Brain-computer interface based consciousness level regulation system and method

Abstract

The application provides a brain-computer interface-based consciousness level regulation system and method, which comprises: a signal sensing module for synchronously collecting multi-modal physiological signals of a user and performing time alignment preprocessing; a central decision module for performing feature extraction and fusion on the preprocessed multi-modal physiological signals, generating a state response index representing a consciousness state, and comparing the index with a dynamic activation threshold in real time to output a control instruction; and a stimulation execution module for applying multi-modal coordinated physical stimulation to the user in response to the control instruction. The central decision module is further configured to adaptively adjust stimulation parameters of the physical stimulation and / or the dynamic activation threshold in the next cycle according to a change trend of the state response index before and after the physical stimulation is applied. The application realizes adaptive closed-loop regulation of "sensing-decision-execution-optimization", and significantly improves the robustness, accuracy and personalized adaptive ability of consciousness level regulation.

Description

Technical Field

[0001] This invention belongs to the field of neural engineering and medical device technology, specifically relating to a consciousness level regulation system and method based on brain-computer interface (BCI). Background Technology

[0002] With the rapid development of brain-computer interfaces (BCIs) and neural engineering technologies, targeted intervention of the target subject's internal brain consciousness state through external physical stimulation has become a research hotspot in the field of consciousness level regulation. Especially in the regulation of consciousness level during long-term low-arousal states of the target subject, establishing a highly efficient and robust closed-loop regulation mechanism is currently key to improving regulation performance.

[0003] Currently, existing techniques for regulating consciousness levels mainly fall into two categories:

[0004] (1) Traditional mainstream mode: open-loop physical stimulation. That is, stimulation (such as transcranial magnetic stimulation, peripheral nerve electrical stimulation) is applied to the target object according to preset fixed time, frequency and intensity parameters, which does not depend on the real-time physiological or neural state feedback of the target object. It belongs to typical non-adaptive regulation. (2) Primary closed-loop feedback mode: some experimental closed-loop regulation methods have recently emerged. These methods usually collect single-modality EEG signals of the target object and set a simple response threshold. When the EEG characteristics exceed the threshold, stimulation is triggered.

[0005] Although the above-mentioned primary closed-loop model introduces the concept of state feedback, the existing technology still has the following fundamental defects in achieving high-precision and high-stability control of consciousness level: (1) One-sidedness and instability of state assessment: The existing closed-loop scheme relies heavily on a single EEG signal modality to judge the state of consciousness. EEG signals are easily interfered with by the external environment and the target object's own body movements, resulting in a low overall signal-to-noise ratio, which makes it difficult to accurately assess the target object's level of consciousness and to serve as a reliable basis for triggering precise control. (2) Coarse timing of intervention and lack of physiological synchronicity: Due to the unreliability of the assessment feedback signal, the existing technology is unable to accurately capture the best time window for the target object's brain to produce a positive physiological response. The triggering of stimulation often has transmission delays or misjudgments, resulting in the external intervention signal being out of sync with the brain's internal neural activity, making it impossible to achieve time locking between the control signal and the positive physiological state, with an effect similar to "blind stimulation". (3) System rigidity and lack of personalization and adaptability The stimulation parameters (such as intensity, frequency, and content) and trigger thresholds of the existing scheme are usually set once and fixed before the system runs. It cannot dynamically adjust online based on individual differences of the target object or changes in the neural evolution state of the same target object at different control periods, and cannot achieve true adaptive personalized control. (4) Fragmented technology, without forming a complete intelligent closed loop: Existing solutions focus on improving single links such as signal acquisition or stimulus application, and fail to seamlessly integrate links such as "high robustness state perception, real-time intelligent decision-making, multi-mode precise execution, and data-driven optimization" into a unified closed loop system, resulting in the inability to form a complete intelligent control closed loop that can learn and continuously improve based on control feedback. Summary of the Invention

[0006] The technical objective of this invention is to address the technical problems of limited and unstable regulatory effectiveness caused by the current methods of regulating consciousness level due to one-sided assessment, crude intervention, and lack of adaptability. The invention provides an intelligent closed-loop consciousness level regulation system and method that can automatically perform precise and personalized regulation based on the user's real-time and stable neurophysiological state.

[0007] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution.

[0008] In a first aspect, embodiments of the present invention provide a consciousness level regulation system based on a brain-computer interface, comprising: a signal sensing module, used to synchronously collect the user's multimodal physiological signals and perform time-aligned preprocessing;

[0009] The central decision-making module is used to extract and fuse features based on the preprocessed multimodal physiological signals to generate a state response index that represents the state of consciousness; compare the state response index with the dynamic activation threshold in real time, and output control commands to the stimulus execution module when the triggering conditions are met;

[0010] The stimulus execution module, in response to the control command, applies multimodal coordinated physical stimulation to the user;

[0011] The central decision-making module is further configured to adaptively adjust the stimulation parameters and / or the dynamic activation threshold of the multi-mode collaborative physical stimulus applied in the next cycle based on the changing trend of the state response index before and after the application of the physical stimulus.

[0012] Secondly, embodiments of the present invention provide a method for regulating consciousness level based on a brain-computer interface, applying a brain-computer interface-based consciousness level regulation system as provided in any possible implementation of the first aspect. The method is executed by a central decision-making module in the system and includes the following steps: receiving multimodal physiological signals synchronously collected from the user and preprocessed with time alignment; extracting and fusing features from the preprocessed multimodal physiological signals to generate a state response index characterizing the state of consciousness; comparing the state response index with a dynamic activation threshold in real time, and outputting a control command to a stimulus execution module to apply multimodal coordinated physical stimulation when a triggering condition is met; and adaptively adjusting the stimulation parameters and / or the dynamic activation threshold of the physical stimulation applied in the next cycle according to the changing trend of the state response index before and after the application of the physical stimulation.

