A closed-loop neuromodulation method based on emotional mismatch biomarkers
Patent Information
- Application Number
- CN202611194612.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-07
- Publication Date
- 2026-10-27
AI Technical Summary
[0004]现有方案的生物标志物大多为单一脑区活动或单一指标，无法区分受试者当前情绪是否与其所处的客观情境相称，无法捕捉到“主观相对客观的背离”，导致刺激时机与靶点的特异性不足、易误触发、副作用较多、疗效受限
[0016]本发明技术方案提供一种基于情绪失配生物标志物的闭环神经调控方法，对同一预处理后神经信号进行通道筛选并特征提取，拆分出客观特征和主观特征；基于有向增强脑区特征进行解码，得到解码后客观特征和解码后主观特征，基于二者之间的背离量作为情绪失配生物标志物，若情绪失配生物标志物超过个体化阈值并持续预设时长，则生成触发信号，并依据有向增强脑区特征确定刺激靶点与刺激参数；基于触发信号、刺激靶点与刺激参数生成对应电荷平衡双相脉冲串的刺激控制信号，并根据跨触点高频刺激伪迹信号确定实际刺激位点，当失配回落时生成减弱或终止的刺激控制信号，反之生成维持或调整的刺激控制信号，解决无法捕捉到主观相对客观的背离的问题，导致刺激时机与靶点的特异性不足、易误触发、副作用较多、疗效受限的问题。
Smart Images
Figure CN122973802A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of biomedical engineering and medical technology, specifically relating to a closed-loop neural regulation method based on emotion mismatch biomarkers. Background Technology
[0002] The core of the neural mechanisms of affective disorders such as depression, anxiety, and post-traumatic stress disorder is that subjective emotional experience is detached from the objective stimulus situation.
[0003] In clinical practice, electrical stimulation is often triggered or adjusted based on biomarkers obtained from the subject's neural signals, thereby achieving closed-loop (adaptive / responsive) neuromodulation of mismatched biomarkers.
[0004] Most existing biomarkers are single brain region activities or single indicators, which cannot distinguish whether the subject's current emotion is commensurate with the objective situation, and cannot capture the "deviation between subjective and objective", resulting in insufficient specificity of stimulation timing and target, easy false triggering, more side effects, and limited efficacy. Summary of the Invention
[0005] To address the aforementioned technical problems, the present invention provides a closed-loop neural modulation method based on emotion mismatch biomarkers, comprising the following steps:
[0006] Channel filtering and feature extraction are performed on the same preprocessed neural signal to separate objective and subjective features. Decoding is performed based on the features of the directed enhanced brain region to obtain the decoded objective features and the decoded subjective features; The deviation between the decoded objective features and the decoded subjective features is calculated as an emotion mismatch biomarker. If the emotion mismatch biomarker exceeds the individualization threshold and continues for a preset duration, a trigger signal is generated, and the stimulation target and stimulation parameters are determined based on the characteristics of the directed enhancement brain region. Stimulation control signals are generated based on trigger signals, stimulation targets, and stimulation parameters to form corresponding charge-balanced biphasic pulse trains. The actual stimulation sites are determined based on cross-contact high-frequency stimulation artifact signals. Changes in emotional mismatch biomarkers are monitored during stimulation pulse intervals or in windows after artifact removal. When mismatch decreases, a weakened or terminated stimulation control signal is generated; otherwise, a maintained or adjusted stimulation control signal is generated.
[0007] Preferably, the targeted enhancement brain region features include objective enhancement from the lateral amygdala to the anterior insula and subjective enhancement from the posterior insula to the striatum.
[0008] Preferably, based on multiple indications such as treatment-resistant depression, anxiety disorder, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, characteristic parameters of the directed enhancement brain region are determined, and the characteristics of the directed enhancement brain region are updated based on the characteristic parameters of the directed enhancement brain region.
[0009] Preferably, stimulation control signal modulation parameters are determined based on multiple indications such as treatment-resistant depression, anxiety disorder, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, and the stimulation control signal is updated based on the stimulation control signal modulation parameters.
[0010] Preferably, the objective feature's corresponding early time window is earlier than the subjective feature's corresponding late time window.
