A personalized closed-loop transcranial electrical stimulation method and system based on electroencephalogram monitoring

Abstract

The application discloses a kind of individualized closed-loop transcranial electrical stimulation method and system based on electroencephalogram monitoring, comprising S1, individual threshold calibration, before stimulation, the attention threshold of individual is established: S1.1, system first acquires electroencephalogram, extracts attention-related electroencephalogram feature A (t);S1.2, before formal regulation, first do calibration, under the condition of quiet rest, active concentration task, optional distraction, baseline is collected, and the concentration threshold limit of the person is calculated;S2, real-time monitoring, in the stimulation, the electroencephalogram of frontal lobe and other regions is continuously monitored, and the state such as not concentrated, transition, concentrated is distinguished, after entering regulation, the current feature is constantly updated by system;S3, multi-state classification and control mapping, compare personal threshold with multi-state classification and generate control mapping instruction;S4, closed-loop stimulation control, stimulation control module receives control mapping instruction, drives tES output, effectively solve the problem that existing open-loop fixed output effect is poor, group threshold misjudgment rate is high, state switching is frequent and safety protection is insufficient.

Description

Technical Field

[0001] This invention relates to the field of neuromodulation technology, specifically a personalized closed-loop transcranial electrical stimulation method and system based on electroencephalogram (EEG) monitoring. Background Technology

[0002] Transcranial electrical stimulation (TES), a non-invasive neuromodulation technique, modulates cortical excitability by applying weak currents to specific brain regions and has been widely used as an adjunct therapy for neuropsychiatric disorders such as attention deficit, cognitive impairment, and mood disorders. However, current TES techniques still have significant shortcomings in practical applications: First, most existing devices employ an open-loop fixed output mode, meaning they provide a constant output according to preset current intensity, frequency, and duration. This lack of dynamic adjustment based on the user's real-time brain function state leads to a mismatch between stimulation timing and the user's actual needs, resulting in poor stimulation effects and a high risk of tolerance. Second, while some existing closed-loop schemes incorporate EEG feedback mechanisms, they often use a uniform group threshold as the standard for state assessment. Due to significant differences in baseline EEG values ​​between individuals, directly applying the group threshold can easily misinterpret normal fluctuations as a decline in attention, and may also lead to other issues. Third, existing solutions often fail to detect genuine attention deficits, leading to frequent errors in stimulus on / off cycles, impacting user experience and treatment effectiveness. Fourth, current solutions primarily use simple on / off states for stimulus output, lacking a smooth transition mechanism. When attention fluctuates around a threshold, frequent switching of stimulus output not only causes subjective discomfort but may also lead to electrical surges due to frequent on / off cycles, posing a safety hazard. Fifth, existing systems lack sufficient monitoring capabilities for EEG signal quality. When electrode contact is poor or signals are interfered with, the system still relies on unreliable characteristic values ​​to determine the state and output stimulation, potentially leading to adverse consequences from blindly applying stimulation. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a transcranial electrical stimulation method and system that can perform personalized threshold calibration based on an individual's EEG baseline, perform multi-state hierarchical closed-loop regulation based on real-time EEG status, and have a sound safety protection mechanism, which can solve the problems in the prior art.

[0004] This invention is achieved through the following technical solution: A personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring, comprising: S1, Individual Threshold Calibration: Establishing an individual's attention threshold before stimulation. S1.1, The system first collects EEG data and extracts attention-related EEG features A(t); S1.2, Before formal adjustment, calibration is performed. Baselines are collected under conditions of quiet rest, active focus on tasks, and optional distraction, and the person's attention threshold is calculated. S2, real-time monitoring, continuously monitors EEG in areas such as the prefrontal cortex during stimulation, distinguishes between three attentional states: inattentive, over-attentive, and focused. After entering the closed-loop regulation stage, the system continuously updates real-time EEG characteristic values. S3, Multi-state hierarchical and control mapping, compares the individual threshold with the multi-state hierarchical and generates control mapping instructions; S4, closed-loop stimulation control: The stimulation control module receives control mapping instructions and drives the tES (transcranial electrical stimulation) output. When the system determines that the stimulation is not focused, it turns on tDCS (transcranial direct current stimulation); when the system determines that the stimulation is too low, it maintains a low output and shortens the duty cycle; when the system determines that the stimulation is too high, it stops stimulation or gradually reduces the current to zero. The entire process can be superimposed with impedance detection, overcurrent protection, maximum continuous stimulation time limit, and emergency stop.

