Electroencephalogram signal-based withdrawal addiction system and method

Abstract

The invention belongs to the technical field of biomedical rehabilitation, and particularly discloses an electroencephalogram-based addiction withdrawal system and method. According to the system, electroencephalogram signals of a user in a normal state and an addiction state are collected, the characteristic difference of the two types of signals is extracted, and reverse adjustment pulses are generated and sent to the brain so as to balance the abnormal addiction electroencephalogram signals; meanwhile, the system has a real-time early warning function, and can timely recognize the addiction recurrence risk. The problems of low signal matching degree, poor adjustment pertinence and early warning lag in the existing addiction withdrawal technology are solved, the stability and safety of the withdrawal effect are improved, and an efficient and accurate technical scheme is provided for addiction withdrawal.

Description

Technical Field

[0001] This invention belongs to the field of biomedical rehabilitation technology, and specifically discloses an addiction withdrawal system and method based on electroencephalogram (EEG) signals. Background Technology

[0002] Current addiction withdrawal techniques (such as drug withdrawal and traditional psychological intervention) and some neuromodulation techniques suffer from the following three specific problems in practical applications, which seriously affect withdrawal effectiveness and user experience:

[0003] Significant biases exist in EEG signal feature extraction: Current technologies largely rely on single-band EEG signals (such as theta waves and delta waves) to determine addiction status, failing to establish a multi-dimensional feature comparison model between "normal" and "addictive" signals, resulting in feature extraction biases exceeding 35%. For example, in nicotine addiction scenarios, judging the degree of addiction solely based on alpha wave intensity can easily misinterpret abnormal alpha waves in a fatigued state as addiction signals, leading to incorrect modulation direction.

[0004] Asynchrony between pulse modulation and EEG signals: Most existing pulse transmitting devices use a fixed frequency (e.g., 50Hz) to transmit pulses, without dynamically adjusting according to the phase and rhythm of the user's real-time EEG signals. This results in a synchronization rate of less than 40% between the pulse and the EEG signals. When the user's EEG signals are in a high-frequency active phase, a fixed low-frequency pulse not only fails to balance the abnormal signals but may also trigger discomfort such as dizziness and headaches, with an incidence rate as high as 28%.

[0005] Delayed early warning of addiction relapse: Existing early warning mechanisms are mostly based on user behavior data (such as relapse behavior records) or single physiological indicators (such as heart rate), without combining early abnormal changes in EEG signals. When users experience addictive psychological cravings but do not actually engage in them, the early warning system cannot identify them in time, resulting in a warning lag of more than 12 hours, missing the best opportunity for intervention, and leading to a persistently high relapse rate. Summary of the Invention

[0006] This invention discloses an addiction withdrawal system based on electroencephalogram (EEG) signals, comprising an EEG signal acquisition module, a physiological signal acquisition module, a signal processing module, a pulse generation module, a pulse emission module, an early warning module, a cloud data module, and a power supply module. The EEG signal acquisition module and the physiological signal acquisition module are respectively connected to the signal processing module for transmitting EEG and physiological signals. The signal processing module is respectively connected to the pulse generation module, the early warning module, and the cloud data module for calculating the characteristic difference degree of EEG signals and outputting optimized parameters. The pulse generation module is connected to the pulse emission module for generating reverse modulation pulses. The early warning module is connected to a user terminal and a monitoring terminal for pushing early warning information. The power supply module supplies power to all modules.

[0007] Furthermore, the signal processing module employs a multi-dimensional EEG signal feature vector extraction algorithm, the algorithm formula of which is as follows: Where F d The difference in EEG signal characteristics between "normal" and "addictive" states, where m is the feature dimension and w is the number of features. k For feature weights, F n,k F is the characteristic value of the normal state. a,k These are the characteristic values ​​of the addictive state.

[0008] Furthermore, the pulse generation module employs a dynamic phase synchronization pulse adjustment mechanism, and the pulse phase calculation formula is as follows: in For pulse phase, For real-time EEG phase, F d,max This is the preset maximum feature difference.

[0009] Furthermore, the pulse generation module also includes a reverse feature matching pulse generation model, which is based on F d The frequency and intensity of the pulses are determined, and the abnormal frequency bands of the addictive EEG signals are reversed to make the addictive signals converge to normal signals.

[0010] Furthermore, the early warning module is an EEG-physiological multimodal fusion early warning system. The system calculates the early warning risk value by combining the rate of change of EEG features with physiological signals. When the risk value exceeds the threshold, an early warning is triggered, and the response time is ≤3 minutes.