[0013] Compared with existing technologies, the brain-computer interface-based consciousness level regulation system and method provided in this invention have achieved the following beneficial technical effects: 1. More comprehensive assessment and more robust decision-making:

[0014] This invention overcomes the limitations of traditional methods that rely on single, unstable EEG signals by simultaneously acquiring and fusing multimodal physiological signals such as EEG, ECG, and EEG, and calculating a comprehensive "state response index." The mutual complementarity and verification of multimodal information significantly improves the accuracy, robustness, and anti-interference ability of consciousness level assessment, providing an unprecedentedly reliable basis for subsequent intelligent decision-making.

[0015] 2. Achieve fully automated, highly efficient, and precise closed-loop intervention:

[0016] The system achieves a fully automated closed loop at the millisecond level, from signal perception and analysis to stimulus triggering. By continuously monitoring the "state response index" and using dynamic thresholds for judgment, the system can accurately capture the "optimal time window" when the user's brain enters a positive response state and automatically apply intervention at that instant. This completely eliminates the delays and subjective errors caused by manual judgment and operation, ensuring high-precision timing between stimulation and the brain's internal neural activity, and greatly improving the accuracy and timeliness of intervention.

[0017] 3. Possesses personalized and adaptive awareness regulation capabilities:

[0018] The core of this system is an adaptive decision engine, which evolves the consciousness level regulation scheme from a "fixed program" to a "personalized intelligent strategy." The engine dynamically adjusts subsequent stimulus parameters (such as electrical stimulation intensity and auditory content) and system thresholds based on the user's real-time physiological feedback (changes in the state response index) to each stimulus. This "evaluation-feedback-optimization" cycle allows consciousness level regulation to "adapt to different times and individuals," truly fitting the user's individual differences and state evolution, breaking through the effectiveness ceiling of traditional fixed-parameter consciousness level regulation methods.

[0019] 4. Establish a data-driven system for sustainable evolution and learning:

[0020] The system comprehensively records the entire lifecycle data (physiological signals, stimulus events, optimization logs) of each level of consciousness regulation and uses this data to continuously update the user's personalized response model (or personalized model). Through long-term learning, the system becomes increasingly accurate with each regulation, prioritizing the application of historically optimal solutions to continuously improve the effectiveness of level-of-consciousness regulation. This constructs a data-driven, self-iteratory intelligent medical system, enabling long-term optimization of the effectiveness of level-of-consciousness regulation.

[0021] 5. The system has a high degree of integration and good scalability and versatility:

[0022] This invention provides a complete integrated hardware and software system solution that seamlessly integrates modules such as multimodal perception, intelligent fusion decision-making, dual-modal stimulus execution, and data management learning, forming a complete intelligent closed loop of "perception-decision-execution-optimization". This highly integrated design not only ensures the feasibility and stability of the solution, but its modular core architecture (such as multimodal fusion and adaptive closed-loop control) can also be extended to other fields requiring precise control of the level of consciousness, such as cognitive enhancement and emotion regulation, demonstrating strong versatility.

[0023] It should be understood that the summary section is not intended to identify key or essential features of the embodiments of this disclosure, nor is it intended to limit the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0024] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of this application in any way. In the drawings:

[0025] Figure 1 A schematic diagram of the framework of a brain-computer interface-based consciousness level regulation system provided for an embodiment;

[0026] Figure 2 This is a schematic diagram of a brain-computer interface-based method for regulating consciousness levels, provided as an example. Detailed Implementation

[0027] To enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application.

[0028] It should be fully understood that the multimodal physiological signals of individual users involved in this application are all information and data authorized by the user or fully authorized by all parties. The use of user information should comply with the privacy policies and practices of the industry that are generally considered to meet or exceed the requirements for maintaining user privacy. The collection, use and processing of related data should comply with relevant laws, regulations and standards, and provide corresponding operation access points for users to choose to authorize or refuse.

[0029] An example is a brain-computer interface-based consciousness level control system, such as... Figure 1 As shown, it includes a signal sensing module, a central decision-making module, and a stimulus execution module.

[0030] The signal sensing module synchronously acquires the user's multimodal physiological signals and performs time-aligned preprocessing. The central decision-making module extracts and fuses features based on the preprocessed multimodal physiological signals to generate a state response index representing the state of consciousness; it compares the state response index with a dynamic activation threshold in real time and outputs control commands to the stimulus execution module when triggering conditions are met. The stimulus execution module responds to the control commands by applying multimodal coordinated physical stimuli to the user. The central decision-making module also adaptively adjusts the stimulation parameters and / or dynamic activation threshold of the multimodal coordinated physical stimuli applied in the next cycle based on the changing trend of the state response index before and after the application of physical stimulation.

[0031] In this embodiment, the multimodal physiological signals include any two or more combinations of electroencephalogram (EEG) signals, electrocardiogram (ECG) signals, skin conductance signals, electromyogram (EMG) signals, blood oxygen saturation signals, and respiratory signals.

[0032] In some embodiments, the acquired multimodal physiological signals include electroencephalogram (EEG) signals, electrocardiogram (ECG) signals, and electrodermal response (TES) signals.