[0005] Further technical solutions include attention-related EEG features A(t) such as: the θ power and β power of the prefrontal electrodes, or their ratio (e.g., θ / β); the α power of the prefrontal electrodes, or its change in the task state relative to the resting state; a combination index of prefrontal electrode engagement and arousal; entropy, kurtosis, etc. of the prefrontal electrodes, and a comprehensive score weighted by signal quality; and the probability of attentional state given by the classifier of the prefrontal electrodes.

[0006] Further technical solutions, such as directly applying group thresholds, may misjudge normal fluctuations as non-concentration and miss abnormal fluctuations. Therefore, individual calibration should be performed before formal closed-loop control. Individual threshold calibration includes: Resting state: Keep eyes open or closed and remain quiet for a few minutes to collect baseline distribution; Focused task state: Perform sustained attention tasks (continuous operation, reading, reaction time, etc.) and collect the feature distribution during periods of relative focus; Optional Distraction Conditions: Estimate the boundary of the "disconcentration" side when attention is briefly lowered due to interference or dual tasks, provided that ethical and safety conditions permit. Generate personal thresholds: When focused on a task, the median of A(t) is M_focus; when resting, the median of A(t) is M_rest; when focused, the threshold of A(t) is T_high; when close to the focused side, T_high = M_focus - k1·σ; the non-focused threshold is T_low; when close to the resting or distracted side, T_low = M_rest + k2·σ; the transition zone, located between T_low and T_high, is used to reduce back-and-forth jitter near the threshold, k1 and k2 are system constants, and σ is the standard deviation of the corresponding feature.

[0007] Further technical solutions, including multi-state hierarchical control mapping, include: In a non-focused state, if A(t) falls on the non-focused side and persists for more than time period W1, tDCS (transcranial direct current stimulation) is initiated; if it is already outputting, it can be maintained, or gradually increased to the target intensity within the safe upper limit; During the transition state, when A(t) is in the transition band, or when it briefly oscillates between non-concentrated and concentrated states, maintain the current output, reduce the duty cycle, or enter the observation countdown to avoid frequent switching. In a concentrated state, A(t) falls on the concentrated side and continues for more than the time period W3, triggering a stop, or first linearly reducing the flow and then stopping, to alleviate subjective discomfort.

[0008] A further technical solution is to suppress noise and chattering. After entering the non-concentrated state, the output should be maintained for at least T1 seconds before being allowed to be turned off due to brief fluctuations; after entering the concentrated state, the output should be stopped for at least T2 seconds before being allowed to be turned on again. T1 and T2 can be set to 5 seconds.

[0009] A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring includes: The EEG acquisition module acquires EEG data in real time; it prioritizes sampling the prefrontal cortex but can also sample other areas; it includes amplification, filtering, analog-to-digital conversion, electrode impedance monitoring, and timestamps; sampling rate examples: 250 Hz, 500 Hz, or 1000 Hz. The preprocessing and feature analysis module performs power frequency notch filtering, bandpass filtering, bad segment removal, and electrooculography / electromyography inhibition on the raw signals sampled by the EEG acquisition module, and then calculates A(t) within a sliding window; the window length is 1-5 seconds in example, and the step size is 0.5-1 seconds in example; then the feature value, signal quality score and timestamp are output. The individual threshold calibration module runs the calibration process, stores individual baselines, and generates thresholds; the results are bound to the subject identifier and can be reused in subsequent sessions or recalibrated as needed.

[0010] The state determination module compares real-time features with individual thresholds, and combines hysteresis and time windows to output a non-centralized state, a centralized state, or a transitional state.

[0011] The stimulation control module generates control commands based on status, safety rules, and prescription settings. It is responsible for ramp raising and lowering, minimum switching interval, cumulative duration or cumulative power consumption statistics, and writing event logs.

[0012] The tES output module performs electrical stimulation and features a constant current tDCS source with interfaces for stimulation and return electrodes. It can also utilize existing tACS / tRNS waveform technology and includes current monitoring and abnormal interruption capabilities.

