increased the contact pressure between the bilateral tightening areas of the headband device and the forehead support area, thereby inducing the contact pressure to migrate from the frontal area to the temporal area. The second task was a temporal-to-frontal migration induction task. The execution method involved guiding the subject to continuously perform slight frowning and frontalis muscle lifting movements while maintaining the overall head posture, and relaxing the bilateral temporomandibular muscles. This increased the contact pressure between the frontal support area and the bilateral tightening areas, thereby inducing the contact pressure to migrate from the temporal area to the frontal area. The third task was a migration transition induction task. The execution method involved guiding the subject to slowly switch between frontal-to-temporal migration induction movements and temporal-to-frontal migration induction movements, or maintaining only low-amplitude, unstable, slight changes in facial tension between the two types of movements, so that the contact pressure between the forehead and temporal areas of the head-mounted device was in a transition process without a stable migration direction. All three sets of induction tasks required subjects to keep their shoulders stable, their heads still, and to refrain from translational or rotational movements, verbal articulation, and vigorous chewing to avoid head movement and mouth movement artifacts being mixed into the calibration data. Before each induction task began, subjects were prompted to enter the corresponding action preparation state. After each induction task, a rest recovery period was arranged to allow the frontotemporal contact state to return to a stable fit before proceeding to the next induction task. Through the above standardized induction method, the frontotemporal contact pressure migration process can be stably triggered without changing the overall wearing position of the head-mounted device, thus providing repeatable calibration samples for subsequent state calibration and individual coefficient fitting.

[0018] For four acquisition sites—left frontal, right frontal, left temporal, and right temporal—physiological electrophysiological signal sequences and interface impedance sequences were acquired simultaneously at the corresponding sites. The interface impedance sequence acquired at the left frontal site is denoted as... The interfacial impedance sequence acquired at the right forehead position is The interface impedance sequence acquired at the left temporal position is The interface impedance sequence acquired at the right temporal position is For each interface impedance sequence, first calculate the median and absolute median difference of the corresponding sequence. The formula for calculating the median of the sequence is: In the formula, The interface impedance sequence is collected at location identifier i, where i is the location identifier, and i can be taken as left frontal, right frontal, left temporal, and right temporal, respectively. This represents the total number of sampling points for a single-channel interface impedance sequence. This is for calculating the median; The formula for calculating the absolute median difference of a sequence is: The reason for using the median and absolute median deviation for statistical calculation is that the median and absolute median deviation are robust statistics that can effectively suppress outlier interference caused by impedance jumps and poor instantaneous contact of electrodes in the early stages of wearing the head-mounted device, avoid normalization bias caused by extreme values ​​affecting the conventional mean and standard deviation statistics, and ensure the stability and robustness of the normalization results. Robust normalization is performed on each interface impedance sequence to obtain the normalized impedance sequence for the corresponding acquisition location. The calculation formula for normalization is as follows: In the formula, Let be the normalized impedance value of the i-th acquisition position at time t.

[0019] For the physiological electrical signals within each time window, a bandpass finite-length unit impulse response filter from 0.5 Hz to 30 Hz is first applied. The filter order is set to 16, and a Hamming window is used. Zero-phase forward and reverse bidirectional filtering is performed to remove baseline drift, power line interference, and high-frequency electromyography noise from the physiological electrical signals. Then, the signal energy of each acquisition channel within the corresponding time window is calculated. The formula for calculating the signal energy of a single channel and a single time window is as follows: In the formula, k is the time window number; Let be the amplitude of the physiological electrical signal at the nth sampling point within the kth time window at the i-th acquisition location; This refers to the number of physiological electrical signal sampling points within a single time window. Take 250; using the sum of squares to calculate signal energy can directly characterize the activity intensity of physiological electrical signals within a time window. The squaring operation can amplify the amplitude difference of the effective signal while suppressing the interference of low-amplitude noise. The signal energy from the two acquisition locations (left and right frontal regions) within the frontal region is summed and aggregated according to the time window to obtain the total frontal region energy for the corresponding time window. The formula for calculating the total frontal region energy is as follows: In the formula, The total energy of the frontal region within the k-th time window. The signal energy at the left frontal position within the k-th time window. The signal energy at the right frontal position within the k-th time window; The signal energy from the two acquisition locations (left and right temporal regions) within the temporal region is summed and aggregated according to the time window to obtain the total temporal region energy for the corresponding time window. The formula for calculating the total temporal region energy is as follows: In the formula, The total energy of the temporal region within the k-th time window. The signal energy at the left temporal location within the k-th time window. The signal energy at the right temporal location within the k-th time window; The reason for conducting energy convergence on the frontal and temporal regions separately is that the physiological electrical activity in the frontal region is directly related to the cognitive load, alertness level, and psychological stress state of the subjects, and is the core representation area of ​​psychoneural activity. The physiological electrical activity in the temporal region is less directly affected by psychoneural activity and can serve as a stable physiological reference area. By using dual-point energy convergence, the impact of single-channel acquisition noise on the regional energy calculation results can be reduced, thereby improving the stability and anti-interference ability of subsequent psychoneural activity load calculation.

[0020] It should be noted that the interface impedance sequence refers to the data sequence formed by the impedance values ​​at the interface between the acquisition electrode and the subject's skin in chronological order during the acquisition of physiological electrical signals. It is used to characterize the tightness, stability, and changes in contact state between the head-mounted acquisition device and the skin of the forehead and temporal regions over time. When the pressure, contact area, or local force state between the electrode and the skin changes, the interface impedance sequence will change accordingly, thus serving as a fundamental characterization of the pressure migration process between the forehead and temporal regions. The total energy of the forehead region refers to the total energy value obtained by calculating the energy of the physiological electrical signals at each acquisition location within the forehead region within a time window and then performing regional aggregation. It is used to characterize the overall strength of the physiological electrical activity in the forehead region within that time window. The total energy of the forehead region is not the instantaneous amplitude of a single electrode channel, but rather the comprehensive result of the electrical activity intensity at multiple acquisition locations in the forehead region within the same time window, reflecting the overall activity level of the forehead region under changes in cognitive load, alertness level, or psychological stress. The total energy of the temporal region refers to the total energy value obtained by calculating the energy of the physiological electrical signals corresponding to each acquisition location in the temporal region within a time window and then performing regional aggregation. It is used to characterize the strength of the overall physiological electrical activity in the temporal region within that time window. The total energy of the temporal region serves as a regional reference quantity corresponding to the total energy of the frontal region. It can be used together with the total energy of the frontal region to construct the relative activity relationship of the region, thereby reducing the impact of single-channel fluctuations and individual baseline amplitude differences on the results of psychoneural activity detection.

[0021] It should be noted that the electrocardiogram (ECG) signal refers to the auxiliary physiological signal used to characterize the electrical activity state of the subject's heart, acquired through an ECG acquisition module or an external wearable physiological monitoring device. Heart rate, ECG rhythm parameters, or heart rate variability parameters can be extracted based on the ECG signal. Blood pressure parameters refer to the auxiliary physiological parameters used to characterize the subject's circulatory physiological state, obtained through a blood pressure acquisition module or an external wearable physiological monitoring device. These parameters may include systolic blood pressure, diastolic blood pressure, or mean arterial pressure. The ECG signal and blood pressure parameters are only used to assist in judging the subject's circulatory physiological fluctuations during the immersive task execution process. They are not necessary inputs for constructing the original psychoneural activity load, nor do they change the processing flow for constructing the psychoneural activity load based on the total energy of the frontal and temporal regions.

