A brain computer interface detection cap and monitoring system for sleep disorder monitoring

Abstract

This invention provides a brain-computer interface (BCI) testing cap and monitoring system for sleep disorder monitoring, belonging to the field of BCI medical technology. The BCI testing cap for sleep disorder monitoring includes a cap body, a flexible electrode array, an integrated signal acquisition unit, an electrode status self-test module, a wireless transmission and local storage module, and a power management module. The cap body is flexible, and a monitoring controller is mounted on its side wall. This invention provides a BCI testing cap and monitoring system for sleep disorder monitoring, achieving low-interference home acquisition of medical-grade multimodal sleep physiological signals, and constructing a closed-loop process from signal acquisition, intelligent analysis, accurate classification to intervention decision-making, providing a complete technical solution for the precise diagnosis and treatment of insomnia disorders.

Description

Technical Field

[0001] This invention relates to the field of brain-computer interface medical technology, specifically to a brain-computer interface detection cap and monitoring system for sleep disorder monitoring. Background Technology

[0002] Currently, nearly 380 million adults in my country suffer from insomnia. Traditional insomnia diagnosis relies on subjective scales and lacks objective neurophysiological markers. Sleep disorders have become a major public health problem. The clinical gold standard, polysomnography (PSG), requires overnight monitoring in hospitals, has more than 10 leads, poor wearing comfort, suffers from the first-night effect, and has high cost per test, making it impossible to achieve large-scale popularization and long-term home-based monitoring. Existing portable EEG monitoring devices often have drawbacks such as strong pressure from rigid electrodes, easy electrode displacement due to body movement during sleep, and weak signal artifact suppression capabilities. Furthermore, they lack closed-loop linkage with AI classification and personalized intervention, making it impossible to achieve accurate and objective monitoring and classification diagnosis of insomnia.

[0003] Therefore, this invention proposes a brain-computer interface detection cap and monitoring system for sleep disorder monitoring. This brain-computer interface sleep monitoring system is flexible, comfortable, suitable for home use, has stable signals, and can be connected to AI for precise classification and neuromodulation, which has become a technical problem that urgently needs to be solved in the field of sleep medicine. Summary of the Invention