[0033] The system simultaneously collects and integrates multimodal physiological signals such as EEG, ECG, and EEG. Through feature extraction and weighted fusion, it generates a comprehensive and more robust state response index as a unified and robust quantitative indicator for assessing the level of consciousness.

[0034] The system continuously monitors the state response index and sets a dynamic activation threshold. Only when the index consistently and stably exceeds the threshold (indicating entry into the effective response window) will the system automatically and instantly trigger stimulation, achieving high-precision time locking between intervention and the brain's active state.

[0035] As an example, a multi-lead dry electrode EEG cap can be used to collect the user's electroencephalogram (EEG) signals, serving as a core signal for assessing cortical arousal and cognitive processing. Patch electrodes can be used to collect the user's electrocardiogram (ECG) signals, including heart rate and heart rate variability, reflecting the balance of the autonomic nervous system. Finger or palm sensors can be used to collect the user's skin conductance signals, directly characterizing the level of sympathetic nerve excitation.

[0036] In some embodiments, electromyographic signals of the user are also collected in specific muscle groups via surface electrodes to monitor minute motor responses to stimuli or to eliminate motion artifacts.

[0037] In this embodiment, the signal sensing module also performs real-time preprocessing on the acquired multimodal physiological signals to improve the signal-to-noise ratio and prepare for feature extraction. As an example, the preprocessing includes:

[0038] For EEG signals: bandpass filtering (e.g., 0.5-45Hz) is performed to preserve physiological rhythms, power frequency notch filtering is used to remove mains interference, and blind source separation algorithm is used to attempt to remove artifacts such as eye movement and ECG.

[0039] For ECG signals: bandpass filtering is performed, R waves are detected, and beat-by-beat heart rate is calculated.

[0040] For the electrodermal signal: low-pass filtering is applied to smooth the signal, and the skin conductance level and its slow fluctuations are extracted.

[0041] Preprocessing also includes time alignment of the acquired physiological signals: raw signals and derived features from all modalities are stamped with a uniform, high-precision system clock. The preprocessed data stream is strictly aligned with stimulus event markers (such as stimulus start / end) on the time axis to ensure the accuracy of causal relationships in subsequent fusion analysis, which is the foundation for achieving precise stimulus-response correlation.

[0042] One of the core aspects of this invention is to construct a comprehensive and robust quantitative index to characterize the level of consciousness, called the state response index.

[0043] In this embodiment, the central decision-making module extracts and fuses features from the preprocessed multimodal physiological signals. For example, from the preprocessed EEG signals, it calculates in real-time the relative power, power ratio, and EEG complexity of specific frequency bands (such as Delta, Theta, Alpha, Beta), providing direct evidence of cortical activity. From the ECG signals, it calculates in real-time the time and frequency domain indices of heart rate variability, reflecting the background state of autonomic nervous system balance. From the electrodermal signal, it calculates in real-time the mean and fluctuation frequency of skin conductance levels, indicating the immediate response to sympathetic nerve excitation.

[0044] The central decision-making module first normalizes each feature to obtain the corresponding sub-index features; then, using the preset fusion weights of each sub-index feature, it performs a weighted summation of each sub-index feature, integrates the feature vectors extracted from different physiological dimensions, and finally outputs a single, continuous scalar value—the state response index; wherein, the fusion weights of each sub-index feature satisfy the preset scaling and normalization conditions.

[0045] As an example, the specific formula for determining the state response index is as follows:

[0046] ;

[0047] This formula is an exemplary implementation, illustrating the fusion path from a multimodal signal to a single exponent. Wherein: The state response index at time t is the higher the value, the higher the "level of consciousness" or "brain responsiveness" judged by the algorithm. This refers to the normalized brain electron index characteristics (also known as the EEG consciousness index). This is a normalized HRV (Heart Rate Variability) sub-index feature (also known as the HRV autonomic nervous index). The normalized GSR sub-index features (also known as the GSR (Galvanic Skin Response) excitability index). , , These are the fusion weights of the EEG consciousness index, HRV autonomic nervous system index, and GSR excitability index, respectively.

[0048] In this embodiment, the calculation formulas and physical meanings of each sub-index feature are as follows:

[0049] 1) EEG Awareness Index :

[0050] It comprehensively reflects the arousal and information processing capabilities of the cerebral cortex. The formula is designed so that its value is positively correlated with the level of consciousness, such as:

[0051] ;

[0052] in: , These are the average powers of the EEG signal in the Theta band (4-8Hz) and Beta band (13-30Hz) within the current sliding time window, respectively. This ratio typically increases as consciousness levels decrease. This is the normalized value of the permutation entropy of the EEG signals within the current time window. Permutation entropy is an indicator of the complexity of a time series; when the person is conscious, the EEG is more complex and the entropy value is higher; when the person is impaired, the EEG is more regular and the entropy value is lower. A harmonic weight (e.g., 0.7) is used to balance the power ratio and complexity features. This is a smoothing term, and it is a very small constant to prevent the denominator from being zero.

[0053] Calculate the baseline period using a sliding time window. and The maximum and minimum values ​​of each are calculated to normalize the real-time value to the [0,1] interval. Finally, the value is calculated using the above formula to make it tend towards 1 when the level of consciousness increases.

[0054] In this embodiment, the length of the sliding time window is set to 2-8 seconds (e.g., preferably 4 seconds), and the sliding step size is set to 0.5-1 seconds to ensure the real-time performance of the evaluation, that is, the 4-second EEG signal is calculated once every 0.5-1 second. index.