[0011] Furthermore, an addiction withdrawal method based on electroencephalogram (EEG) signals includes the following steps:

[0012] S1: The EEG signal acquisition module collects the user's EEG signals in normal state. n EEG signals S in an addictive state a ;

[0013] S2: The signal processing module uses a multi-dimensional EEG signal feature vector extraction algorithm to calculate... The difference in "normal-addictive" signal characteristics was obtained;

[0014] S3: The pulse generation module, based on F of S2... d With real-time EEG phase The pulse phase is calculated through a dynamic phase synchronization pulse adjustment mechanism. And combine it with the reverse feature matching model to generate pulse parameters;

[0015] S4: The pulse emission module sends the reverse modulation pulse generated by S3 to the user's brain to modulate the abnormal brain electrical signals of addiction;

[0016] S5: The early warning module collects the rate of change of EEG characteristics and physiological signals in real time, calculates the early warning risk value, and pushes early warning information when the threshold is exceeded;

[0017] S6: The signal processing module uploads the adjustment data to the cloud data module, optimizes the algorithm parameters through federated learning, and iteratively updates the system.

[0018] Furthermore, in step S4, the pulse type of the pulse emission module includes transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), or transcranial alternating current stimulation (tACS), and the brain regions acted upon are the prefrontal, parietal, or temporal lobes associated with addiction.

[0019] Furthermore, in step S5, the calculation basis for the warning risk value includes: the rate of change of EEG characteristics (such as an increase of >10% in β waves within 10 minutes), a decrease of >5% in heart rate variability, or an increase of >15% in skin conductance.

[0020] Furthermore, step S4 also includes an adaptive pulse intensity feedback adjustment step: collecting the user's physiological feedback (tremor frequency, discomfort score) through a somatosensory sensor, lowering the pulse intensity when the feedback exceeds the safe range, and raising the intensity when the feedback is normal and the adjustment is slow, with the intensity adjustment range being 0.05-0.1mA / time.

[0021] Furthermore, in step S6, the cloud data module collects "normal-addictive" EEG signal samples of different addiction types and optimizes the feature weights w through federated learning. k A pulse-based addiction withdrawal system, characterized by improving algorithm adaptability to different users by ≥30%, comprises an EEG signal acquisition module, a physiological signal acquisition module, a signal processing module, a pulse generation module, a pulse emission module, an early warning module, a cloud data module, and a power supply module. The EEG signal acquisition module and the physiological signal acquisition module are respectively connected to the signal processing module for transmitting EEG and physiological signals. The signal processing module is respectively connected to the pulse generation module, the early warning module, and the cloud data module for calculating the characteristic difference of EEG signals and outputting optimized parameters. The pulse generation module is connected to the pulse emission module for generating reverse regulation pulses. The early warning module is connected to the user terminal and the monitoring terminal for pushing early warning information. The power supply module supplies power to all modules.

[0022] Beneficial effects:

[0023] This system collects EEG signals from users in both normal and addictive states, extracts the characteristic differences between the two types of signals, generates inverse modulation pulses, and sends them to the brain to balance the abnormal EEG signals associated with addiction. It also features a real-time warning function to promptly identify the risk of addiction relapse. This invention solves the problems of low signal matching, poor modulation targeting, and delayed warnings in existing addiction withdrawal technologies, thus improving the stability and safety of withdrawal effects. Attached Figure Description

[0024] Figure 1 Overall architecture diagram of the pulse withdrawal addiction system. Example

[0025] A system and method for addiction withdrawal based on electroencephalogram (EEG) signals includes an EEG signal acquisition module, a signal processing module, a pulse generation module, a pulse emission module, an early warning module, a cloud data module, and a power supply module. These modules are connected via a data bus to collaboratively complete the entire process of "signal acquisition - feature extraction - pulse regulation - early warning iteration." The corresponding withdrawal method includes six core steps: signal acquisition, feature comparison, pulse generation, synchronization regulation, early warning monitoring, and algorithm iteration. The following are the six inventive points:

[0026] To address the issue of large feature extraction bias, a multi-dimensional EEG signal feature vector extraction algorithm was designed. This algorithm collects data from the user's normal state (S...). n ) and addiction state (S a The algorithm extracts features from EEG signals under various brain regions based on three dimensions: frequency intensity (θ, δ, α, β, γ waves), signal rhythm stability, and cross-brain region synchronicity. A feature vector is then constructed, and the dissimilarity is calculated. The algorithm formula is as follows:

[0027]

[0028] Among them, F d The difference in EEG signal characteristics between "normal" and "addictive" is represented by m, which is the feature dimension (8 in this invention, including 5 frequency band intensities, 2 rhythm indicators, and 1 cross-regional synchronicity indicator), and w. k F represents the weight of the k-th feature (obtained through machine learning training, ranging from 0.1 to 0.3). n,k F is the k-th eigenvalue under normal conditions. a,k This represents the k-th feature value in the addictive state.