[0011] Preferably, the objective features include the stimulus salience feature and the quadratic feature of the absolute price value, and the subjective features include the cumulative integral feature corresponding to the late time window.
[0012] Preferably, the directional enhancement brain region features are obtained based on a nonlinear phase coupling metric.
[0013] Preferably, the nonlinear phase coupling metric employs directional phase transfer entropy or symbol transfer entropy.
[0014] Preferably, the directed phase transfer entropy is calculated in the theta frequency band.
[0015] Preferably, the formula for calculating the deviation amount is as follows: M=max(0,( S) max(0, O)) Where O represents the objective features after decoding, and S represents the subjective features after decoding.
[0016] This invention provides a closed-loop neural modulation method based on emotion mismatch biomarkers. It involves channel screening and feature extraction of the same preprocessed neural signal, separating objective and subjective features. Decoding is performed based on features of directed-enhanced brain regions, yielding decoded objective and subjective features. The deviation between these features is used as an emotion mismatch biomarker. If the emotion mismatch biomarker exceeds an individualized threshold and persists for a preset duration, a trigger signal is generated. The stimulation target and parameters are determined based on the directed-enhanced brain region features. A stimulation control signal for a corresponding charge-balanced biphasic pulse train is generated based on the trigger signal, stimulation target, and stimulation parameters. The actual stimulation site is determined based on cross-contact high-frequency stimulation artifact signals. When the mismatch subsides, a weakened or terminated stimulation control signal is generated; conversely, a maintained or adjusted stimulation control signal is generated. This method addresses the problem of failing to capture the deviation between subjective and objective features, leading to insufficient specificity of stimulation timing and target, easy false triggering, numerous side effects, and limited therapeutic efficacy. Attached Figure Description
[0017] Figure 1 This is a schematic diagram of a closed-loop neural modulation method based on emotion mismatch biomarkers. Figure 2 A schematic diagram showing the curves of objective emotion components and subjective emotion components; Figure 3 This is a schematic diagram of brain regions and directed routes. Figure 4 A schematic diagram of a charge-balanced two-phase pulse train; Figure 5 This is a schematic diagram of a directed routing process. Detailed Implementation
[0018] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.
[0019] like Figure 1 As shown in the figure, an embodiment of the present invention provides a closed-loop neural modulation method based on emotion mismatch biomarkers for the neural modulation of mental disorders and chronic pain, comprising the following steps: S1 acquires neural signals from multiple candidate brain regions (including the lateral amygdala as an objective source, the posterior insula as a subjective source, and directional enhancement brain regions such as the anterior insula and striatum, as well as candidate stimulation target regions) at a sampling rate of 2 kHz. Power frequency notch filtering and in-rod bipolar heavy reference are applied to the signal to suppress common-mode interference.
[0020] S2: Preprocessing of neural signals to obtain preprocessed neural signals, including resampling, high-frequency activity envelope extraction and decibel normalization, as well as channel and sample quality screening and missing value labeling.
[0021] Preprocessing includes: resampling the acquired signal to a broadband bandwidth of approximately 1000 Hz; extracting the high-frequency activity (HFA, 70–150 Hz) envelope and normalizing the pre-stimulation baseline in decibels; and triple-channel quality screening. (i) Sample amplitudes exceeding the range (e.g., |x|>1×10) 6 Channels with a percentage greater than 1% (μV) are classified as clipping channels and are removed. (ii) Channels with a maximum absolute value of the robustness standard score of HFA greater than 5 or a kurtosis greater than 10 are classified as dirty channels and removed. (iii) On each frequency subband, samples that deviate from the median absolute deviation (MAD) by more than 2.5 times the MAD are identified as outliers and marked as missing. The missing marks are propagated along the processing chain.
[0022] After multi-channel preprocessing, neural signals that passed quality screening were divided according to their source brain regions: objective emotion components were obtained from channels in objective source brain regions (lateral amygdala), and subjective emotion components were obtained from channels in subjective source brain regions (posterior insula).
[0023] Feature extraction is performed on the objective and subjective emotion components to separate objective and subjective features.