[0055] 2) HRV autonomic nervous index :

[0056] This index reflects the activity of the parasympathetic (vagus) nerve in the autonomic nervous system. Increased parasympathetic tone is associated with relaxation and recovery, and may reflect active physiological regulation under certain arousal paradigms. As an example, the formula is as follows:

[0057] ;

[0058] in, Here, represents the normalized high-frequency and low-frequency power of the heart rate variability signal within the current sliding time window, respectively. High-frequency power is highly correlated with respiratory sinus arrhythmia and is a marker of parasympathetic activity.

[0059] Spectral analysis of the beat-by-beat heart rate sequence yielded low-frequency (LF) and high-frequency (HF) power. After normalization within the baseline period, the power was calculated using the aforementioned formula. An increase in this exponent may indicate a relative enhancement of parasympathetic activity.

[0060] 3) GSR ​​Excitability Index :

[0061] This index directly characterizes the excitation level of the sympathetic nervous system. Upon effective stimulation, skin conductance may show a transient or sustained increase. As an example, the formula is as follows:

[0062] ;

[0063] in, This is the normalized value of skin conductance level within the current time window. Direct extraction. The mean of the values ​​was normalized over the baseline period.

[0064] The fusion weights reflect the relative importance of each modality feature in the comprehensive evaluation. They are typically based on prior knowledge and optimized through machine learning from data. In this embodiment, the fusion weights are determined through the following offline optimization steps: acquiring multimodal physiological signals collected synchronously at multiple evaluation time points and constructing multimodal sub-index feature vectors; acquiring the user's external reference evaluation score at the corresponding time point; establishing a linear mapping model between the multimodal sub-index feature vectors and the reference evaluation score; and finding the optimal solution that satisfies the preset scaling and normalization conditions for the fusion weights, with the optimization objective of minimizing the mean square error between the inner product of the multimodal sub-index feature vectors and the fusion weights and the reference evaluation score. This optimal solution is then used as the preset fusion weights.

[0065] It should be noted that the modal sub-index features (such as brain electronics index features) described in the embodiments of the present invention are... HRV sub-index characteristics GSR sub-index characteristics The form is not limited to this; it can be a single scalar value or a multidimensional feature vector or matrix composed of multiple feature dimensions.

[0066] Accordingly, the preset fusion weights are adaptively matched according to the dimensions of the sub-index features: when the sub-index features are scalar values, the corresponding fusion weights are single scalar coefficients; when the sub-index features are multi-dimensional feature vectors or matrices, the corresponding fusion weights are weighted vectors or mapping matrices of the same dimensions.

[0067] In some embodiments, the preset scaling and normalization condition specifically refers to the fact that the fusion weights of each sub-index feature are all non-negative real numbers and their sum equals 1, expressed as: This ensures that the final output scalar exponent is strictly limited to a fixed scalar range of 0 to 1.

[0068] As an example, the specific method for obtaining the fusion weights is as follows:

[0069] By establishing a correlation between the daily mean of the State Response Index (SRI) and changes in clinical gold standard consciousness level scale scores (such as the CRS-R score), a supervised learning model is used to automatically find an optimal set of fusion weights, ensuring that the daily variation trend of the SRI can best explain or predict changes in the scale scores. Essentially, this uses the effect of clinical consciousness level regulation as a "mentor signal" to guide the optimization of the assessment model.

[0070] Data Alignment and Preparation: Within a consciousness level regulation cycle, multiple assessment time points are selected, and the consciousness level regulation data for each day is extracted to calculate the data for all consciousness level regulation periods within that day. , and Let the eigenvectors On the same day or within a closely adjacent time window (such as before / after the regulation of consciousness level on the same day), clinicians assess users using standardized scales (such as the Coma Recovery Scale Revised Version) as external reference scores to obtain the total scale score for that day. .

[0071] Model assumptions and There exists a linear mapping relationship between them, that is:

[0072] ;

[0073] in, Here, represents the coefficients of the three modal indicators, y represents the scale score, x represents the state response index corresponding to the scale assessment time, and j and k represent different assessment time points. and These represent the user's State Response Index (SRI) at time points k and j, respectively. This represents the exponential change in the state response from time point j to time point k; and These represent the user's scale ratings at time points k and j, respectively. This represents the change in a user's scale rating from time point j to time point k; For constant terms; To estimate the change in scale ratings for users from time point j to time point k.

[0074] The weights are optimized to minimize the loss, and the loss is designed as follows:

[0075] ;

[0076] Where D is the total number of difference pairs; the constraint is: ,and .

[0077] The State Response Index comprehensively reflects the excitability of the user's cerebral cortex, the arousal level of the autonomic nervous system, and the activity level of the sympathetic nervous system. It is a more stable and interference-resistant quantitative indicator of a comprehensive state of consciousness / awakeness than any single modality feature; its upward trend is interpreted as an improvement in the level of consciousness or brain responsiveness.

[0078] In this embodiment, the stimulus execution module employs a phased approach (including a primary continuous stimulus phase and a secondary conditional reinforcement stimulus phase) and a multimodal synergistic stimulus strategy.

[0079] As an example, personalized auditory stimulation is preferred for primary continuous stimulation. The system maintains a personalized sound library for each user (such as recordings from loved ones, music the user enjoyed before the illness, natural sounds, etc.). Specifically, at the start of consciousness level modulation, the system selects and plays the stimulus. Its function is to gently and continuously modulate the brain's arousal state, creating a "time window" for subsequent precise intervention. Personalized auditory stimulation aims to better improve the user's consciousness level indicators, designing personalized adaptive strategies and adjusting parameters such as the relevant stimulus sound source and volume.