[0022] It should be noted that this embodiment uses the regional energy comparison between the frontal and temporal regions to construct the psychoneural activity load, rather than directly determining changes in psychoneural activity based on the absolute signal amplitude of a single electrode channel. Instead, it characterizes the relative activity relationship between the frontal and temporal regions at the same time. The frontal region corresponds to the prefrontal region and is more directly affected by cognitive task engagement, changes in alertness level, and stress state. In this embodiment, the temporal region is mainly used to provide a lateral reference that is relatively independent of the frontal region in the head-mounted acquisition structure. By using a dual-region comparison method of the frontal and temporal regions, the influence of individual scalp resistance, channel gain differences, and public environmental disturbances on the absolute amplitude of a single channel can be reduced while maintaining a small number of electrodes and a simple head-mounted device structure. This makes the subsequently constructed original psychoneural activity load more suitable for use as a relative change detection quantity.

[0023] In one embodiment of the present invention, a frontal compression surrogate quantity and a temporal compression surrogate quantity are constructed based on a normalized impedance sequence, and a frontotemporal contact pressure migration index is calculated based on the frontal compression surrogate quantity and the temporal compression surrogate quantity, including: For each time window, first calculate the mean of the normalized impedance sequence corresponding to each acquisition position in the frontal region within the current time window, and the mean of the normalized impedance sequence corresponding to each acquisition position in the temporal region within the current time window; then calculate the mean of the normalized impedance for a single acquisition position and a single time window using the following formula: In the formula, This represents the normalized mean impedance corresponding to the k-th time window. This represents the time range corresponding to the k-th time window. This represents the number of sampling points for the interface impedance sequence within a single time window. Let be the normalized impedance value of the i-th sampling location at time t. Calculation using the mean within a time window eliminates instantaneous sampling noise in the interface impedance sequence, smooths random impedance fluctuations, and ensures that the impedance mean is aligned with the physiological electrical signal energy calculation within the same time window in the time dimension, avoiding subsequent migration characterization deviations caused by temporal misalignment. Furthermore, since the sampling rate of the interface impedance sequence is set to 1 Hz in this embodiment, and the length of a single time window is 1 second, there is one impedance sampling point within a single time window. In this implementation, the normalized impedance mean within the time window and the mean impedance at that time window are... The normalized impedance sample values ​​within the window are the same; the unified representation of the mean within the time window is still used. The purpose is to keep the interface impedance processing flow consistent with the time window processing flow of physiological electrical signals in terms of data structure, so as to facilitate the subsequent completion of multi-source feature alignment and vectorization representation with the same time window number. At the same time, this unified representation also retains compatibility with other alternative implementation methods. In implementations where the interface impedance sequence sampling rate is higher than 1 Hz or the time window length is greater than 1 second, the mean within the time window can correspond to the average value of multiple impedance sampling points, which is used to further smooth impedance fluctuations.

[0024] The average impedance values ​​of the time window at the two acquisition positions on the left and right frontal sides within the frontal region are averaged to obtain the average impedance window value of the frontal region. The calculation formula is as follows: In the formula, This represents the mean value of the pre-load impedance window within the k-th time window. The normalized mean impedance of the left forehead position within the k-th time window. The normalized average impedance of the right forehead position within the k-th time window is used. Averaging the average impedance of the two points in the forehead region can reduce the local error caused by the contact fluctuation of a single electrode, obtain the overall interface impedance change characterization of the forehead region, match the overall contact pressure change characteristics of the forehead support area of ​​the head-mounted device, and avoid misjudgment of the pressing state caused by single-point abnormality. The average impedance values ​​of the time windows at the two acquisition locations (left and right temporal regions) are averaged to obtain the average impedance window value of the temporal region. The calculation formula is as follows: In the formula, The mean value of the temporal impedance window within the k-th time window. The normalized mean impedance of the left temporal location within the k-th time window is given. The normalized average impedance of the right temporal position within the k-th time window is used. By averaging the average impedance of the two points in the temporal region, the symmetrical force characteristics of the bilateral temporal binding areas of the head-mounted device can be matched, eliminating the impedance deviation caused by unilateral head posture fine-tuning, obtaining the overall interface impedance change characterization of the temporal region, and ensuring that the impedance characterization of the frontal region and the temporal region are statistically equivalent.

[0025] Based on the physical relationship that enhanced local contact compression leads to a relative decrease in interface impedance, the compression proxy for the denomination region is constructed according to the average impedance window value of the denomination region. The calculation formula is as follows: In the formula, For the amount of agent compression corresponding to the k-th time window, The average value of the forehead impedance window within the k-th time window is used. The reason for taking a negative value for the average value of the forehead impedance window to construct the compression proxy quantity is that when the contact compression between the electrode and the skin increases, the effective contact area of ​​the electrode-skin interface increases, and the interface impedance will decrease accordingly. The normalized impedance mean will show a negative shift. Taking a negative value can make the magnitude of the compression proxy quantity positively correlated with the contact compression degree. The larger the value, the higher the contact compression degree of the forehead area, which intuitively represents the change in contact pressure in the forehead area and facilitates the subsequent calculation of the migration direction and degree. The temporal region compression surrogate quantity is constructed based on the mean value of the temporal region impedance window, and the calculation formula is as follows: In the formula, The temporal compression proxy corresponding to the k-th time window. The average value of the temporal region impedance window within the k-th time window is used. By taking the negative value of the average value of the temporal region impedance window to construct the compression surrogate quantity, the magnitude of the temporal region compression surrogate quantity can be positively correlated with the degree of temporal region contact compression, which is consistent with the representation logic of the frontal region compression surrogate quantity and ensures that the two are directly comparable. Based on the difference and amplitude relationship between the temporal compression surrogate volume and the frontal compression surrogate volume, the frontotemporal contact pressure migration index, which characterizes the relative migration direction and degree of pressure between the frontal and temporal regions, is calculated. The calculation formula is as follows: In the formula, The frontotemporal contact pressure migration index corresponds to the k-th time window. The temporal compression proxy corresponding to the k-th time window. For the amount of agent compression corresponding to the k-th time window, To prevent extremely small values, a fixed value is used. The numerator uses the difference between the temporal compression surrogate and the frontal compression surrogate, which directly characterizes the direction of contact pressure migration: when the difference is positive, it indicates that the temporal compression is relatively stronger than the frontal compression, and the contact pressure migrates from the frontal to the temporal region; when the difference is negative, it indicates that the frontal compression is relatively stronger than the temporal compression, and the contact pressure migrates from the temporal to the frontal region; when the difference approaches zero, it indicates that there is no relative change in the compression between the frontal and temporal regions, and there is no migration of contact pressure. The denominator uses the sum of the absolute values ​​of the temporal and frontal compression surrogates, plus a minimum value to prevent zero, which normalizes the migration difference and eliminates the influence of baseline differences in compression levels among different subjects and at different collection times, allowing the migration index to be compared across individuals and time periods. The migration index is constructed by combining the difference and the normalized amplitude, which can simultaneously characterize the relative migration direction and relative migration degree of contact pressure, distinguishing both the bidirectional trend of pressure migration and quantifying the intensity of migration.