[0013] The safety and human-computer interaction module provides emergency stop, impedance over-limit alarm, overcurrent / open circuit protection, maximum continuous stimulation time limit, daily cumulative limit, and discomfort feedback button; the interface displays the current status, whether stimulation is in progress, remaining stimulation time, and calibration wizard.

[0014] The data storage and configuration module saves calibration parameters, session logs, anonymization characteristic curves, and device configurations to facilitate review and algorithm iteration; when personal information is involved, it is handled in accordance with applicable privacy and ethical requirements.

[0015] Further technical solutions, The beneficial effects of this invention are as follows: First, the personalized closed-loop transcranial electrical stimulation method and system based on EEG monitoring provided by this invention establishes personalized attention threshold boundaries by calibrating individual thresholds for users under resting, focused task, and optional distraction conditions before formal closed-loop modulation. It then performs multi-state hierarchical comparisons based on real-time monitored EEG characteristics and individual thresholds to generate corresponding control mapping commands to drive the transcranial electrical stimulation output. Simultaneously, it sets up a smooth transition state processing mechanism and multiple safety protection rules, achieving precise matching between the stimulation output and the user's real-time brain function state. This effectively solves the problems of poor output performance, high group threshold misjudgment rate, frequent state switching, and insufficient safety protection in existing open-loop fixed output methods, significantly improving the personalization, modulation accuracy, and safety of transcranial electrical stimulation.

[0016] Second, by calibrating individual thresholds before formal intervention, baseline EEG data of users are collected under quiet rest, active focus, and optional distraction conditions. Attention-related EEG features are extracted and their medians and standard deviations are calculated. This generates personalized concentration and non-concentration thresholds, making the state judgment criteria based on the user's own EEG feature distribution rather than a uniform group threshold. This effectively avoids misjudgments and omissions caused by individual differences in EEG baselines, and improves the accuracy of state judgment and the targeting of intervention.

[0017] Third, by setting up a multi-state hierarchical and control mapping mechanism, real-time EEG characteristics are compared with individual thresholds and divided into three levels: inattentive state, transitional state, and focused state, each corresponding to a different stimulation control strategy: in the inattentive state, transcranial direct current stimulation is initiated or maintained; in the transitional state, the current output is maintained or the duty cycle is reduced to avoid frequent switching; in the focused state, stimulation is stopped or the current is gradually reduced to zero. This allows the stimulation output to be dynamically graded and adjusted according to the user's real-time attention level, rather than a simple on / off switch, effectively reducing stimulation jitter during state fluctuations and improving user comfort and control smoothness.

[0018] Fourth, by introducing a hysteresis mechanism and time window constraints in state switching, the system sets a minimum output duration after entering a non-concentrated state before allowing it to be turned off, and a minimum stop duration after entering a concentrated state before allowing it to be turned on. This effectively suppresses the jitter phenomenon caused by EEG signal noise, avoids frequent start-stop of stimulation due to short-term fluctuations, makes the stimulation output more stable and reliable, and reduces the user's subjective discomfort.

[0019] Fifth, the system synchronously calculates the quality of EEG signals during real-time monitoring. When the signal quality is unqualified, it pauses the state switching and issues an alarm, avoiding blind stimulation. This effectively prevents inappropriate stimulation output caused by factors such as poor electrode contact, electromyography interference, or electrooculography artifacts, ensuring the reliability and safety of closed-loop control. Attached Figure Description

[0020] For ease of explanation, the present invention will be described in detail below with reference to specific embodiments and accompanying drawings.

[0021] Figure 1 This is a block diagram of a personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring according to the present invention. Figure 2 A schematic diagram of the individual threshold calibration process; Figure 3 This is a schematic diagram of the closed-loop control process. Detailed Implementation

[0022] like Figures 1-3 As shown, the present invention will be described in detail. For ease of description, the directions mentioned below are defined as follows: the directions of up, down, left, right, front, and back mentioned below are the same as... Figure 1 The projection relationships are consistent in the up, down, left, right, front, and back directions.