[0080] The core of the primary continuous stimulation strategy is to construct a closed loop of "exploration-evaluation-optimization-monitoring" to ensure that the primary auditory stimulation used by the system is always the most effective and appropriately dosed for the current user. In the initial stage of consciousness level modulation, different sound stimuli will be tried, automatically locking in the material that can best enhance the user's consciousness level; the stimulation volume will be fine-tuned based on real-time effect feedback; when the effect declines, the stimulation program will be automatically re-evaluated and switched.

[0081] In this embodiment, the secondary conditional reinforcement stimulation is preferably transcutaneous electrical nerve stimulation (TENS). The system automatically triggers when the state response index (representing the level of consciousness) reaches a threshold and remains stable for a preset minimum duration. The target is preferably a peripheral nerve with proven neuromodulation effects, such as the median nerve or the vagus nerve. Its function is to apply a precise, controllable physical stimulus (usually a commonly used arousal stimulation method such as median nerve or transvagus nerve stimulation) on top of the primary continuous stimulation at the optimal moment when the brain enters a positive response state, aiming to produce a "time-locked" reinforcement effect on neural activity.

[0082] In some embodiments, when the central decision module makes adaptive adjustments (using an adaptive decision engine), it specifically uses the increase in the state response index before and after a single physical stimulus trigger as a reward signal; if the reward signal is lower than the preset expectation, it increases the intensity of the physical stimulus or switches the stimulus content within a safe threshold range, and adaptively lowers the dynamic activation threshold; if the reward signal continues to be higher than the preset expectation, it maintains the current stimulus parameters and raises the dynamic activation threshold to match the user's state evolution.

[0083] In this embodiment, the dynamic activation threshold does not use a fixed value as the trigger threshold. Instead, it dynamically calculates and adjusts a personalized activation threshold based on each user's real-time baseline level and historical awareness level. As an example, it may include:

[0084] 1) Establishment of personalized baselines and Z-score normalization: This step establishes individualized "benchmarks" and "statistical baselines" for the entire dynamic activation threshold system, including:

[0085] Data Acquisition: At the start of each level of conscious regulation, the system automatically acquires multimodal physiological signals for a preset duration (e.g., 2-5 minutes) in a resting state without any level of conscious regulation stimulation.

[0086] Feature calculation and fusion: This section of the resting signal is processed in real time to calculate the time series of the state response index.

[0087] Statistical baseline calculation: The mean and standard deviation of the State Responsiveness Index (SRI) are calculated for this resting period time series. The mean reflects the user's current baseline level of awareness, and the standard deviation reflects the inherent range of fluctuations in their physiological state.

[0088] Z-score normalization: Throughout the subsequent intervention period, the system converts all real-time calculated SRI values ​​into Z-scores based on the mean and standard deviation of this resting period, as shown in the following formula:

[0089] ;

[0090] Z(t) represents the degree to which the real-time SRI deviates from the individual's resting baseline, expressed in standard deviations. Z(t) = 1 means that the real-time level of consciousness is 1 standard deviation higher than the individual's resting level. This represents the mean of the state response index within the resting period time series. This represents the labeled difference of the state response index within the resting period time series.

[0091] 2) Setting, truncation and mapping of dynamic activation threshold: This step defines the “threshold line” for triggering stimuli, which is dynamically adjustable.

[0092] In this embodiment, the initialization and dynamic adjustment of the dynamic activation threshold are as follows:

[0093] Initialization: In the first round of consciousness level modulation or when historical data is lacking, the system adopts a conservative default Z-score threshold (e.g., This means that a significant response is considered to have been achieved only when a user's real-time level of awareness (Z score) consistently exceeds their resting baseline by two standard deviations.

[0094] Dynamic adjustment is based on the following: The system maintains a recent historical window of consciousness level regulation (e.g., data from the previous 5 consciousness level regulation periods). The Z-score threshold parameter is automatically adjusted based on the mean and variance distribution of the Z-scores during the consciousness level regulation periods within this window.

[0095] Adaptive adjustment: If recent user feedback has been generally weak, it indicates that the default threshold is too high and difficult to achieve. The system can fine-tune and lower the Z-score threshold. (e.g., from 2 to 1.5) to reduce the difficulty of triggering and maintain the accessibility of consciousness level control.

[0096] Increased Challenge: If users consistently reach and exceed the current threshold in the near future, indicating improved ability, the system may slightly increase the Z-score threshold. (e.g., from 2 to 2.3) to provide a moderate challenge and promote further recovery.

[0097] 3) Z-score truncation and normalization mapping: To ensure system stability and transform the Z-score into a more intuitive "probability of achievement", the following processing is required:

[0098] Truncating: The real-time Z-score is truncated, for example, by limiting it to the interval [−3, +3] or [0, +3] (to focus on positive improvement). The truncation boundary can be slightly adjusted based on the actual distribution of recent Z-scores. For example, if historical data shows that a user's Z-score has never been lower than -2, the lower bound can be adjusted to -2.

[0099] Mapping to the 0-1 interval: The truncated Z score is linearly mapped to the [0,1] interval to obtain the normalized response level.

[0100]

[0101] in: It is the truncated Z-score value. It is the Z-score cutoff boundary.

[0102] The dynamically adjusted Z-score threshold This can also be mapped using the above formula:

[0103] ;

[0104] It is the truncated Z-score threshold. It is the Z-score cutoff threshold boundary.