[0026] It should be noted that the forehead compression surrogate quantity refers to a characterization quantity constructed based on the statistical results of the normalized impedance sequence of the corresponding acquisition position in the forehead region within a time window. It is used to reflect the relative compression degree of the head-mounted acquisition device at the forehead-skin contact interface. Since an increase in the compression degree of the forehead contact interface leads to an increase in the effective contact area between the electrode and the skin, and a more stable contact state, the interface impedance decreases accordingly. Therefore, the change in normalized impedance can be converted into a surrogate representation positively correlated with the compression degree, thus describing the change in forehead contact pressure in a calculable manner. The temporal compression surrogate quantity refers to a characterization quantity constructed based on the statistical results of the normalized impedance sequence of the corresponding acquisition position in the temporal region within a time window. It is used to reflect the relative compression degree of the head-mounted acquisition device at the temporal-skin contact interface. Since the lateral compression change of the temporal region constriction structure directly affects the contact state and interface impedance level between the electrode and the skin, the change in normalized impedance can be used to indirectly characterize the temporal region compression degree, allowing the temporal region contact pressure change to participate in subsequent migration analysis in a unified numerical form. It should be noted that the frontotemporal contact pressure migration index is a characterization quantity constructed based on the relative difference between the frontal and temporal compression surrogate quantities. It is used to describe the direction and degree of contact pressure migration between the frontal and temporal regions. When the index increases in one direction, it indicates that the contact pressure migrates relatively towards the corresponding region and increases the compression degree of that region. When the index is close to zero, it indicates that the relative compression relationship between the frontal and temporal regions is basically stable. This index can transform the contact pressure redistribution process induced by stress between the frontotemporal region into a quantitative result that can be continuously analyzed.

[0027] In one embodiment of the present invention, the frontotemporal reverse linkage coefficient is calculated based on a normalized impedance sequence; the original psychoneural activity load is constructed based on the total energy of the frontal region and the total energy of the temporal region; a feature vector is formed by combining the original psychoneural activity load, the frontotemporal contact pressure migration index, and the frontotemporal reverse linkage coefficient; the feature vector is then used for state identification to obtain a pressure migration state sequence, including: For each time window to be calculated, starting from the current time window as the endpoint, four consecutive adjacent time windows are selected forward, forming a local analysis window together with the current time window. The total length of the local analysis window is five consecutive non-overlapping time windows, corresponding to a total duration of 5 seconds. Using a unidirectional sliding local analysis window with the current time window as the endpoint adapts to the real-time detection requirements of head-mounted acquisition scenarios. Simultaneously, the 5-second window length matches the duration characteristics of frontotemporal contact pressure migration under stress, effectively suppressing correlation coefficient calculation errors caused by random impedance fluctuations in a single time window, while ensuring the detection sensitivity of the pressure migration linkage relationship and preventing the smoothing of linkage change characteristics caused by excessively long windows. It should also be noted that in this embodiment, the local analysis window uses five consecutive time windows... The purpose of setting the time windows is not to model long-term trends, but to capture the short-term linkage process of frontotemporal contact pressure migration in real-time detection scenarios. Frontotemporal contact pressure migration is a short-term continuous process formed by low-amplitude facial tension changes, local force redistribution of the head-mounted device, and changes in electrode-skin interface impedance. It usually maintains a relatively consistent migration direction and linkage relationship within a few seconds. Using five consecutive time windows to construct the local analysis window can, on the one hand, form a minimum calculable local impedance mean sequence without introducing future time data, meeting the real-time detection requirements; on the other hand, five consecutive points are sufficient to distinguish between random single-point perturbations and frontotemporal reverse linkage changes with continuous directionality, thus providing a stable local sequence basis for the Pearson correlation coefficient. For implementations requiring higher smoothness, the local analysis window can be extended to six to ten consecutive time windows to further improve the robustness of the linkage coefficient without changing the core principle of the invention.

[0028] After constructing the local analysis window, within that window, the mean value of the frontage impedance window corresponding to each time window is extracted in chronological order, forming a frontage impedance mean value sequence. The formula for calculating the frontage impedance mean value sequence is as follows: In the formula, This is the sequence of average impedance values ​​in the local analysis window corresponding to the k-th time window; This represents the mean value of the approximate impedance window corresponding to the kn-th time window. Take integers from 0 to 4; construct a sequence using the average impedance window value after averaging two points in the front area. This can continue the overall regional characterization logic of the aforementioned steps, eliminate the sequence fluctuations caused by single electrode contact anomalies, and ensure that the sequence can accurately reflect the overall interface impedance change trend of the front area and match the overall change characteristics of the front area contact pressure. Within the same local analysis window, the mean temporal impedance window value corresponding to each time window is extracted in chronological order to form a temporal impedance mean value sequence. The formula for calculating the temporal impedance mean value sequence is as follows: In the formula, This is the sequence of mean temporal impedance values ​​within the local analysis window corresponding to the k-th time window; The mean value of the temporal impedance window corresponding to the kn-th time window, where n is an integer from 0 to 4; the sequence is constructed by averaging the impedance window values ​​of two points in the temporal region, which can ensure that the temporal impedance change sequence and the frontal impedance change sequence are equivalent in statistical dimension and representation logic, eliminate the sequence bias caused by the asymmetrical force on both sides of the temporal region, reflect the overall interface impedance change trend of the temporal region, and provide equivalent input data for subsequent calculation of the reverse linkage relationship; Negating each element in the temporal impedance mean sequence yields the negative temporal impedance mean sequence, calculated using the following formula: In the formula, This represents the negative temporal impedance mean sequence corresponding to the k-th time window. The reason for negativening the temporal impedance mean sequence is that the frontotemporal contact pressure migration is characterized by the inverse relationship between the contact pressure in the frontal and temporal regions. The corresponding interface impedance changes show an inverse linkage relationship, that is, when the frontal region's compression increases, the frontal region's impedance decreases, and when the temporal region's compression decreases, the temporal region's impedance increases. By negativening the temporal impedance sequence, the original inverse linkage relationship can be transformed into a positive correlation relationship. The strength of the inverse linkage between the two can be directly quantified by calculating the correlation coefficient, making the calculation logic more intuitive and the results more directly interpretable.

[0029] Calculate the Pearson correlation coefficient between the mean frontal impedance sequence and the negativeed mean temporal impedance sequence. Use this correlation coefficient as the frontotemporal inverse linkage coefficient, characterizing the inverse relationship between frontal and temporal impedance changes. The formula for calculating the Pearson correlation coefficient is: In the formula, This is the frontotemporal inverse linkage coefficient corresponding to the k-th time window; Mean impedance sequence of the pre-region The (n+1)th element in; The mean temporal impedance sequence after taking the negative value. The (n+1)th element in; Mean impedance sequence of the pre-region The arithmetic mean of all elements in the set; The mean temporal impedance sequence after taking the negative value. The arithmetic mean of all elements in the sequence; the Pearson correlation coefficient can accurately quantify the degree of linear correlation between two sequences; this calculation method can directly quantify the linkage relationship into a fixed range of values, which is convenient for subsequent feature vector construction and state recognition, and can effectively distinguish between systematic impedance changes caused by pressure migration and independent impedance fluctuations caused by single-electrode random noise; It should be noted that the frontotemporal inverse linkage coefficient is a characterization quantity constructed based on the normalized impedance change relationship between the frontal and temporal regions. It is used to reflect whether the contact state between the frontal and temporal regions exhibits a linkage change with each other within the same time period. When the contact compression of the frontal region increases while the contact compression of the temporal region decreases, or when the contact compression of the temporal region increases while the contact compression of the frontal region decreases, the impedance changes of the frontal and temporal regions will show opposite trends. The frontotemporal inverse linkage coefficient is used to quantitatively describe the strength of this inverse coordinated change, thereby helping to determine whether the frontotemporal contact pressure migration has continuity and structure, rather than being a local anomaly caused by a single point random fluctuation.