[0029] This algorithm reduces feature extraction bias to below 8%, far lower than the 35% of existing technologies, ensuring the accuracy of feature difference recognition.

[0030] To address the issue of asynchronous pulse and EEG signals, a dynamic phase-synchronized pulse modulation mechanism is proposed. The signal processing module monitors the phase of the user's current EEG signal in real time. (Updated every 10ms), the pulse generation module determines the pulse based on the feature difference F. d This generates a pulse signal that is phase-reversed with the EEG signal, ensuring that the pulse transmission time is precisely aligned with the abnormal phase point of the EEG signal. The pulse phase adjustment formula is as follows:

[0031]

[0032] in, F represents the real-time phase of the pulse signal, π represents the reverse phase offset, and F... d,maxThe maximum feature variance was preset (determined to be 1.2 based on a large number of samples). It is a dynamic compensation phase, used to fine-tune the pulse phase according to the severity of addiction.

[0033] This mechanism increases the synchronization rate between pulses and EEG signals to over 92% and reduces the incidence of adverse reactions to below 3%.

[0034] Based on the feature differences between "normal" and "addictive" signals, a reverse feature matching pulse generation model is constructed. This model uses the feature vector F of normal EEG signals. n To calculate the addiction signal F a With F n The system generates "deviation-compensating" pulse parameters (frequency, intensity, and duration) based on the direction of the deviation: if the β-wave intensity in the addiction signal is too high (exceeding the normal range by 20%), a low-frequency (8-12Hz) pulse is generated to suppress the β-wave; if the θ-wave intensity is abnormal (below the normal range by 15%), a pulse of specific intensity (0.5-1.2mA) is generated to enhance the θ-wave. The pulse parameters are linearly correlated with the characteristic deviation, ensuring targeted adjustment and increasing the convergence speed of the addiction signal to a normal signal by 50%.

[0035] To address the issue of delayed early warning, an EEG-physiological multimodal fusion early warning system was designed. This system simultaneously collects EEG signals (rate of change of addiction-related features) and physiological signals (heart rate variability, skin conductance) to establish an early warning risk value calculation model. When the rate of change of EEG features exceeds a threshold (e.g., a 15% increase in beta wave intensity within 10 minutes) and the physiological signals are abnormal (an 8% decrease in heart rate variability), the system triggers an early warning and pushes a notification to the user and supervisor via a mobile app. The early warning response time is controlled within 3 minutes, significantly shorter than the 12 hours of existing technologies.

[0036] An adaptive feedback adjustment function is integrated into the pulse emission module. Through a wearable somatosensory sensor, the system collects real-time physiological feedback from the user regarding the pulses (such as muscle tremor frequency and head discomfort score). When the feedback value exceeds the safe range (e.g., tremor frequency > 5 times / second), the pulse intensity is automatically reduced (decreased by 0.1mA each time); when the feedback is normal and the EEG signal converges slowly in the normal direction, the pulse intensity is appropriately increased (increased by 0.05mA each time). This module improves the safety of pulse adjustment to 99%, avoiding discomfort caused by inappropriate intensity.

[0037] A cloud-based EEG signal database has been established, containing "normal-addictive" EEG signal samples from different addiction types (nicotine, alcohol, gaming, etc.) and different populations (age, gender). The system regularly uploads users' adjustment data (feature changes, impulse parameters, withdrawal effects) to the cloud. Through federated learning algorithms, the feature extraction weights (w_k) and impulse parameter matching models are optimized, improving the algorithm's adaptability to different users by 40%. This enables bidirectional algorithm iteration between individuals and groups, continuously optimizing withdrawal effects.

[0038] Example 2

[0039] System deployment: The user wears a head-mounted EEG acquisition device (containing 6 electrodes covering the frontal and temporal lobes) and a wrist physiological sensor (measuring heart rate variability and skin conductance). The pulse emission module acts on the left dorsolateral prefrontal lobe (addiction-related brain region) via transcranial magnetic stimulation (TMS) and connects to the cloud data module.