[0105] Obtain the corresponding normalized dynamic activation threshold (The initial value is usually set to 0.5, corresponding to the median Z-score.) This normalized dynamic activation threshold. It is the core dynamic variable that the system ultimately uses to determine whether to trigger, and it normalizes the response level. With normalized dynamic activation threshold The comparison is performed, and control instructions are output to the stimulus execution module when the triggering conditions are met.

[0106] In this embodiment, adaptive optimization of stimulation parameters may include:

[0107] 1) Auditory stimulation: The volume can be dynamically adjusted or different types of sounds can be switched according to the response speed of the real-time state response index in order to find the optimal stimulation content.

[0108] 2) Electrical stimulation: Its key parameters (such as current intensity, frequency, and pulse width) constitute an optimizable parameter space. The engine employs a feedback-based optimization algorithm: the magnitude and duration of the increase in the state response index after each stimulus trigger are used as a "reward signal." If the reward is lower than expected, the stimulation intensity is exploratoryly increased or other parameters are adjusted within a safe range; if the reward remains high, the parameters are maintained or fine-tuned to find the lowest effective dose. This allows the consciousness-level modulation scheme to continuously evolve towards the direction most effective for a specific user.

[0109] The optimal stimulation scheme selection process of this system is a data-driven, closed-loop feedback, and dynamically adaptive intelligent process. Its core technology lies in the adaptive decision engine, which dynamically adjusts the stimulation type, content, parameters, and combination sequence based on real-time calculated changes in the state response index to maximize the effect of consciousness level regulation. The core objective of the adaptive adjustment strategy is to measure the magnitude, stability, and duration of the increase in the user's state response index during consciousness level regulation. The real-time state response index serves as a direct, quantifiable feedback signal. Its rate of increase, peak level, and duration exceeding the threshold are all used as key indicators to evaluate the effectiveness of the current stimulation scheme.

[0110] In some embodiments, real-time dynamic optimization and adaptive adjustment are performed within a single level of consciousness modulation.

[0111] Adaptive adjustments can include stimulus paradigm switching. The system has a built-in, extensible stimulus paradigm library. During awareness level regulation, if the currently used stimulus paradigm fails to effectively increase SRI or reach the trigger condition within a preset time, the decision engine can automatically switch to an alternative paradigm according to preset rules to explore more effective stimuli for the current user.

[0112] Optionally, adaptive adjustment also includes real-time parameter fine-tuning. For a selected stimulus paradigm, its parameters (such as the volume of auditory stimulation, the intensity of electrical stimulation, and the frequency) are defined within a safe, adjustable range. The adaptive decision engine adjusts the parameters in small steps (e.g., increasing the intensity if the response is weak, and decreasing it slightly if the response is too strong) when the same stimulus is applied again, based on the immediate response of the SRI to the previous stimulus (e.g., the magnitude of the increase).

[0113] In some embodiments, the central decision-making module is also used to establish and continuously update a personalized stimulus-response map for each user, recording the average SRI response effect corresponding to different stimulus paradigms and parameter combinations in historical consciousness level regulation. Based on this map, at the beginning of each new consciousness level regulation, the system's central decision-making module can prioritize recommending the stimulus scheme that historical data shows is most effective for the user as the starting scheme, thereby achieving "the more regulation, the more accurate."

[0114] In some embodiments, the stimulus execution module is also used to perform primary-secondary stimulus coordination: a fixed coordination pattern of "primary auditory stimulation + secondary conditioned electrical stimulation". Auditory stimulation is used to create and maintain an elevated arousal background, while electrical stimulation intervenes precisely to reinforce the arousal when the SRI reaches its optimal window. The coordination of the two is scheduled by a unified SRI index and dynamic thresholds.

[0115] Dynamic activation threshold management: As a "valve" for triggering secondary stimuli, the dynamic activation threshold itself is also an optimizable parameter. The adaptive decision engine of the central decision module dynamically adjusts this threshold based on the user's recent response level, ensuring that it is always at a moderately challenging level that is "just within reach," avoiding ineffective or excessive stimulation due to a fixed threshold.

[0116] This invention also provides a method for regulating consciousness level based on a brain-computer interface system. The method utilizes the consciousness level regulation system based on a brain-computer interface system as described in the above embodiments. The method is executed by a central decision-making module within the system and includes the following steps: receiving multimodal physiological signals synchronously collected from the user and preprocessed with time alignment; extracting and fusing features from the preprocessed multimodal physiological signals to generate a state response index characterizing the state of consciousness; comparing the state response index with a dynamic activation threshold in real time, and outputting a control command to the stimulus execution module to apply multimodal coordinated physical stimulation when the triggering condition is met; and adaptively adjusting the stimulation parameters and / or dynamic activation threshold of the physical stimulus applied in the next cycle based on the changing trend of the state response index before and after the application of physical stimulation.

[0117] This method realizes an intelligent closed-loop consciousness level regulation process of "perception-decision-execution-optimization".

[0118] In specific embodiments, such as Figure 2 As shown, the method includes:

[0119] S1: Preparation for Level of Consciousness Regulation and System Initialization. The user wears an integrated signal acquisition and stimulation device. After system startup, multi-channel signal quality detection is automatically performed to ensure that the signal-to-noise ratio of each physiological signal (such as EEG and ECG) meets preset requirements.

[0120] S2: Personalized Baseline Automatic Establishment. A resting-state baseline is acquired without external stimulation. The system automatically collects physiological signals for a preset duration (baseline period, e.g., 2-5 minutes). Based on this signal, the mean and standard deviation of the user's initial, personalized state response index (i.e., consciousness level assessment index) are calculated. Using this as a benchmark, the state response index during the intervention period is normalized based on Z-scores. This is then used to initialize the dynamic activation threshold for subsequent decisions. For example, the Z-score of the state response index is truncated to ±3 and normalized to the 0-1 range, with the initial dynamic activation threshold set at 0.5.