[0030] For each time window, an energy contrast value is constructed based on the ratio of total energy in the frontal region to total energy in the temporal region. The formula for calculating the energy contrast value is as follows: In the formula, This represents the energy comparison value corresponding to the k-th time window. This represents the total energy of the pre-region corresponding to the k-th time window. The total energy in the temporal region corresponds to the k-th time window. The physiological electrical activity in the frontal region is directly related to the cognitive load, alertness level, and psychological stress state of the subject. It is the core representation area of ​​psychoneural activity. The physiological electrical activity in the temporal region is less directly affected by the target psychoneural activity and can be used as a stable physiological reference area. The ratio calculation can eliminate the systematic bias caused by individual differences in the amplitude of basic physiological electrical signals, differences in the channel gain of the acquisition system, and environmental common-mode interference, and realize the standardized representation of the intensity of electrical activity in the frontal region relative to the temporal region. Logarithmic transformation of the energy contrast quantity yields the original psychoneural activity load, calculated using the following formula: In the formula, This represents the original psychoneural activity load corresponding to the k-th time window; For natural logarithm operations, This represents the energy contrast value corresponding to the k-th time window. The energy contrast value exhibits a positively skewed distribution, and the change in the ratio is asymmetric, meaning that the ratio values ​​corresponding to a doubling or decreasing of energy in the frontal region relative to the temporal region differ significantly. The natural logarithmic transformation converts the multiplicative change in the ratio into an additive change, making the data distribution closer to a normal distribution. Simultaneously, it transforms the positive and negative changes in the ratio into symmetrical positive and negative values: when the energy in the frontal region is stronger than that in the temporal region, the original psychoneural activity load is positive; when the energy in the frontal region is weaker than that in the temporal region, the original psychoneural activity load is negative. This provides an intuitive and symmetrical representation of the relative change trend of psychoneural activity, adapting to the subsequent needs of change calculation and state recognition. It should be noted that the raw psychoneural activity load is a quantitative result constructed based on the relative relationship between the total energy in the frontal region and the total energy in the temporal region, used to characterize the original strength of psychoneural activity within the current time window. It essentially reflects the change in the physiological electrical activity level of the frontal region relative to the temporal region, and is used to describe the initial representation of the subject's cognitive load, alertness level, or psychological stress-related activities at the current moment. It is called the raw psychoneural activity load because this quantity has not yet eliminated the spurious effects caused by the interface impedance changes and regional signal distribution shifts caused by the frontotemporal contact pressure migration. Therefore, it includes the real psychoneural activity component as well as the interference component introduced by the contact pressure migration.

[0031] The change in the frontotemporal contact pressure migration index between adjacent time windows is extracted, and the calculation formula is as follows: In the formula, This represents the change in the frontotemporal contact pressure migration index corresponding to the k-th time window. The frontotemporal contact pressure migration index corresponds to the preceding adjacent time window of the k-th time window. For the first time window of the sequence when k=1, there is no preceding adjacent time window. The difference between the frontotemporal contact pressure migration index of the first time window and the median of the frontotemporal contact pressure migration index of all time windows in the reference baseline stage is used as the corresponding change. The first-order difference can accurately characterize the temporal change rate and direction of frontotemporal contact pressure migration, capture the dynamic process of pressure migration, eliminate the static offset effect caused by individual baseline differences, improve the sensitivity of the recognition of the start and end of pressure migration, and provide dynamic feature support for state recognition. The change in the original psychoneural activity load between adjacent time windows is extracted synchronously, and the calculation formula is as follows: In the formula, This represents the change in the original psychoneural activity load corresponding to the k-th time window. The original psychoneural activity load corresponds to the preceding adjacent time window of the k-th time window. For the first time window of the sequence when k=1, there is no preceding adjacent time window. The difference between the original psychoneural activity load of the first time window and the median of the original psychoneural activity load of all time windows in the reference baseline stage is used as the corresponding change. Using the first difference can effectively capture the instantaneous fluctuations and trends of psychoneural activity, eliminate the influence of slow baseline drift, distinguish between the steady-state level and dynamic changes of psychoneural activity, and form a temporal correspondence with the change of the migration index, which is convenient for distinguishing between real psychoneural activity changes and pseudo changes caused by stress migration in subsequent state identification.

[0032] The original psychoneural activity load, the frontotemporal contact pressure migration index, the frontotemporal reverse linkage coefficient, the change in the frontotemporal contact pressure migration index, and the change in the original psychoneural activity load are collectively composed of a feature vector. The formula for combining the feature vector is: In the formula, This is the five-dimensional feature vector corresponding to the k-th time window. The five features cover five dimensions: the steady-state level of psychoneural activity, the static degree of stress migration, the linkage characteristics of stress migration, the dynamic rate of change of stress migration, and the dynamic rate of change of psychoneural activity. It can accurately characterize the core pattern of frontotemporal contact stress migration and the changing characteristics of psychoneural activity. The combination of multi-dimensional features can effectively improve the accuracy of subsequent state recognition and avoid misjudgment caused by a single feature.

[0033] A pre-trained Hidden Markov Model (HMM) is used to classify the temporal state of the feature vector sequence. The core structure and state definition of the HMM are predefined, and the number of hidden states is fixed at 4, corresponding to the stable contact state, frontotemporal migration state, temporal-frontotemporal migration state, and migration transition state, respectively. The dimension of the observation vector is fixed at 5, matching the dimension of the feature vector. The state changes of frontotemporal contact pressure migration have strong temporal dependence. The pressure migration state of the current time window is determined by the state of the previous time moment, rather than an independent random event. The HMM can effectively capture the state transition patterns in the temporal sequence and has strong robustness to noise in the observed features. Compared with the static classification method with a single time window, it can significantly improve the continuity and accuracy of state classification and avoid misjudgment caused by feature fluctuations in a single time window.

[0034] The three core parameters of a Hidden Markov Model are defined as the initial state probability vector, the state transition probability matrix, and the observation probability distribution; the formula for calculating the initial state probability vector is: In the formula, This is the initial state probability vector; The probability that the first time window of the sequence is in a stable, fitted state; The probability that the first time window of the sequence is in a frontotemporal migration state; The probability that the first time window of the sequence is in a temporal-to-frontal migration state; This represents the probability that the sequence is in a transitional state during its first time window. The initial value of the initial state probability vector is set to... The reason for setting this initial value is that the initial stage of the acquisition sequence is when the subject is in a resting state, and the head-mounted device is in a stable fit with the skin. The probability of pressure migration is extremely low. Assigning a high initial probability to the stable fit state can ensure the accuracy of the initial state classification, which is in line with the actual process characteristics of head-mounted acquisition. The formula for calculating the state transition probability matrix is: In the formula, It is a 4x4 state transition probability matrix. Let i be the probability of transitioning from the i-th hidden state to the j-th hidden state. i and j are integers from 1 to 4, corresponding to the stable fitting state, the fronto-temporal migration state, the temporal-frontal migration state, and the transition state, respectively. The initial values ​​of the state transition probability matrix are set according to the physical logic of state transitions: the probability of transitioning from the stable fit state to itself is set to 0.92, the probability of transitioning to the transitional state is set to 0.08, and the direct transition probability to the fronto-temporal and temporal-frontostate transition states is set to 0; the probability of transitioning from the fronto-temporal transition state to itself is set to 0.85, the probability of transitioning to the transitional state is set to 0.15, and the direct transition probability to other states is set to 0; the probability of transitioning from the temporal-frontostate transition to itself is set to 0.85, the probability of transitioning to the transitional state is set to 0.15, and the direct transition probability to other states is set to 0; the probability of transitioning from the transitional state to itself is set to 0.6, the probability of transitioning to the stable fit state is set to 0.3, and the transition probabilities to the fronto-temporal and temporal-frontostate transition states are set to 0.05 respectively. The reason for setting this initial transition probability is that the state change of frontotemporal contact pressure migration is a continuous process. It cannot jump directly from a stable contact state to a strong migration state. It must go through a migration transition state. At the same time, the stable contact state and the strong migration state have strong self-maintenance and it is difficult for frequent state jumps to occur. This initial setting matches the physical characteristics of pressure migration and can effectively constrain the rationality of state transition. The observation probability distribution adopts a single Gaussian distribution, and the five-dimensional observation vector corresponding to each hidden state follows an independent Gaussian distribution. The calculation formula is as follows: In the formula, Let be the observation probability density function corresponding to the j-th hidden state; The input is a five-dimensional feature vector; Let be the five-dimensional observation mean vector corresponding to the j-th hidden state; Let be the 5x5 observation covariance matrix corresponding to the j-th hidden state; For exponentiation; superscript is the matrix transpose operator; superscript -1 is the matrix inversion operator; the reason for using a single Gaussian distribution as the observation probability distribution is that the distribution of the five-dimensional feature vector in the corresponding hidden state has a unimodal characteristic. The single Gaussian distribution can accurately fit the feature distribution in each state. At the same time, the model structure is simple, the training and inference computation is small, and it is suitable for the real-time detection requirements of head-mounted acquisition devices.