[0004] This invention addresses the technical problems existing in the prior art by providing a brain-computer interface (BCI) detection cap and monitoring system for sleep disorder monitoring. The aim is to overcome the aforementioned deficiencies of the prior art and provide a BCI detection cap and monitoring system for sleep disorder monitoring that achieves low-interference home acquisition of medical-grade multimodal sleep physiological signals and constructs a closed-loop process from signal acquisition, intelligent analysis, accurate classification to intervention decision-making, providing a complete technical solution for the precise diagnosis and treatment of insomnia disorders.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A brain-computer interface testing cap for sleep disorder monitoring includes a testing cap body, which is made of a flexible cap body. A monitoring controller body is installed on the side wall of the testing cap body. The monitoring controller body is equipped with an integrated signal acquisition unit, an electrode status self-test module, a wireless transmission and local storage module, and a power management module. The front end of the testing cap body is provided with a brim, and the rear end of the testing cap body is provided with an adjustable tightness adjustment buckle. A flexible electrode array is installed inside the testing cap body. The flexible cap body is made of highly elastic, breathable, and skin-friendly fabric. The inner side of the cap body is provided with a sliding silicone electrode fixing seat that corresponds one-to-one with the electrode. The electrode position can be adjusted along the warp and weft of the cap body to adapt to the scalp curvature of different head shapes. The electrode module adopts a magnetic quick-release structure to ensure connection stability and easy disassembly and assembly. The cap body can be removed, washed, and disinfected separately. The flexible electrode array adopts a multimodal electrode layout, including 12-channel EEG electrodes, 2-channel EOG electrodes, and 2-channel EMG electrodes. The flexible electrode array uses nano-silver carbon nanotube composite flexible dry electrodes, which can form a stable electrode electrolyte interface with the scalp without the need for conductive cream. The electrode status self-test module realizes real-time detection of electrode-scalp contact impedance based on AC constant current source injection method. The module monitors the contact impedance value of each electrode in real time, and triggers a contact abnormality reminder when the contact impedance is >10kΩ. The integrated signal acquisition unit is used to synchronously acquire EEG, EOG, and EMG multimodal physiological signals. The unit has a built-in low-noise instrumentation amplifier, a programmable bandpass filter circuit, a 50Hz power frequency notch filter circuit, and a 24-bit high-precision A / D converter. The monitoring controller body is the core control unit, which integrates a brain-computer interface system for monitoring sleep disorders. The wireless transmission and local storage module adopts the Bluetooth transmission protocol and supports symmetric encryption algorithm to encrypt the collected data throughout the entire link. The module has built-in non-volatile flash memory, which can completely store more than 10 days of complete monitoring data in the absence of network. After the network is restored, it will automatically verify and retransmit the missing data. The power management module uses a lithium polymer battery and has a built-in four-fold protection circuit for overcharge, over-discharge, overcurrent, and short circuit. The battery capacity is ≥500mAh to meet the continuous monitoring requirements of ≥8 hours. A brain-computer interface monitoring system for monitoring sleep disorders, the monitoring system includes the brain-computer interface detection cap of claim 1, the monitoring system includes an edge computing layer, a cloud intelligence layer, an application interaction layer, and a closed-loop intervention layer, and each layer realizes bidirectional data flow through a standardized encrypted interface; The edge computing layer is a local terminal processing module used to receive encrypted data uploaded by the detection cap, and to complete decryption, adaptive artifact removal, signal standardization preprocessing, and local data backup. The cloud-based intelligent layer is the MianShu AI Cloud Platform, which includes a data management submodule, an AI sleep automatic segmentation submodule, an insomnia accurate classification submodule, and a diagnosis and treatment decision support submodule. The AI ​​sleep automatic segmentation submodule is based on a Transformer-CNN fusion model to realize automatic sleep cycle segmentation, and the insomnia accurate classification submodule is based on a graph neural network (GNN) to build a multimodal fusion classification model. The application interaction layer includes a doctor-side management module and a patient-side display module, which are used for monitoring report visualization, treatment plan interaction, and full-cycle follow-up management. The closed-loop intervention layer is a neuromodulation adaptation module, used to generate personalized transcranial magnetic stimulation (rTMS) treatment parameters based on insomnia classification results, and to complete the objective quantitative evaluation of treatment efficacy and iterative optimization of the treatment plan.

[0006] The beneficial effects of this invention are: This invention provides a brain-computer interface testing cap for sleep disorder monitoring. The testing cap adopts flexible dry electrodes and a lightweight design, which can complete multimodal signal acquisition that meets clinical standards in a natural home sleep environment. It completely solves the first-night effect and portability defects of traditional PSG, while ensuring the clinical diagnostic validity of the data.

[0007] This invention provides a brain-computer interface monitoring system for sleep disorder monitoring. This monitoring system breaks through the limitation of existing devices that only monitor but do not diagnose and treat. It constructs a complete closed loop from signal acquisition, sleep staging, accurate classification to intervention decision-making and efficacy evaluation, promoting the transformation of insomnia diagnosis and treatment from experience-driven to evidence-driven, and solving the core pain point of homogenization in clinical treatment.

[0008] This invention provides a brain-computer interface detection cap and monitoring system for sleep disorder monitoring, which features high reliability and scenario adaptability. It adopts a dual mode of wireless real-time transmission and local backup storage to prevent data loss due to network outages. The device and system are suitable for multiple scenarios such as home screening, in-hospital diagnosis, primary healthcare, and multi-center research, which can effectively promote the downward flow of sleep medical resources. At the same time, the AI ​​model is trained and optimized based on the multimodal clinical database of insomnia in the Chinese population, which has stronger localization adaptability compared with overseas models.