[0024] like Figure 2 As shown, the early time window corresponding to the objective emotion component is earlier than the late time window corresponding to the subjective emotion component. The early time window is located on the order of approximately 0.1 seconds after the stimulus, while the late time window is located on the order of approximately 0.7 seconds after the stimulus.
[0025] The objective features are those calculated within the early time window, including the stimulus salience features and the quadratic features (U-shaped) of the absolute value of the price; the subjective features are those calculated within the late time window, including the cumulative integral features of high-frequency activity accumulated over time.
[0026] The stimulus saliency feature refers to the increase in high-frequency activity relative to the pre-stimulus baseline (in decibels), used to characterize the saliency of the corresponding brain region's response to the stimulus event; the quadratic feature of the absolute value of the valence refers to the quadratic term after taking the absolute value of the emotional valence of the stimulus. Since objective encoding increases both the positive and negative ends of emotional arousal, and positive valence is difficult to decode linearly by sign, its saliency is characterized by the quadratic term of the absolute value of the valence (U-shaped); the emotional valence is given by the mapping from high-frequency activity in the early time window to valence, obtained by training with a stimulus paradigm labeled with valence during the individualized calibration phase. During inference, the valence estimate is obtained by applying this mapping to the current high-frequency activity in the early time window, and then taking the quadratic term of the absolute value.
[0027] The cumulative integral feature refers to the amount obtained by integrating the high-frequency activity envelope over time within the late time window, which is used to characterize the time accumulation of subjective experience.
[0028] S3 and S4: Based on the features of the directed enhancement brain region, the objective and subjective features are decoded respectively to obtain the decoded objective features and the decoded subjective features.
[0029] like Figure 3As shown, the objective emotional component corresponds to the objective source brain region 12 (i.e., the first brain region), and the subjective emotional component corresponds to the subjective source brain region 13 (i.e., the second brain region). The directional enhancement brain region features include the lateral amygdala 12 to the anterior insula 14 for enhancing objectivity and the posterior insula 13 to the striatum 15 for enhancing subjectivity.
[0030] The directional enhancement brain region features are obtained by nonlinear phase coupling measurement. The phase transfer direction of the brain region is estimated in the theta band (the neural oscillation band of about 4-8 Hz) using directional phase transfer entropy. The number of phase regions is 4 and the number of substitution data is 500. The directionally stable directional connection edges are preserved, so as to stably obtain the direction from the lateral amygdala to the anterior insula for objective enhancement and the direction from the posterior insula to the striatum for subjective enhancement.
[0031] Decoding further includes: under the constraint of the directed connection (lateral amygdala to anterior insula) used to enhance objectivity, performing linear discrimination or regression on the objective features to obtain the decoded objective features O; under the constraint of the directed connection (posterior insula to striatum) used to enhance subjectivity, decoding the subjective features to obtain the decoded subjective features S; the "constraint" is to include the high-frequency activity of the downstream brain region (anterior insula) in the corresponding time window, or the phase transmission intensity of the directed connection edge, as an additional input or weighting term into the objective decoder. Similarly, the subjective path includes the striatal activity or the intensity of the posterior insula to striatum edge into the subjective decoder, so that the decoded features are given by the joint activity of upstream and downstream, rather than determined by a single source brain region.
[0032] S5: Calculate the deviation between the decoded objective features and the decoded subjective features as an emotion mismatch biomarker. If the emotion mismatch biomarker does not meet the preset triggering conditions, i.e. does not exceed the individualized threshold and continues for a preset duration, then continue the detection.
[0033] The deviation is defined as M=max(0,( S) max(0, O)) calculation, where O is the objective feature after decoding and S is the subjective feature after decoding.
[0034] The deviation M = max(0,( S) max(0, O)) represents the deviation of subjective negativity from the objective situation: when the objective emotion is non-negative ( O≤0) and subjective emotion is negative ( When S>0, M>0, and the larger M is, the stronger the deviation between subjective negativity and objective non-negativity; based on this, the individualized threshold and duration are used to jointly determine whether to generate a trigger signal.