[0035] After defining the model structure and initial parameters, the Baum-Welch algorithm was used to perform unsupervised pre-training of the Hidden Markov Model. The pre-training dataset consisted of model training data from 30 historically collected healthy subjects in a head-mounted imaging scenario. This model training data was independent of the current subject's single detection process. The model training data covered the resting stable fit phase, the artificially induced frontotemporal pressure migration phase, the artificially induced temporal-frontal pressure migration phase, and the pressure migration transition phase. The training time for each healthy subject was no less than 10 minutes, and the number of iterations was set to 100. The iteration convergence threshold was set to a preset log-likelihood change threshold. When the number of iterations reached the upper limit or the change in log-likelihood was less than the preset log-likelihood change threshold, the training was terminated. Training is stopped, resulting in a pre-trained Hidden Markov Model (HMM). A preset log-likelihood change threshold is used to determine whether the HMM training has converged. When the change in log-likelihood between two consecutive iterations is less than 0.0001, the model training is considered to have reached the convergence condition. The Baum-Welch algorithm is used as the standard unsupervised training algorithm for the HMM, which can automatically optimize the core parameters of the model based on the observation sequence without the need for manual labeling of the state in each time window, thus reducing the labeling cost of training data. The calibration data collection segment of the current subject in a single detection process is used to fit the state coefficients corresponding to the current subject and is not used to retrain the HMM, thereby distinguishing the historical model pre-training process from the current subject's individualized state coefficient fitting process.

[0036] After model training is complete, the feature vector sequence arranged chronologically during the data collection process is input into the trained Hidden Markov Model. The Viterbi algorithm is used to solve for the optimal state path, obtaining the pressure migration state corresponding to each time window. First, the forward probability variable is defined. , representing the maximum forward probability that the k-th time window is in the j-th hidden state, is calculated using the following formula: In the formula, The maximum forward probability of the first time window being in the j-th hidden state; This is the j-th element in the initial state probability vector; The feature vector of the first time window The observation probability in the j-th hidden state; For a time window k>1, the iterative formula for calculating the forward probability is: In the formula, The maximum forward probability of being in the j-th hidden state during the k-th time window; This is for calculating the maximum value. The maximum forward probability of being in the i-th hidden state during the (k-1)-th time window; Let be the probability of transitioning from the i-th hidden state to the j-th hidden state; The feature vector of the k-th time window The observation probability in the j-th hidden state; Define a backtrack pointer variable Record the index of the optimal state at the previous time when the k-th time window is in the j-th hidden state. The calculation formula is: In the formula, This is the backtracking pointer for the j-th hidden state corresponding to the k-th time window; This is an operation that takes the index corresponding to the maximum value.

[0037] After completing the forward probability iteration calculation for all time windows, starting from the last time window, the optimal state sequence is solved in reverse using a backtracking pointer. First, the optimal state for the last time window is determined using the following formula: In the formula, K is the total number of time windows in the acquisition sequence; Number the optimal state corresponding to the last time window; Starting from the (K-1)th time window, backtrack to find the optimal state for each time window. The calculation formula is as follows: In the formula, The optimal state number corresponding to the k-th time window; This is the backtracking pointer value for the optimal state corresponding to the (k+1)th time window. The Viterbi algorithm can solve the globally optimal hidden state sequence based on the entire observation sequence, rather than the local optimal classification of a single time window. This can ensure the continuity and global rationality of the state sequence, effectively suppress the misjudgment of the state caused by the feature noise of a single time window, and adapt to the needs of time-series state classification.

[0038] Each time window's state number is mapped to a corresponding pressure migration state: state number 1 corresponds to a stable fit state, state number 2 corresponds to a frontotemporal migration state, state number 3 corresponds to a temporal-frontal migration state, and state number 4 corresponds to a transitional migration state, resulting in a pressure migration state sequence arranged according to the time window. Specifically, the stable fit state corresponds to a scenario where the head-mounted device maintains stable contact with the skin in the frontotemporal region, with no significant pressure migration; the frontotemporal migration state corresponds to a scenario where contact pressure shifts from the frontal region to the temporal region, with a significant increase in temporal pressure relative to the frontal region; the temporal-frontal migration state corresponds to a scenario where contact pressure shifts from the temporal region to the frontal region, with a significant increase in frontal pressure relative to the temporal region; and the transitional migration state corresponds to a scenario where contact pressure is in a transitional phase between a stable state and a strong migration state, with no stable migration direction.

[0039] Furthermore, the pre-trained Hidden Markov Model (HMM) is used to provide the initial state space structure, initial transition logic, and initial observation distribution constraints for identifying the current subject's stress migration state. Before entering the formal task execution phase for the current subject, the feature vector sequence obtained by the current subject during the reference baseline phase and the calibration data acquisition phase is input into the pre-trained HMM to obtain the initial state identification result corresponding to the current subject. Subsequently, using the initial state identification result as the basis for individualized state allocation, the feature vectors of the current subject in each time window within the calibration data acquisition phase are classified into states, forming individual calibration sample sets corresponding to the frontotemporal migration state, temporal-frontal migration state, migration transition state, and stable fit state. Preferably, while keeping the number of latent states and state definitions unchanged, an individual adaptive update can be performed on the observation mean vector and observation covariance matrix in the pre-trained model using the reference baseline stage and calibration data acquisition segment data of the current subject. This makes the state observation distribution more consistent with the current subject's facial structure, skin impedance characteristics, and differences in the tightness of the head-mounted device. In embodiments without individual adaptive updates, the pre-trained model can be used directly to complete state recognition, and the recognition results can then be used to provide individual state labels for state coefficient fitting. Through the above processing, a closed data flow relationship is formed between the pre-trained model, current subject state recognition, individual calibration sample division, and state coefficient fitting. That is, the pre-trained model first provides state recognition capability, and the state recognition results are then used as the grouping basis for the current subject's linear regression fitting, thereby ensuring that the fitting object of the state coefficients is consistent with the state classification results in definition.

[0040] In one embodiment of the present invention, a migration interference component is constructed based on a stress migration state sequence, and the original psychoneural activity load is corrected according to the migration interference component to obtain the corrected psychoneural activity load, including: The stress migration states are categorized, with fronto-temporal migration, temporal-frontal migration, and transitional migration states all grouped into migration-related states. Stable fit states are classified as interference-free states. A state-specific directional correction logic is used to remove interference from the original psychoneural activity load. This classification correction logic allows correction to be performed only during periods of stress migration interference, while preserving the original psychoneural activity representation in the interference-free stable fit state. This avoids excessive smoothing and loss of effective information caused by uniform correction throughout the entire time period, maximizing the preservation of real physiological changes and improving the reliability of the detection results.