[0040] Signal acquisition and feature extraction: EEG signals were collected from users 72 hours before quitting smoking (addictive state) and 1 week after quitting smoking (normal state). The multi-dimensional feature vector extraction algorithm was used to calculate (F_d = 0.82) and it was found that the intensity of β waves (22-30Hz) was 28% higher and the intensity of θ waves (4-8Hz) was 16% lower during nicotine addiction.

[0041] Pulse modulation: The dynamic phase-synchronized pulse modulation mechanism generates pulses with a frequency of 10Hz and an initial intensity of 0.8mA based on real-time EEG phase. The model is reverse-phase matched with the abnormal phase of the beta wave; the reverse feature matching model is used to suppress the beta wave and enhance the theta wave, and is adjusted twice a day for 30 minutes each time.

[0042] Early warning and feedback: On the 10th day, the system detected a 12% increase in wave intensity and a 7% decrease in heart rate variability within 10 minutes. The multimodal fusion early warning system triggered a pre-user beta alert and pushed a "psychological craving risk" reminder, allowing regulators to intervene psychologically in a timely manner. The adaptive module reduced the pulse intensity to 0.7mA based on user feedback of "mild dizziness".

[0043] Results Comparison: Compared with traditional nicotine replacement therapy (such as nicotine patches), this system increases the success rate of quitting within 4 weeks by 45% and reduces the incidence of adverse reactions such as dizziness and nausea by 60%.

[0044] Alcohol addiction withdrawal scenarios

[0045] System configuration: Transcranial direct current stimulation (tDCS) pulse emission mode is used to act on the medial prefrontal cortex. The EEG acquisition module is equipped with orbitofrontal electrodes to focus on monitoring changes in delta waves (0.5-4Hz) (the core frequency band associated with alcohol addiction).

[0046] Features and Modulation: EEG signals were collected from users in normal and alcohol-dependent states, and F was calculated. d =0.95, it was found that the intensity of delta waves was 35% higher and the cross-brain region synchronization was 22% lower when alcohol addiction was present; the reverse feature matching model generated a 15Hz pulse to suppress delta waves, and the dynamic synchronization mechanism made the phase synchronization rate between the pulse and delta waves reach 94%.

[0047] Early warning and iteration: In the third week, when a user had the thought of "wanting to drink alcohol", the system triggered an early warning 4 minutes in advance through changes in EEG characteristics (a 10% increase in delta wave intensity); the cloud system optimized the feature weight (w_k) based on the user's data, increasing the delta wave weight from 0.25 to 0.28, and the subsequent adjustment accuracy improved by 12%.

[0048] Effect Comparison: Compared with drug withdrawal (such as naltrexone), this system has no drug side effects, reduces the relapse rate by 50% in 3 months, and shortens the time for EEG signals to return to normal by 30%.

[0049] Game addiction withdrawal scenario

[0050] System Application: For adolescent game addicts, non-invasive EEG acquisition (dry electrodes) is used, the pulse emission module is transcranial alternating current stimulation (tACS) and acts on the parietal lobe (attention-related brain region), and physiological sensors are added to increase eye tracking (monitoring game-related visual stimulus response).

[0051] Characteristics and Modulation: In a state of gaming addiction, the intensity of users' gamma waves (30-45Hz) is 23% higher than in the normal state, and their rhythmic stability is poor; multi-dimensional feature algorithm calculation (F d =0.78, the inverse model generates a 25Hz pulse modulated γ wave with a dynamic synchronization rate of 91%.

[0052] Warning and feedback: When the user comes into contact with game-related screens, eye tracking shows an increase in fixation time and an 8% increase in EEG gamma wave intensity. The warning system immediately reminds the user to "avoid exposure to addictive stimuli". The adaptive module controls the pulse intensity between 0.3-0.6mA based on the tolerance of adolescents.

[0053] Comparison of effects: Compared with traditional psychological interventions, this system reduces the daily gaming time of teenagers by 70%, improves attention concentration (assessed by EEG alpha wave stability) by 55%, and achieves a 90% satisfaction rate among parents regarding the withdrawal effect.

Claims

1. A brainwave-based addiction withdrawal system, characterized in that, It includes an EEG signal acquisition module, a physiological signal acquisition module, a signal processing module, a pulse generation module, a pulse emission module, an early warning module, a cloud data module, and a power supply module; The EEG signal acquisition module and the physiological signal acquisition module are respectively connected to the signal processing module for transmitting EEG signals and physiological signals; the signal processing module is respectively connected to the pulse generation module, the early warning module and the cloud data module for calculating the characteristic difference degree of EEG signals and outputting optimization parameters; the pulse generation module is connected to the pulse transmission module for generating reverse modulation pulses; the early warning module is connected to the user terminal and the monitoring terminal for pushing early warning information. The power module supplies power to each module.