[0009] Based on the above technical solution, the present invention can be further improved as follows.

[0010] Furthermore, the adaptive artifact removal module of the edge computing layer achieves artifact separation and removal based on the FastICA independent component analysis algorithm. Its core decomposition formula is: ; in For multi-channel observation signal matrix, It is a mixed matrix. Given the source signal matrix, the module uses the FastICA algorithm to separate artifact components such as eye movement, electromyography, and power frequency interference. After removing artifacts, a pure EEG signal is reconstructed. The algorithm's convergence threshold is 1e-6, and the maximum number of iterations is 1000.

[0011] Furthermore, the core attention mechanism of the AI ​​sleep auto-segmentation submodule in the cloud-based intelligent layer satisfies the formula: ; in For query matrix, For the key matrix, For value matrices, The feature vector dimension is used; the model first extracts the temporal-frequency domain local features of the preprocessed EEG signal through CNN, then captures the long-term sleep cycle dependent features through the Transformer attention mechanism, and finally divides the sleep signal into 5 stages according to the AASM sleep stage standard: wakefulness, N1 stage, N2 stage, N3 deep sleep stage, and REM rapid eye movement stage. The core node feature update formula for the insomnia precision sub-module of the cloud-based intelligent layer to construct the multimodal fusion sub-model is as follows: ; in For the first Layered brain network nodes eigenvectors, For nodes The set of adjacent nodes, For neighborhood feature aggregation function, , For the first Learnable weights and biases of layers Using a nonlinear activation function, the model constructs a brain functional connectivity network with EEG channels as nodes, integrates sleep stage characteristics, EEG power spectrum characteristics, and clinical scale characteristics, and mines neurophysiological biomarkers of insomnia to achieve objective and accurate classification of insomnia subtypes.

[0012] Furthermore, the insomnia precise subtyping submodule of the cloud-based intelligent layer incorporates a multimodal feature fusion unit, which achieves adaptive fusion of multi-source features based on an attention-weighted mechanism. The fusion formula is as follows: ; For the first Each single-modal feature vector includes EEG frequency domain features, sleep structure features, and clinical scale features. For the first Adaptive attention weights for each modality, and satisfying The weight calculation formula is: ; in This is the mapping function for a multilayer perceptron.

[0013] Furthermore, the AI ​​sleep auto-switching submodule of the cloud-based intelligent layer has a built-in EEG feature extraction unit that extracts the power spectral density (PSD) features of the EEG signal based on the Welch method. The calculation formula is as follows: ; in Number of signal segments The energy of the window function, For the Hanning window function, For the first segment time domain signal, For Fast Fourier Transform, the absolute and relative energy characteristics of five sleep-related frequency bands—δ (0.5~4Hz), θ (4~8Hz), α (8~13Hz), β (13~30Hz), and standard sleep σ = 12~14Hz—are extracted based on the PSD results.

[0014] Furthermore, the closed-loop intervention layer incorporates a personalized parameter generation unit. Based on the patient's insomnia subtype and EEG characteristics, the personalized parameter generation unit outputs core parameters such as stimulation target, stimulation frequency, stimulation intensity, stimulation pulse number, and treatment course. After treatment, it automatically compares the changes in sleep structure parameters and EEG biomarkers before and after treatment to complete an objective quantitative assessment of the therapeutic effect, and iteratively optimizes the parameter model based on the assessment results.

[0015] Furthermore, the data management submodule follows medical data security standards, performs full lifecycle access control and anonymization processing on patient data, and the anonymization processing meets irreversible anonymization rules, which can be directly connected to the hospital's electronic medical record system.