[0041] For each type of migration-related state, corresponding state coefficients are fitted. These state coefficients include a first fitting coefficient and a second fitting coefficient, which correspond to the weights of the two components of the migration interference component, respectively. For the current subject, during the calibration data acquisition phase after the reference baseline stage, three standardized pressure migration induction tasks are set up, corresponding to the frontotemporal migration state, the temporal-frontal migration state, and the migration transition state, respectively. Each task lasts for 30 seconds, with a 60-second rest recovery period between tasks to ensure that the contact state between the head-mounted device and the skin returns to a stable fit. During the calibration data acquisition phase, physiological electrical signals and interface impedance sequences of each channel are simultaneously acquired. Following the aforementioned steps, the original psychoneural activity load, frontotemporal contact pressure migration index, and migration index change for the corresponding time window are calculated. At the same time, the corresponding states are labeled using the aforementioned hidden Markov model to obtain the calibration dataset corresponding to each type of migration-related state. Since there are individual differences in facial structure, head-mounted device tightness, and skin impedance characteristics among different subjects, the state coefficients fitted based on individual calibration data can accurately match the individual's pressure migration and signal interference mapping relationship, eliminating correction errors caused by individual differences.

[0042] For each type of migration-related state ,in Linear regression models were constructed to correspond to the frontotemporal migration state, the temporal-frontal migration state, and the transitional migration state, respectively. The original psychoneural activity load was used as the target variable, and the frontotemporal contact pressure migration index and the change in the migration index were used as independent variables. The formula for calculating the linear regression model is as follows: In the formula, k represents the state. Time window number; For state The corresponding first fitting coefficient; For state The corresponding second fitting coefficient; The residual term represents the true psychoneural activity component in the original psychoneural activity load that is unaffected by stress migration. The reason for using a linear regression model to construct the mapping relationship is that the interference of stress migration on the original load exhibits a linear correlation characteristic, that is, the higher the degree of stress migration and the faster the migration rate, the greater the pseudo-influence on the load. The linear model can accurately capture this linear mapping relationship. At the same time, the model structure is simple, the fitting calculation is small, it is suitable for real-time detection requirements, and it is highly interpretable, making it easy to distinguish between interference components and true signal components. The optimal first and second fitting coefficients are obtained using the least squares method, and the solution formula is as follows: In the formula, It is the set of all calibration time windows belonging to state s; The operation is to select the value of the independent variable that minimizes the objective function; the objective function is the sum of squares of the fitting residuals; the least squares method can minimize the sum of squares of the fitting residuals, ensuring that the linear model obtained by fitting can explain the changes in the original psychoneural activity load caused by stress migration to the greatest extent, thereby accurately separating the interference components from the real signal components. The solution process has a fast convergence speed, strong numerical stability, and is suitable for fitting individual calibration data.

[0043] After completing the state coefficient fitting, for each time window to be processed in the acquisition process, first determine the pressure migration state corresponding to the current time window; when the state corresponding to the current time window is a migration-related state, construct a migration disturbance component based on the fitting coefficient corresponding to the current state. The calculation formula for the migration disturbance component is: In the formula, This is the migration interference component corresponding to the k-th time window; the migration interference component consists of two parts: the first component is... and The product of these components, constructed based on the frontotemporal contact pressure migration index, is used to characterize the fixed pseudo-influence of the steady-state degree of pressure migration on the original load; the second component is... and The product of the frontotemporal contact pressure migration index, constructed based on the change in adjacent time windows, is used to characterize the transient pseudo-effect of the dynamic change rate of pressure migration on the original load. The interference of pressure migration on physiological signals includes both steady-state interference and transient interference. Steady-state interference is caused by the continuous difference in pressure distribution, while transient interference is caused by the dynamic change process of pressure migration. The dual-component structure can cover both types of interference and has higher interference stripping accuracy than the single-component structure, avoiding interference residue caused by single-dimensional correction. It should be noted that the migration interference component refers to a quantitative component constructed based on the pressure migration state sequence and the corresponding frontotemporal contact pressure migration characteristics, used to characterize the spurious effects of frontotemporal contact pressure migration on the original psychoneural activity load. It does not reflect the subject's actual psychoneural activity itself, but rather the additional disturbance introduced by the head-mounted acquisition device when contact pressure redistribution occurs between the frontal and temporal regions, due to changes in interface impedance, changes in electrode contact state, and regional signal distribution shift. The migration interference component is used to describe this non-target change component caused by contact pressure migration, and is subsequently corrected for difference with the original psychoneural activity load, thereby separating the real psychoneural activity changes from the spurious changes caused by pressure migration as much as possible.

[0044] The original psychoneural activity load is differentially corrected based on the migration interference component to obtain the corrected psychoneural activity load. The correction calculation formula is as follows: In the formula, is the corrected psychoneural activity quantity corresponding to the k-th time window; the migration interference component is the superimposed pseudo-change in the original psychoneural activity load quantity caused by stress migration. By directly subtracting the interference component, the pseudo-effects caused by stress migration can be accurately removed, and the real psychoneural activity components in the original load quantity can be retained.

[0045] When the current time window corresponds to a stable contact state, it is determined that there is no significant frontotemporal contact pressure migration interference in the current time period. Therefore, no correction operation is required, and the original psychoneural activity load is directly used as the corrected psychoneural activity load. In the stable contact state, the contact state between the electrode and the skin is stable, and there are no interface impedance changes or spurious changes in signal energy distribution caused by pressure migration. The original psychoneural activity load can truly reflect the psychoneural activity state of the subject. Directly retaining the original value can avoid unnecessary correction operations that could damage the real signal and ensure the authenticity of the test results.

[0046] In one embodiment of the present invention, the corrected psychoneural activity level is compared with a reference baseline to obtain a relative offset, and the psychoneural activity detection result is determined based on the relative offset, including: Extract the corrected psychoneural activity levels corresponding to each time window within the reference baseline phase to form a reference baseline sequence. Calculate the statistical central value of the reference baseline sequence and use this statistical central value as the baseline reference value. The statistical central value is calculated using the median, and the calculation formula is as follows: In the formula, Used as a baseline reference value; This is a corrected sequence of psychoneural activity within the baseline phase; The reference baseline phase includes the total number of time windows. Using the median as the statistical center value can effectively suppress occasional electrode contact fluctuations and outlier interference caused by minor movements of the subjects within the reference baseline phase. Compared with the arithmetic mean, it can avoid the influence of extreme values ​​on the baseline reference value, ensuring that the baseline reference value can accurately reflect the basic psychoneural activity level of the subjects and improve the stability and accuracy of subsequent relative offset calculations. The offset of the corrected psychoneural activity level within the current time window relative to the baseline reference value is used as the relative offset, calculated using the following formula: In the formula, This represents the relative offset corresponding to the k-th time window. By using the difference between the current corrected psychoneural activity level and the baseline reference value, the differences in the individual baseline psychoneural activity levels of the subjects can be eliminated. The corrected psychoneural activity levels of different subjects can be converted into relative changes relative to their own baseline, achieving a unified judgment standard across individuals. At the same time, it can intuitively represent the direction and magnitude of the current psychoneural activity offset relative to the baseline level: a positive value represents that the current psychoneural activity level is higher than the baseline level, a negative value represents that the current psychoneural activity level is lower than the baseline level, and a zero value represents that it is consistent with the baseline level.