[0023] Example 1: A personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring according to the present invention includes: A1, Preparation Stage: A1.1, Establish subject files; A1.2, Wear EEG (electroencephalography) and tES (transcranial electrical stimulation) electrodes to confirm that the impedance is acceptable; A1.3, Select stimulation prescription, electrode montage, maximum current, longest session duration, state strategy level, etc.; A2, Individual Attention Threshold Calibration: A2.1, adopt a resting state for several minutes; A2.2, focus on the task for several minutes; A2.3, Optional: Short-term interference condition; A2.4, calculate T_low, T_high, and the transition band; A2.5, after confirmation or fine-tuning, lock and enter closed loop; A3, Closed-loop regulation: A3.1 calculates A(t) and signal quality in real time. If the signal quality is not up to standard, it pauses the state switching, only issues an alarm, and does not blindly add stimulation. A3.2, determine the state of non-concentration, concentration, and transition; A3.3, In a decentralized state: Start or maintain tDCS; A3.4, In a concentrated state: stop, or reduce flow first and then stop; A3.5, During the transition state: Execute the transition strategy; A3.6, if any safety condition is triggered, the output will stop immediately; A3.7 After the session termination condition is met, the flow will be reduced normally and the data will be saved.

[0024] Example 2, based on Example 1, further specifies the following: During the prefrontal cortex monitoring and the start / stop closed loop of the anodic tDCS, EEG: collects Fp1 and Fp2, and references the earlobe, mastoid process, or Fpz; sampling rate 500Hz.

[0025] Features: The power ratio R = θ / β of θ (4-7Hz) and β (13-30Hz) is calculated every 2-second window and weighted according to signal quality; the increase of R when attention decreases is just a common direction; the direction can also be automatically determined according to the calibration data during implementation.

[0026] Calibration: After a 3-minute rest and a 3-minute continuous attention task, the 25%, 50%, and 75% quantiles of their respective power ratios R are used to generate T_low and T_high.

[0027] tDCS: Anode F3, cathode F4; target current 1.0mA; current rise / fall 15-30 seconds each; maximum continuous output per session 10 minutes; maximum cumulative stimulation per session 20 minutes.

[0028] Control: If R is higher than T_low for 5 consecutive seconds, check S1 and start outputting; if R is lower than T_high for 5 consecutive seconds, check S3 and stop outputting; if R is between the two, check S2, maintain the previous valid action and open a 10-second observation window.

[0029] Example 3, based on Example 2, further defines the following: when performing multi-state grading and intensity adjustment, it is not just about turning on / off; the non-concentrated state corresponds to 1.0mA, the transitional state corresponds to 0.5mA or 50% duty cycle intermittent output, and the concentrated state S3 corresponds to 0mA.

[0030] Example 4, based on Example 3, further specifies the following: During the calibration phase, data with task labels is collected, and a lightweight model (logistic regression, linear discriminant analysis, small-scale gradient boosting, or an online-updable probabilistic model) is trained to output the "disaggregated probability" P. Individual probability boundaries P_low and P_high are used instead of T_low and T_high, while the rest of the control framework remains unchanged.

[0031] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions conceived without creative effort should be included within the scope of protection of the present invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring, characterized in that, include: S1, Individual Threshold Calibration: Before stimulation, an individual's attention threshold is established: The system first collects EEG data, extracts attention-related EEG features A(t), and collects baseline data for resting state and active focus task state respectively. The baseline for distracted state can be selected. Based on the baseline distribution, the individualized non-concentration threshold T_low and concentration threshold T_high are calculated, and a transition zone is formed between T_low and T_high. S2, Real-time EEG monitoring stage: During the closed-loop regulation process, EEG signals are continuously collected, EEG characteristics A(t) are calculated in real time, and the quality of EEG signals is evaluated. When the signal quality is substandard, pause the state determination and stimulus adjustment and output an alarm. S3, Multi-state hierarchical determination: The real-time EEG feature A(t) is compared with the personalized threshold, and combined with the duration constraint, the states of inattentiveness, transition, and concentration are divided. S4, graded closed-loop stimulation control: Execute the corresponding transcranial electrical stimulation strategy according to the judgment result: When the state is judged to be non-concentrated, start or maintain the transcranial direct current stimulation output; when the state is judged to be transitional, adopt a buffer strategy of maintaining output and reducing output duty cycle to avoid frequent stimulation switching; when the state is judged to be concentrated, stop stimulation or cut off the output after linear decurrent. The entire control process synchronously executes impedance detection, overcurrent protection, maximum stimulation time limit constraint, and emergency stop safety logic.