[0121] S3: Main stimulus-monitoring-decision cycle. This is the core closed loop, which is executed cyclically until the preset total level of consciousness regulation duration is reached.

[0122] S3.1 Applying primary auditory stimulation: The system stimulation execution module selects content from the personalized sound library and applies continuous, adjustable auditory stimulation through an auditory stimulation device (such as bone conduction headphones) to initially enhance the user's brain arousal.

[0123] S3.2 Real-time synchronous acquisition and processing of multimodal signals: While applying stimulation, the system signal sensing module simultaneously acquires multiple physiological signals. After preprocessing by the signal sensing module, the central decision-making module extracts features and calculates the real-time state response index.

[0124] S3.3 Intelligent Threshold Judgment and Closed-Loop Trigger: The central decision engine compares the "State Response Index" and the "Dynamic Activation Threshold" in real time. When the index continuously exceeds the threshold and remains at a preset minimum duration (e.g., 8 seconds), it is determined that the user has entered an active response state, and secondary electrical stimulation is automatically and immediately triggered.

[0125] S3.4 Adaptive Parameter Tuning: Within a single cycle or between cycles, the system's central decision module analyzes the changes in evaluation indicators (such as the slope and peak value) before and after triggering the stimulus, and fine-tunes the stimulus parameters (such as electrical stimulation intensity and auditory stimulation content) and / or decision thresholds for the next cycle based on preset optimization rules.

[0126] S4: End of Consciousness Level Regulation and Data Archiving. After the total duration of consciousness level regulation is reached, the system automatically stops all stimulation. The complete set of synchronized data for this consciousness level regulation (original signals, evaluation indicators, all event markers, and adjusted parameters) is encrypted and stored, and used to update the user's personalized response model, providing an optimization basis for subsequent consciousness level regulation.

[0127] The following section, using a fictional case study of "User Z" to illustrate the modulation of consciousness levels, explains in detail how the above technical solution works:

[0128] (1) Scene setting, including:

[0129] User status: User Z, impaired consciousness after traumatic brain injury.

[0130] Stimulus library: Auditory A: Recordings of loved ones calling out, Auditory B: User-preferred light music, Auditory C: Rhythmic natural beats.

[0131] Initial parameters: electrical stimulation intensity 10mA, dynamic activation threshold SRI>0.65 (must be sustained for 8 seconds).

[0132] (2) Example process, including initiation and testing: The level of consciousness regulation begins, and the system defaults to starting with auditory A. After playing for 2 minutes, the monitoring showed that the SRI fluctuated slightly between 0.5 and 0.6, and the trigger condition was not met.

[0133] The adaptive decision engine of the central decision-making module determines that the current stimulus is ineffective and automatically switches to auditory stimulus B.

[0134] Identifying effective stimuli and initial triggers: After switching to auditory B, the user's SRI steadily increased to 0.7 within 1 minute and remained stable above 0.65 for 10 seconds.

[0135] Engine decision: Triggering conditions met. While continuously playing auditory stimulus B, automatically trigger 10mA electrical stimulation. Simultaneously, internally mark auditory stimulus B as "high-response stimulus for this level of consciousness modulation".

[0136] Parameter adaptive optimization: During electrical stimulation, the SRI further climbed to a peak of 0.82. After stimulation, the SRI stabilized at a new plateau of 0.72.

[0137] Engine analysis: This combination of "auditory B+ electrical stimulation" produced a strong response (SRI increased by 0.12).

[0138] Optimization decisions: The electrical stimulation intensity was slightly increased from 10mA to 10.5mA to attempt to enhance the effect next time; based on the improved SRI platform, the dynamic activation threshold was adaptively increased from 0.65 to 0.68 to match the user's improved state.

[0139] Locking and Optimization in the Cycle: In subsequent cycles of conscious level modulation, the system locks auditory stimulus B as the primary primary stimulus. When the SRI continues to exceed the new threshold of 0.68, the system triggers electrical stimulation with an optimized intensity of 10.5 mA.

[0140] Long-term learning: After this round of consciousness level modulation, the system significantly improved the weight score of the combination of auditory B and the "intensity 10.5mA" parameter in user Z's personalized model. In the next consciousness level modulation, the system will prioritize the combination of auditory B and 10.5mA electrical stimulation parameters as the initial scheme, thus skipping the exploration phase and directly applying the historically optimal scheme.

[0141] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or physical entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, or a tablet computer, or any combination of these devices.

[0142] The above provides a detailed description of the brain-computer interface-based consciousness level control system and method provided in this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the concept of this application and should not be construed as a limitation on the scope of protection of this application.

Claims

1. A consciousness-level regulation system based on a brain-computer interface, characterized in that, include: The signal sensing module is used to synchronously acquire the user's multimodal physiological signals and perform time-aligned preprocessing. The central decision-making module is used to extract and fuse features based on the preprocessed multimodal physiological signals to generate a state response index that represents the state of consciousness; compare the state response index with the dynamic activation threshold in real time, and output control commands to the stimulus execution module when the triggering conditions are met; The stimulus execution module, in response to the control command, applies multimodal coordinated physical stimulation to the user; The central decision-making module is further configured to adaptively adjust the stimulation parameters and / or the dynamic activation threshold of the multi-mode collaborative physical stimulus applied in the next cycle based on the changing trend of the state response index before and after the application of the physical stimulus.