[0047] A baseline offset sequence is constructed based on the difference between the corrected psychoneural activity level and the baseline reference value for each time window within the reference baseline phase. Each element in the baseline offset sequence is the offset value obtained by subtracting the baseline reference value from the corrected psychoneural activity level for the corresponding time window within the reference baseline phase. An upper and lower decision limit are determined based on the distribution range of the baseline offset sequence. The upper decision limit is the upper quartile of the baseline offset sequence, and the lower decision limit is the lower quartile of the baseline offset sequence. The reason for using the baseline offset sequence to determine the upper and lower decision limits is that the relative offset of the current time window is the offset of the corrected psychoneural activity level within the current time window relative to the baseline reference value, which is an offset. Accordingly, the upper and lower decision limits should also be determined based on the offset distribution within the reference baseline period, so that the relative offset of the current time window and the decision limits are on the same numerical scale, thereby avoiding the problem of inconsistent decision scale caused by directly comparing the offset with the original value of the corrected psychoneural activity level. The quartiles are calculated based on the actual distribution of the baseline offset sequence and are robust statistics, unaffected by extreme outliers within the reference baseline period. The interval between the upper and lower quartiles represents the normal offset fluctuation range of the subject relative to the baseline reference value within the reference baseline period. Relative offsets exceeding this interval can be judged as statistically significant changes.

[0048] A tiered judgment is performed on the relative offset of each time window: when the relative offset is greater than the upper judgment threshold, a detection result representing enhanced psychoneural activity is obtained, corresponding to a psychoneural activity state in which the subject experiences increased cognitive load, heightened alertness, or psychological stress response; when the relative offset is less than the lower judgment threshold, a detection result representing weakened psychoneural activity is obtained, corresponding to a psychoneural activity state in which the subject experiences decreased cognitive load, reduced alertness, or relaxation; when the relative offset is between the upper and lower judgment thresholds, a detection result representing stable psychoneural activity is obtained, corresponding to a psychoneural activity in which the subject's psychoneural activity is within the normal offset fluctuation range of the baseline level, without significant enhancement or weakening changes. The three-level tiered judgment logic can achieve adaptive judgment based on the actual offset distribution of the individual baseline, adapting to individual differences among different subjects. At the same time, the judgment logic is simple and clear, and can directly output interpretable detection results, adapting to the real-time assessment needs of scenarios such as immersive training and emergency decision-making drills.

[0049] In one optional implementation, when simultaneously acquiring electrocardiogram (ECG) signals and / or blood pressure parameters, after obtaining detection results of enhanced, weakened, or stable psychoneural activity, auxiliary physiological prompts can be further output based on the ECG signals and / or blood pressure parameters. Specifically, when the heart rate relative to the reference baseline stage increases by more than a preset heart rate fluctuation range, or the heart rate variability is lower than the preset lower limit corresponding to the reference baseline stage, or the blood pressure parameter deviates from the blood pressure statistical center value relative to the reference baseline stage by more than a preset blood pressure fluctuation range, circulatory physiological fluctuation prompts are output. The circulatory physiological fluctuation prompts are used to assist in interpreting the current psychoneural activity detection results and do not change the psychoneural activity detection results determined based on the relative deviation. The heart rate statistical center value, heart rate variability reference value, and blood pressure statistical center value are considered as reference values ​​for the ECG signals and / or blood pressure parameters. The calculated center values ​​are all determined based on the time window sequence of the corresponding parameters within the reference baseline period. The preset heart rate fluctuation range, preset blood pressure fluctuation range, and preset lower limit corresponding to heart rate variability can be determined based on the quartile range of the corresponding parameters within the reference baseline period, or based on the mean and standard deviation of the corresponding parameters within the reference baseline period. When using the quartile range for determination, the interval between the upper and lower quartiles of the corresponding parameters within the reference baseline period is taken as the normal fluctuation range, and the lower quartile of the heart rate variability parameter is taken as the preset lower limit corresponding to heart rate variability. When using the mean and standard deviation for determination, the interval formed by the upper and lower standard deviations of the mean of the corresponding parameters within the reference baseline period is taken as the normal fluctuation range, and the value obtained by subtracting the preset standard deviation from the mean of the heart rate variability parameter is taken as the preset lower limit corresponding to heart rate variability.

[0050] It should be further clarified that the increased, decreased, and stable psychoneural activity output in this embodiment are all grading results of the relative physiological changes of the current subject during this testing process, and not subjective psychological diagnostic conclusions given independently of the testing scenario. Increased psychoneural activity refers to a significant positive shift in the corrected psychoneural activity level relative to the reference baseline, indicating that the current frontal region's physiological electrical activity intensity exceeds the normal fluctuation range of the baseline, corresponding to a trend of increased cognitive load, increased alertness, or increased psychological stress in the current task stage. Decreased psychoneural activity refers to a significant negative shift in the corrected psychoneural activity level relative to the reference baseline, indicating that the current frontal region's physiological electrical activity intensity decreases beyond the normal fluctuation range of the baseline, corresponding to a trend of decreased cognitive load, decreased alertness, or relaxation in the current task stage. Stable psychoneural activity means that the corrected psychoneural activity level is within the normal fluctuation range defined by the interquartile range of the reference baseline, indicating that the current subject's psychoneural activity level has not deviated significantly from the individual's baseline level. Because this invention identifies and eliminates the interface spurious effects caused by stress-locked frontotemporal contact pressure migration before outputting the grading results, the corrected psychoneural activity level used for grading, compared to the uncorrected original psychoneural activity load, better reflects the subject's true relative psychoneural activity changes. The three-level output conclusion of this invention is based on the combined effects of individual baseline reference, comparison of relative physiological electrical activity in the frontotemporal region, and directional correction for migration interference. Its technical meaning is an objective grading of the relative trend of psychoneural activity changes within the current testing process, rather than a medical diagnosis of a specific mental illness or psychological disorder.

[0051] Example 2, as Figure 2 As shown, the present invention proposes a comprehensive physiological and electrocardiogram (ECG) detection system for assessing psychoneural activity, used to execute a comprehensive physiological and ECG detection method for assessing psychoneural activity as described in Embodiment 1, comprising: The data processing module is used to collect physiological electrical signals and interface impedance sequences of the frontal and temporal regions, normalize the interface impedance sequences to obtain normalized impedance sequences, and extract the total energy of the frontal and temporal regions according to time windows. The index construction module is used to construct the frontal and temporal compression surrogate quantities based on the normalized impedance sequence, and to calculate the frontotemporal contact pressure migration index based on the frontal and temporal compression surrogate quantities. The state recognition module is used to calculate the frontotemporal reverse linkage coefficient based on the normalized impedance sequence, construct the original psychoneural activity load based on the total energy of the frontal region and the total energy of the temporal region, form a feature vector by combining the original psychoneural activity load, the frontotemporal contact pressure migration index and the frontotemporal reverse linkage coefficient, and perform state recognition on the feature vector to obtain the pressure migration state sequence. The load correction module is used to construct migration interference components based on the pressure migration state sequence, and to correct the original psychoneural activity load based on the migration interference components to obtain the corrected psychoneural activity. The evaluation output module compares the corrected psychoneural activity level with the reference baseline to obtain the relative offset, determines the psychoneural activity detection result based on the relative offset, generates auxiliary physiological prompts based on ECG-related auxiliary physiological parameters, and combines the psychoneural activity detection result and auxiliary physiological prompts as a comprehensive detection output.