2. The system according to claim 1, characterized in that, The signal processing module employs a multi-dimensional EEG signal feature vector extraction algorithm, the formula of which is: Where F d The difference in EEG signal characteristics between "normal" and "addictive" is represented by m, where m is the feature dimension and w is the number of features. k For feature weights, F n,k F is the characteristic value of the normal state. a,k These are the characteristic values ​​of the addictive state.

3. The system according to claim 1, characterized in that, The pulse generation module employs a dynamic phase synchronization pulse adjustment mechanism, and the pulse phase calculation formula is as follows: in( For pulse phase, For real-time EEG phase, F d,max This is the preset maximum feature difference.

4. The system according to claim 1, characterized in that, The pulse generation module also includes an inverse feature matching pulse generation model, which is based on F d The frequency and intensity of the pulses are determined, and the abnormal frequency bands of the addictive EEG signals are reversed to make the addictive signals converge to normal signals.

5. The system according to claim 1, characterized in that, The warning module is an EEG-physiological multimodal fusion warning system. The system calculates the warning risk value by combining the rate of change of EEG features with physiological signals. When the risk value exceeds the threshold, the warning is triggered and the response time is ≤3 minutes.

6. A method for addiction withdrawal based on electroencephalogram (EEG) signals, characterized in that, Includes the following steps: S1: The EEG signal acquisition module collects the user's EEG signals in normal state. n EEG signals S in an addictive state a ; S2: The signal processing module uses a multi-dimensional EEG signal feature vector extraction algorithm to calculate... Obtain the difference in "normal-addictive" signal characteristics; S3: The pulse generation module, based on F of S2... d With real-time EEG phase Calculate pulse phase using dynamic phase synchronization pulse adjustment mechanism And combine it with the reverse feature matching model to generate pulse parameters; S4: The pulse emission module sends the reverse modulation pulse generated by S3 to the user's brain to modulate the abnormal brain electrical signals of addiction; S5: The early warning module collects the rate of change of EEG characteristics and physiological signals in real time, calculates the early warning risk value, and pushes early warning information when the threshold is exceeded; S6: The signal processing module uploads the adjustment data to the cloud data module, optimizes the algorithm parameters through federated learning, and iteratively updates the system.

7. The method according to claim 6, characterized in that, In step S4, the pulse type of the pulse delivery module includes transcranial magnetic stimulation (TMS), transcranial direct current stimulation (tDCS), or transcranial alternating current stimulation (tACS), and the target brain region is the prefrontal lobe, parietal lobe, or temporal lobe associated with addiction.

8. The method according to claim 6, characterized in that, In step S5, the calculation basis for the warning risk value includes: the rate of change of EEG characteristics (such as an increase of >10% in β waves within 10 minutes), a decrease of >5% in heart rate variability, or an increase of >15% in skin conductance.

9. The method according to claim 6, characterized in that, Step S4 also includes an adaptive pulse intensity feedback adjustment step: collecting the user's physiological feedback (tremor frequency, discomfort score) through a somatosensory sensor, lowering the pulse intensity when the feedback exceeds the safe range, and raising the intensity when the feedback is normal and the adjustment is slow, with the intensity adjustment range being 0.05-0.1mA / time.

10. The method according to claim 6, characterized in that, In step S6, the cloud data module collects "normal-addictive" EEG signal samples of different addiction types and optimizes the feature weights w through federated learning. k This improves the algorithm's adaptability to different users by ≥30%. A pulse-based addiction withdrawal system, characterized in that, It includes an EEG signal acquisition module, a physiological signal acquisition module, a signal processing module, a pulse generation module, a pulse emission module, an early warning module, a cloud data module, and a power supply module; The EEG signal acquisition module and the physiological signal acquisition module are respectively connected to the signal processing module for transmitting EEG signals and physiological signals; the signal processing module is respectively connected to the pulse generation module, the early warning module and the cloud data module for calculating the characteristic difference degree of EEG signals and outputting optimization parameters; the pulse generation module is connected to the pulse transmission module for generating reverse modulation pulses; the early warning module is connected to the user terminal and the monitoring terminal for pushing early warning information. The power module supplies power to each module.

Images