[0016] Furthermore, the patient-side module of the application interaction layer supports functions such as viewing sleep reports, sleep quality scoring, health education, and follow-up reminders, while the doctor-side module supports functions such as batch patient management, report review, issuance of treatment plans, and multi-center scientific research data statistics. It can generate sleep monitoring reports and subtype diagnosis reports that conform to clinical standards. Attached Figure Description

[0017] Figure 1 This is a structural block diagram of the brain-computer interface monitoring system for sleep disorder monitoring according to the present invention; Figure 2 This is a front view of the overall shape of the brain-computer interface detection cap for sleep disorder monitoring according to the present invention. Figure 3 This is a rear view of the overall shape of the brain-computer interface detection cap for sleep disorder monitoring according to the present invention. Figure 4 This is an internal view of the overall shape of the brain-computer interface detection cap for sleep disorder monitoring according to the present invention. Figure 5 This is a flowchart illustrating the working steps of the brain-computer interface monitoring system for sleep disorder monitoring according to the present invention.

[0018] The attached diagram lists the components represented by each number as follows: 1. Detection cap body; 2. Monitoring controller body; 3. Cap brim; 4. Tightness adjustment buckle; 5. Flexible electrode array. Detailed Implementation

[0019] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0020] The present invention provides the following preferred embodiments. Example

[0021] like Figure 1-4 As shown, this embodiment of the invention provides a brain-computer interface testing cap for sleep disorder monitoring. The testing cap includes a main body 1, a monitoring controller body 2 installed on the side wall of the main body 1, and an integrated signal acquisition unit, an electrode status self-test module, a wireless transmission and local storage module, and a power management module installed inside the monitoring controller body 2. The front end of the main body 1 is provided with a brim 3, and the rear end of the main body 1 is provided with an adjustable tightness adjustment buckle 4. A flexible electrode array 5 is installed inside the main body 1. The functional structure of the testing cap includes a flexible cap body, a flexible electrode array, an integrated signal acquisition unit, an electrode status self-test module, a wireless transmission and local storage module, and a power management module. The flexible cap body is made of highly elastic, breathable, and skin-friendly fabric. The inner side of the cap is provided with a sliding silicone electrode fixing seat corresponding to each electrode. The electrode position can be adjusted along the warp and weft of the cap body to adapt to the scalp curvature of different head shapes. The electrode module adopts a magnetic quick-release structure to ensure connection stability and ease of disassembly and assembly. The cap body can be disassembled, washed, and disinfected separately. The flexible electrode array adopts a multimodal electrode layout, including 12-channel EEG electrodes, 2-channel EOG electrodes, and 2-channel EMG electrodes. The flexible electrode array uses nano-silver carbon nanotube composite flexible dry electrodes, which can form a stable electrode electrolyte interface with the scalp without the need for conductive cream. The electrode status self-test module realizes real-time detection of electrode-scalp contact impedance based on AC constant current source injection method. The module monitors the contact impedance value of each electrode in real time, and triggers a contact abnormality reminder when the contact impedance is >10kΩ. The integrated signal acquisition unit is used to synchronously acquire EEG, EOG, and EMG multimodal physiological signals. The unit has a built-in low-noise instrumentation amplifier, a programmable bandpass filter circuit, a 50Hz power frequency notch filter circuit, and a 24-bit high-precision A / D converter. The monitoring controller body is the core control unit, which integrates a brain-computer interface system for monitoring sleep disorders. The wireless transmission and local storage module adopts the Bluetooth 5.2 low-power transmission protocol and supports the AES-256 symmetric encryption algorithm to encrypt the collected data throughout the entire link. The module has built-in non-volatile flash memory, which can store more than 10 days of complete monitoring data when the network is disconnected. After the network is restored, it will automatically verify and retransmit the missing data. The power management module uses a 3.7V lithium polymer battery and has a built-in four-fold protection circuit for overcharge, over-discharge, overcurrent, and short circuit. The battery capacity is ≥500mAh to meet the continuous monitoring requirements of ≥8 hours. Example