[0035] S6: If the preset triggering conditions are met, a trigger signal is generated, and the stimulation target and stimulation parameters are determined based on the characteristics of the directed enhancement brain region.
[0036] Targeting is performed according to the direction of the directed connections: with the goal of reshaping the direction of the abnormal circuit, the upstream or downstream brain regions with the corresponding directed enhancement of brain region features are selected as the stimulation target points, which are stable in direction and have the strongest association with the pathological circuit.
[0037] S7: Generates stimulation control signals based on trigger signals, stimulation targets, and stimulation parameters.
[0038] The stimulation control signal generated by S7 corresponds to a biphasic pulse train with charge balance, such as a frequency of 50 Hz and a pulse width of 0.5 ms. The train length and amplitude are determined by individualized calibration to ensure that the net charge is zero and to avoid electrode polarization.
[0039] S8: Generate the stimulation control signal for the corresponding charge-balanced biphasic pulse train (e.g., 50 Hz, pulse width 0.5 ms, etc.). Figure 4 As shown in the figure, the actual stimulation site is confirmed based on the cross-contact high-frequency stimulation artifact signal. The changes in mismatch biomarkers are monitored during the stimulation pulse interval or the window after artifact removal. When the mismatch falls back, a weakened or terminated stimulation control signal is generated, and vice versa, a maintenance or adjustment stimulation control signal is generated.
[0040] In S3, based on multiple indications such as treatment-resistant depression, anxiety disorders, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, the characteristic parameters of the directed enhancement brain regions are determined, and the characteristics of the directed enhancement brain regions are updated based on the characteristic parameters of the directed enhancement brain regions.
[0041] In S7, based on multiple indications such as treatment-resistant depression, anxiety disorders, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, stimulation control signal modulation parameters are determined, and stimulation control signals are updated based on stimulation control signal modulation parameters.
[0042] This invention also provides a closed-loop neural modulation system based on emotion mismatch biomarkers, employing the closed-loop neural modulation method based on emotion mismatch biomarkers as described above. The system includes: Signal acquisition module 1 acquires neural signals.
[0043] Preprocessing module 2 preprocesses the neural signals to obtain preprocessed neural signals.
[0044] The channel quality screening module 11 performs channel quality screening on the preprocessed neural signals and outputs multi-channel signals that have passed the screening; the objective emotion component and the subjective emotion component are obtained from the channels of the objective source brain region and the subjective source brain region, respectively.
[0045] Feature extraction module 3 extracts features from the objective and subjective emotion components, separating them into objective and subjective features.
[0046] Directed connectivity module 6 stores directed enhanced brain region features obtained based on nonlinear phase coupling metrics. The nonlinear phase coupling metrics employ directed phase transfer entropy (dPTE) or symbol transfer entropy (STE).
[0047] Objective emotion decoding module 4 decodes the objective emotion component of objective features based on the features of the directed enhanced brain region, and obtains the decoded objective features.
[0048] Subjective emotion decoding module 5 decodes the subjective emotion component of subjective features based on the features of the directed enhancement brain region, and obtains the decoded subjective features.
[0049] The mismatch biomarker module 7 uses the deviation between the objective features and subjective features after decoding as an emotion mismatch biomarker. If the emotion mismatch biomarker does not meet the preset triggering conditions, i.e. does not exceed the individualized threshold and continues for a preset duration, the detection continues. If the preset triggering conditions are met, a trigger signal is generated, and the stimulation target and stimulation parameters are determined based on the characteristics of the directed enhancement brain region.
[0050] Stimulation decision module 8 generates stimulation control signals based on trigger signals, stimulation targets, and stimulation parameters.
[0051] Stimulation module 9 generates stimulation control signals based on corresponding charge balance biphasic pulse trains and confirms the actual stimulation sites based on cross-contact high-frequency stimulation artifact signals. It monitors changes in mismatch biomarkers during stimulation pulse intervals or in windows after artifact removal. When mismatch falls back, it generates a weakened or terminated stimulation control signal; otherwise, it generates a maintained or adjusted stimulation control signal.