2. The consciousness level regulation system based on brain-computer interface according to claim 1, characterized in that, The multimodal physiological signals include any two or more combinations of electroencephalogram (EEG) signals, electrocardiogram (ECG) signals, skin conductance signals, electromyogram (EMG) signals, blood oxygen saturation signals, and respiratory signals.

3. The consciousness level regulation system based on brain-computer interface according to claim 1, characterized in that, The central decision-making module generates a state response index that represents the state of consciousness, specifically including: Based on the multimodal physiological signals, the corresponding features are extracted and normalized to determine the corresponding sub-index features; Using the preset fusion weights of each sub-index feature, the sub-index features are weighted and summed to output the state response index; wherein the fusion weights of each sub-index feature satisfy the preset scaling and normalization conditions.

4. The consciousness level regulation system based on brain-computer interface according to claim 3, characterized in that, The fusion weights are determined through the following offline optimization steps: Multimodal physiological signals were acquired synchronously at multiple assessment time points, and a multimodal sub-index feature vector was constructed. Obtain the user's external reference evaluation score at the corresponding point in time; Establish a linear mapping model between the multimodal sub-index feature vector and the reference evaluation score; with the optimization objective of minimizing the mean square error between the inner product of the multimodal sub-index feature vector and the fusion weight and the reference evaluation score, solve for the optimal solution that satisfies the preset scaling and normalization condition of the fusion weight, and use it as the preset fusion weight.

5. The consciousness level regulation system based on brain-computer interface according to claim 3, characterized in that, The multimodal physiological signals include electroencephalogram (EEG) signals; When the central decision-making module calculates the brain electronics index characteristics, it is specifically used for: Calculate the ratio of the average power of the Theta band to the average power of the Beta band of the EEG signal within the current time window; Calculate the normalized permutation entropy value of the EEG signals within the current time window; Using preset harmonic weights, the reciprocal of the ratio and the normalized permutation entropy value are weighted and harmonicly calculated to obtain the brain electronic index feature, wherein the value of the brain electronic index feature is positively correlated with cortical arousal.

6. The consciousness level regulation system based on brain-computer interface according to claim 1, characterized in that, The signal sensing module is also used to collect multimodal physiological signals for a preset duration and perform time-aligned preprocessing in a resting state without the application of the physical stimulus. When the central decision-making module compares the state response index with the dynamic activation threshold in real time, it is specifically used for: Calculate the state response index time series corresponding to a preset duration under resting conditions; calculate the mean and standard deviation of the state response index within the state response index time series; During subsequent intervention, the real-time calculated state response index is converted into a Z-score based on the mean and standard deviation; the Z-score is truncated and linearly mapped to a scalar interval of 0 to 1 to obtain a normalized response level; according to the distribution of normalized response levels within the selected historical intervention window, the initial Z-score threshold is adjusted accordingly, and the adjusted Z-score threshold parameter is converted into a core dynamic variable through the same truncation and mapping rules to obtain a normalized Z-score threshold as the dynamic activation threshold; The current real-time normalized response level is compared with the normalized Z-score threshold.

7. The consciousness level regulation system based on brain-computer interface according to claim 1, characterized in that, The multimodal coordinated physical stimuli applied by the stimulus execution module include: primary continuous stimulation and secondary conditioned reinforcement stimulation; When the central decision-making module compares data in real time, the triggering condition is that the state response index exceeds the dynamic activation threshold and is maintained for a preset minimum duration. If the triggering condition is met, the user is determined to have entered an active response state, and then the corresponding control command is output to control the stimulus execution module to apply the secondary conditional reinforcement stimulus in real time while maintaining the primary continuous stimulus, thereby achieving time locking between the intervention signal and the brain's active response state.

8. The consciousness level regulation system based on brain-computer interface according to claim 7, characterized in that, The primary continuous stimulation is auditory stimulation derived from a personalized sound library and applied through an auditory stimulation device; the secondary conditioned reinforcement stimulation is peripheral nerve electrical stimulation and applied through an electrical stimulation device attached to a target point on the median nerve or auricular vagus nerve.

9. The consciousness level regulation system based on brain-computer interface according to claim 1, characterized in that, When the central decision-making module makes adaptive adjustments, it is specifically used to: take the increase in the state response index and / or the duration before and after a single triggering of the physical stimulus as a reward signal; If the reward signal is lower than the preset expectation, the intensity of the physical stimulus is increased or the stimulus content is switched within the safety threshold range, and the dynamic activation threshold is adaptively lowered. If the reward signal continues to exceed the preset expectation, the current stimulation parameters are maintained, and the dynamic activation threshold is increased to match the evolution of the user's state.

10. A method for regulating the level of consciousness based on a brain-computer interface system, characterized in that, The method of the brain-computer interface-based consciousness level regulation system as described in any one of claims 1 to 9, wherein the method is executed by the central decision-making module in the system, includes the following steps: receiving multimodal physiological signals synchronously collected from the user and preprocessed with time alignment; extracting and fusing features from the preprocessed multimodal physiological signals to generate a state response index characterizing the state of consciousness; comparing the state response index with a dynamic activation threshold in real time, and outputting a control command to the stimulus execution module to apply multimodal coordinated physical stimulation when the triggering condition is met; and adaptively adjusting the stimulation parameters and / or the dynamic activation threshold of the physical stimulation applied in the next cycle according to the changing trend of the state response index before and after the application of the physical stimulation.

Images