[0022] like Figure 1-4 As shown, a brain-computer interface monitoring system for sleep disorder monitoring is disclosed. The monitoring system includes the brain-computer interface detection cap of claim 1. The monitoring system includes an edge computing layer, a cloud intelligence layer, an application interaction layer, and a closed-loop intervention layer. Each layer realizes bidirectional data flow through a standardized encrypted interface. The edge computing layer is a local terminal processing module used to receive encrypted data uploaded by the detection cap, and to complete decryption, adaptive artifact removal, signal standardization preprocessing, and local data backup. The cloud-based intelligent layer is the MianShu AI Cloud Platform, which includes a data management submodule, an AI sleep automatic segmentation submodule, an insomnia accurate classification submodule, and a diagnosis and treatment decision support submodule. The AI ​​sleep automatic segmentation submodule is based on a Transformer-CNN fusion model to realize automatic sleep cycle segmentation, and the insomnia accurate classification submodule is based on a graph neural network (GNN) to build a multimodal fusion classification model. The application interaction layer includes a doctor-side management module and a patient-side display module, which are used for monitoring report visualization, treatment plan interaction, and full-cycle follow-up management. The closed-loop intervention layer is a neuromodulation adaptation module, used to generate personalized transcranial magnetic stimulation (rTMS) treatment parameters based on insomnia classification results, and to complete the objective quantitative evaluation of treatment efficacy and iterative optimization of the treatment plan. The adaptive artifact removal module of the edge computing layer achieves artifact separation and removal based on the FastICA independent component analysis algorithm. Its core decomposition formula is: ; in For multi-channel observation signal matrix, It is a mixed matrix. Given the source signal matrix, the module uses the FastICA algorithm to separate artifact components such as eye movement, electromyography, and power frequency interference. After removing artifacts, it reconstructs a pure EEG signal. The algorithm's convergence threshold is 1e-6, and the maximum number of iterations is 1000. The core attention mechanism of the AI ​​sleep auto-segmentation submodule in the cloud-based intelligent layer satisfies the formula: ; in For query matrix, For the key matrix, For value matrices, The feature vector dimension is used; the model first extracts the temporal-frequency domain local features of the preprocessed EEG signal through CNN, then captures the long-term sleep cycle dependent features through the Transformer attention mechanism, and finally divides the sleep signal into 5 stages according to the AASM sleep stage standard: wakefulness, N1 stage, N2 stage, N3 deep sleep stage, and REM rapid eye movement stage. The core node feature update formula for the insomnia precision sub-module of the cloud-based intelligent layer to construct the multimodal fusion sub-model is as follows: ; in For the first Layered brain network nodes eigenvectors, For nodes The set of adjacent nodes, For neighborhood feature aggregation function, , For the first Learnable weights and biases of layers As a nonlinear activation function, the model constructs a brain functional connectivity network with EEG channels as nodes, integrates sleep stage characteristics, EEG power spectrum characteristics, and clinical scale characteristics, and mines neurophysiological biomarkers of insomnia to achieve objective and accurate classification of insomnia subtypes; The insomnia precise subtyping submodule of the cloud-based intelligent layer has a built-in multimodal feature fusion unit, which achieves adaptive fusion of multi-source features based on an attention weighting mechanism. The fusion formula is as follows: ; For the first Each single-modal feature vector includes EEG frequency domain features, sleep structure features, and clinical scale features. For the first Adaptive attention weights for each modality, and satisfying The weight calculation formula is: ; in For multilayer perceptron mapping functions; The AI ​​sleep auto-switching submodule of the cloud-based intelligent layer has a built-in EEG feature extraction unit that extracts the power spectral density (PSD) feature of the EEG signal based on the Welch method. The calculation formula is as follows: ; in Number of signal segments The energy of the window function, For the Hanning window function, For the i-th time-domain signal, For Fast Fourier Transform, the absolute and relative energy characteristics of five sleep-related frequency bands, namely δ (0.5~4Hz), θ (4~8Hz), α (8~13Hz), β (13~30Hz), and standard sleep σ = 12~14Hz, are extracted based on the PSD results. The closed-loop intervention layer has a built-in personalized parameter generation unit. Based on the patient's insomnia subtype and EEG characteristics, the personalized parameter generation unit outputs core parameters such as stimulation target, stimulation frequency, stimulation intensity, stimulation pulse number, and treatment course. After the treatment is completed, the changes in sleep structure parameters and EEG characteristic markers before and after the treatment are automatically compared to complete the objective quantitative evaluation of the efficacy. The parameter model is iteratively optimized based on the evaluation results. The data management submodule follows medical data security standards, performs full lifecycle access control and desensitization processing on patient data, and the desensitization processing meets the irreversible anonymization rules, which can be directly connected to the hospital's electronic medical record system; The patient-side module of the application interaction layer supports functions such as viewing sleep reports, sleep quality scoring, health education, and follow-up reminders. The doctor-side module supports functions such as batch patient management, report review, issuance of treatment plans, and multi-center scientific research data statistics. It can generate sleep monitoring reports and subtype diagnosis reports that conform to clinical standards. Example