[0052] The individualized calibration module 10 is used to store the characteristic parameters of the directed enhanced brain regions and the stimulation control signal modulation parameters corresponding to various indications such as treatment-resistant depression, anxiety disorders, post-traumatic stress disorder, obsessive-compulsive disorder and chronic pain. The objective emotion decoding module 4 updates the characteristics of the directed enhanced brain regions based on the characteristic parameters of the directed enhanced brain regions, and the stimulation decision module 8 updates the stimulation control signals based on the stimulation control signal modulation parameters.
[0053] The embodiments of this application directly capture the core of maladaptation, namely "subjective emotional experience deviating from the objective stimulus situation," and precisely trigger and target neural modulation accordingly, thereby improving the specificity and targeting of closed-loop stimulation and reducing false triggering and side effects.
[0054] like Figure 5As shown in Example 1 (treatment-resistant depression): This example applies the method to signals collected by intracranial electrodes covering the lateral amygdala, posterior insula, ventral / ventral striatum of the internal capsule, and infragenual cingulate gyrus. When mismatch biomarkers exceed individualized thresholds and persist for a preset duration, the infragenual cingulate gyrus, ventral / ventral striatum of the internal capsule, etc., are identified as stimulation targets based on directed connectivity, generating corresponding stimulation control signals. Compared to triggering only by exceeding the power threshold of a single brain region's frequency band, this example can generate stimulation control signals more specifically when pathological emotional states occur.
[0055] Example 2 (Non-invasive variant): The two components are approximated by the spatiotemporal pattern localized or learned by scalp EEG / MEG combined source localization and mismatch is calculated; the accuracy is compromised but the applicability is wider.
Claims
1. A closed-loop neural modulation method based on emotion mismatch biomarkers, characterized in that, Includes the following steps: Channel filtering and feature extraction are performed on the same preprocessed neural signal to separate objective and subjective features. Decoding is performed based on the features of the directed enhanced brain region to obtain the decoded objective features and the decoded subjective features; The deviation between the decoded objective features and the decoded subjective features is calculated as an emotion mismatch biomarker. If the emotion mismatch biomarker exceeds the individualization threshold and continues for a preset duration, a trigger signal is generated, and the stimulation target and stimulation parameters are determined based on the characteristics of the directed enhancement brain region. Stimulation control signals are generated based on trigger signals, stimulation targets, and stimulation parameters to form corresponding charge-balanced biphasic pulse trains. The actual stimulation sites are determined based on cross-contact high-frequency stimulation artifact signals. Changes in emotional mismatch biomarkers are monitored during stimulation pulse intervals or in windows after artifact removal. When mismatch decreases, a weakened or terminated stimulation control signal is generated; otherwise, a maintained or adjusted stimulation control signal is generated.
2. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, The targeted enhancement features include objective enhancement from the lateral amygdala to the anterior insula and subjective enhancement from the posterior insula to the striatum.
3. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, Based on multiple indications such as treatment-resistant depression, anxiety disorders, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, characteristic parameters of directed enhancement brain regions are determined, and the characteristics of the directed enhancement brain regions are updated based on these characteristic parameters.
4. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, Based on multiple indications such as treatment-resistant depression, anxiety disorders, post-traumatic stress disorder, obsessive-compulsive disorder, and chronic pain, stimulation control signal modulation parameters are determined, and the stimulation control signal is updated based on the stimulation control signal modulation parameters.
5. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, The objective feature corresponds to an earlier time window than the subjective feature corresponds to a later time window.
6. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 5, characterized in that, The objective features include the saliency of the stimulus and the quadratic features of the absolute value of the price, while the subjective features include the cumulative integral features corresponding to the late time window.
7. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, The directional enhancement brain region features are obtained based on a nonlinear phase coupling metric.
8. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, The nonlinear phase coupling metric employs either directed phase transfer entropy or symbol transfer entropy.
9. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 8, characterized in that, The directed phase transfer entropy is calculated over the theta frequency band.
10. The closed-loop neural modulation method based on emotion mismatch biomarkers as described in claim 1, characterized in that, The formula for calculating the deviation amount is as follows: M=max(0,( S) max(0, O)) Where O represents the objective features after decoding, and S represents the subjective features after decoding.