2. The personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring according to claim 1, characterized in that: Attention-related EEG features A(t) include: the θ power, β power, or the ratio θ / β of the prefrontal electrodes; the α power of the prefrontal electrodes, or the amount of change of the α power in the task state relative to the resting state.

3. The personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring according to claim 1, characterized in that: Individual threshold calibration includes: Resting state: Keep eyes open or closed and remain quiet for a few minutes to collect baseline distribution; Focused task state: Performing sustained attention tasks and collecting feature distributions during periods of relative focus; Optional distraction condition: When attention is briefly lowered due to interference or dual tasks, estimate the boundary of the "disengagement" side; Generate personal thresholds: When focused on a task, the median of A(t) is M_focus; when resting, the median of A(t) is M_rest; when focused, the threshold of A(t) is T_high; when close to the focused side, T_high = M_focus - k1·σ; the non-focused threshold is T_low; when close to the resting or distracted side, T_low = M_rest + k2·σ; the transition zone, located between T_low and T_high, is used to reduce back-and-forth jitter near the threshold, k1 and k2 are system constants, and σ is the standard deviation of the corresponding feature.

4. The personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring according to claim 1, characterized in that: Multi-state hierarchical and control mapping includes: In the non-concentrated state, if A(t) falls on the non-concentrated side and continues for more than the time period W1, transcranial direct current stimulation (tDCS) is initiated; if it is already outputting, it is maintained, or gradually increased to the target intensity within the safe upper limit. During the transition state, when A(t) is in the transition band, or when it briefly oscillates between non-concentrated and concentrated states, maintain the current output, reduce the duty cycle, or enter the observation countdown to avoid frequent switching. In a concentrated state, A(t) falls on the concentrated side and continues for more than the time period W3, triggering a stop, or first linearly reducing the current and then stopping.

5. A personalized closed-loop transcranial electrical stimulation method based on electroencephalogram (EEG) monitoring according to claim 4, characterized in that: After entering a non-concentrated state, output for at least T1 seconds before being allowed to be turned off due to brief fluctuations; after entering a concentrated state, stop for at least T2 seconds before being allowed to be turned on again.

6. A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring, characterized in that, include: The EEG acquisition module acquires EEG data in real time, including amplification, filtering, analog-to-digital conversion, electrode impedance monitoring, and timestamps. The preprocessing and feature analysis module performs power frequency notch filtering, bandpass filtering, bad segment removal, and electrooculography / electromyography inhibition on the raw signals sampled by the EEG acquisition module. Then, it calculates A(t) within a sliding window and outputs feature values, signal quality scores, and timestamps. The individual threshold calibration module runs the calibration process, stores individual baselines, and generates thresholds; the results are bound to the subject identifier and can be reused in subsequent sessions or recalibrated as needed. The state determination module compares real-time features with individual thresholds, combines hysteresis and time windows, and outputs a non-centralized state, a centralized state, or a transitional state. The stimulation control module generates control commands based on status, safety rules, and prescription settings to achieve ramp raising and lowering, minimum switching interval, cumulative duration or cumulative power consumption statistics, and writes event logs. The tES output module performs electrical stimulation and features a constant current tDCS source with interfaces for stimulation and return electrodes.

7. A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring according to claim 6, characterized in that: EEG acquisition module, sampling the prefrontal cortex.

8. A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring according to claim 6, characterized in that: The preprocessing and feature analysis module calculates A(t) within a sliding window, with an example window length of 1-5 seconds and an example step size of 0.5-1 seconds, and then outputs the feature value, signal quality score, and timestamp.

9. A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring according to claim 6, characterized in that: It also includes a safety and human-computer interaction module, providing emergency stop, impedance over-limit alarm, overcurrent / open circuit protection, maximum continuous stimulation time limit, daily cumulative limit, and discomfort feedback button. The interface displays the current status, whether stimulation is in progress, remaining stimulation time, and calibration wizard.

10. A personalized closed-loop transcranial electrical stimulation system based on electroencephalogram (EEG) monitoring according to claim 6, characterized in that: It also includes a data storage and configuration module to save calibration parameters, session logs, anonymized characteristic curves, and device configurations.

Images