[0052] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A comprehensive physiological and electrocardiographic method for assessing psychoneural activity, characterized in that, Includes the following steps: S1. Collect physiological electrical signals and interface impedance sequences of the frontal and temporal regions, normalize the interface impedance sequences to obtain normalized impedance sequences, and extract the total energy of the frontal and temporal regions according to the time window. S2. Construct the frontal and temporal compression surrogate quantities based on the normalized impedance sequence, and calculate the frontotemporal contact pressure migration index based on the frontal and temporal compression surrogate quantities. S3. Calculate the frontotemporal reverse linkage coefficient based on the normalized impedance sequence, construct the original psychoneural activity load based on the total energy of the frontal region and the total energy of the temporal region, form a feature vector based on the combination of the original psychoneural activity load, the frontotemporal contact pressure migration index and the frontotemporal reverse linkage coefficient, and perform state recognition on the feature vector to obtain the pressure migration state sequence. S4. Construct migration interference components based on the stress migration state sequence, and correct the original psychoneural activity load based on the migration interference components to obtain the corrected psychoneural activity load. S5. Compare the corrected psychoneural activity level with the reference baseline to obtain the relative offset, determine the psychoneural activity detection result based on the relative offset, generate auxiliary physiological prompts based on ECG-related auxiliary physiological parameters, and use the psychoneural activity detection result and auxiliary physiological prompts together as a comprehensive detection output.

2. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 1, characterized in that, The interface impedance sequence is normalized to obtain a normalized impedance sequence, and the total energy of the frontal region and the total energy of the temporal region are extracted according to the time window, including: Physiological electrical signals and interfacial impedance sequences were collected at corresponding locations on the left frontal, right frontal, left temporal, and right temporal sides, respectively. Calculate the median and absolute median difference of the interface impedance sequence corresponding to each acquisition location; Calculate the difference between the current time series value of the interface impedance sequence at each acquisition location and the corresponding median, and divide by the corresponding absolute median difference to obtain the normalized impedance sequence at each acquisition location. The signal energy of the physiological electrical signals corresponding to each collection location in the frontal region is extracted according to the time window, and the total energy of the frontal region is obtained by pooling them together. The signal energy of the physiological electrical signals corresponding to each acquisition location in the temporal region is extracted according to the time window, and the total energy of the temporal region is obtained by summing them.

3. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 1, characterized in that, Frontal and temporal compression surrogate quantities are constructed based on normalized impedance sequences, and the frontotemporal contact pressure migration index is calculated based on these quantities, including: Calculate the mean value of the normalized impedance sequence corresponding to each acquisition position in the quota area within the current time window, and construct the quota area compression proxy based on the mean value; The mean value of the normalized impedance sequence corresponding to each acquisition location in the temporal region within the current time window is calculated, and the temporal region compression proxy is constructed based on the mean value. Based on the difference and amplitude relationship between the temporal compression surrogate and the frontal compression surrogate, the frontotemporal contact pressure migration index, which characterizes the relative migration direction and degree of pressure between the frontal and temporal regions, is calculated.

4. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 1, characterized in that, The frontotemporal inverse linkage coefficient is calculated based on the normalized impedance sequence, including: A local analysis window is constructed based on the current time window and its adjacent time windows; Within the local analysis window, the average time window value of the normalized impedance sequence corresponding to each acquisition location in the frontal region is averaged to obtain the average impedance sequence of the frontal region. Within the local analysis window, the average time window value of the normalized impedance sequence corresponding to each acquisition location in the temporal region is averaged to obtain the temporal region impedance mean sequence. The correlation coefficient between the mean impedance sequence of the frontal region and the negative sequence of the mean impedance sequence of the temporal region is calculated, and the correlation coefficient is used as the frontotemporal inverse linkage coefficient to characterize the inverse relationship between the changes in impedance of the frontal region and the temporal region.

5. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 1, characterized in that, The original psychoneural activity load is constructed based on the total energy of the frontal and temporal regions. A feature vector is formed by combining the original psychoneural activity load, the frontotemporal contact pressure migration index, and the frontotemporal reverse linkage coefficient, including: The energy comparison data for the corresponding time window is constructed based on the ratio of total energy in the frontal region to total energy in the temporal region. Logarithmic transformation of the energy comparison quantity yields the original psychoneural activity load; The changes in the frontotemporal contact pressure migration index and the changes in the original psychoneural activity load between adjacent time windows are extracted. The original psychoneural activity load, the frontotemporal contact pressure migration index, the frontotemporal reverse linkage coefficient, the changes in the frontotemporal contact pressure migration index, and the changes in the original psychoneural activity load are combined to form a feature vector.

6. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 5, characterized in that, State recognition of feature vectors includes: The feature vectors are input into the Hidden Markov Model for state classification to obtain the stress migration state sequence corresponding to each time window. The pressure migration state sequence includes a stable fit state, a frontal-to-temporal migration state, a temporal-to-frontal migration state, and a migration transition state.

7. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 6, characterized in that, Migration interference components are constructed based on the stress migration state sequence, and the original psychoneural activity load is corrected according to the migration interference components, including: For the frontotemporal migration state, temporal-frontal migration state, or migration transition state corresponding to the current time window, the migration interference component is constructed based on the state coefficients obtained by fitting the data from the calibration data acquisition segment after the reference baseline stage of the current subject. The migration interference component includes a first component based on the frontotemporal contact pressure migration index and a second component based on the change of the frontotemporal contact pressure migration index between adjacent time windows. The original psychoneural activity load is differentially corrected based on the migration interference component to obtain the corrected psychoneural activity load. When the current time window corresponds to a stable fit state, the original psychoneural activity load is directly used as the corrected psychoneural activity load.

8. The physiological and electrocardiographic comprehensive detection method for assessing psychoneural activity according to claim 1, characterized in that, The corrected psychoneural activity level is compared with a reference baseline to obtain a relative offset, and the psychoneural activity detection results are determined based on the relative offset, including: Calculate the statistical center value of the corrected psychoneural activity for each time window within the reference baseline period, and use it as the baseline reference value; The offset of the corrected psychoneural activity level within the current time window relative to the baseline reference value is used as the relative offset. The upper and lower decision limits are determined based on the offset distribution range of the corrected psychoneural activity amount relative to the baseline reference value for each time window within the reference baseline phase. When the relative offset is greater than the upper judgment threshold, a detection result representing enhanced psychoneural activity is obtained; When the relative offset is less than the lower decision threshold, a detection result representing a weakening of psychoneural activity is obtained; When the relative offset is between the upper and lower decision limits, a detection result representing stable psychoneural activity is obtained.

9. A comprehensive physiological and electrocardiographic detection system for assessing psychoneural activity, used to perform the comprehensive physiological and electrocardiographic detection method for assessing psychoneural activity as described in any one of claims 1-8, characterized in that, Specifically, it includes: The data processing module is used to collect physiological electrical signals and interface impedance sequences of the frontal and temporal regions, normalize the interface impedance sequences to obtain normalized impedance sequences, and extract the total energy of the frontal and temporal regions according to time windows. The index construction module is used to construct the frontal and temporal compression surrogate quantities based on the normalized impedance sequence, and to calculate the frontotemporal contact pressure migration index based on the frontal and temporal compression surrogate quantities. The state recognition module is used to calculate the frontotemporal reverse linkage coefficient based on the normalized impedance sequence, construct the original psychoneural activity load based on the total energy of the frontal region and the total energy of the temporal region, form a feature vector by combining the original psychoneural activity load, the frontotemporal contact pressure migration index and the frontotemporal reverse linkage coefficient, and perform state recognition on the feature vector to obtain the pressure migration state sequence. The load correction module is used to construct migration interference components based on the pressure migration state sequence, and to correct the original psychoneural activity load based on the migration interference components to obtain the corrected psychoneural activity. The evaluation output module compares the corrected psychoneural activity level with the reference baseline to obtain the relative offset, determines the psychoneural activity detection result based on the relative offset, generates auxiliary physiological prompts based on ECG-related auxiliary physiological parameters, and combines the psychoneural activity detection result and auxiliary physiological prompts as a comprehensive detection output.

Images