[0023] like Figure 5 As shown, the monitoring system of the brain-computer interface detection cap for sleep disorder monitoring provided by the present invention has the following specific operating steps: Step 1: After completing the testing cap application, the system performs a full-process self-test. Once the self-test is passed, the system enters standby mode. Step 2: The detection cap simultaneously collects raw physiological signals of EEG, EOG, and EMG from the subject during natural sleep, preprocesses them in hardware, and then encrypts and transmits them to the edge computing layer. Step 3: The edge computing layer performs adaptive artifact removal and standardization preprocessing on the original signal, simultaneously completes local data backup, and uploads the valid data to the cloud AI platform. If the network is down, the preprocessed data is completely stored on the local terminal. After the network is restored, it is automatically re-uploaded to the cloud platform without affecting the normal execution of the monitoring process. Step 4: The cloud-based AI platform automatically stages the sleep cycle, performs multimodal fusion analysis to accurately classify insomnia, and generates personalized treatment suggestions; Step 5: The system generates standardized monitoring and classification reports, which are pushed to both the patient and doctor's ends, supporting doctors to review and adjust treatment plans online; Step 6: Based on the confirmed treatment plan, the closed-loop intervention layer generates personalized neuromodulation parameters and synchronizes them to the treatment device. After the targeted therapy is completed, the efficacy is objectively evaluated and the treatment plan is iterated through follow-up monitoring. Step 7: Complete the entire process of data anonymization and archiving, establish a personal sleep health record, and execute the preset long-term follow-up plan.

[0024] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0025] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0026] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A brain-computer interface testing cap for sleep disorder monitoring, comprising a testing cap body (1), characterized in that: The main body (1) of the test cap is made of a flexible cap body, and a monitoring controller body (2) is installed on its side wall. The monitoring controller body (2) is equipped with an integrated signal acquisition unit, an electrode status self-test module, a wireless transmission and local storage module, and a power management module. The front end of the test cap body (1) is provided with a brim (3), and the rear end of the test cap body (1) is provided with an adjustable tightness adjustment buckle (4). The interior of the test cap body (1) is equipped with a flexible electrode array (5). The flexible cap body is made of a high-elasticity, breathable and skin-friendly fabric. The inner side of the cap body is provided with a sliding silicone electrode fixing seat, which can adjust the electrode position along the warp and weft of the cap body to adapt to the scalp curvature of different head shapes. The electrode module adopts a magnetic quick-release structure to take into account both connection stability and ease of disassembly and assembly. The cap body can be disassembled and disinfected separately. The flexible electrode array adopts a multimodal electrode layout, including 12-channel EEG electrodes, 2-channel EOG electrodes, and 2-channel EMG electrodes. The flexible electrode array uses nano-silver carbon nanotube composite flexible dry electrodes, which can form a stable electrode electrolyte interface with the scalp without the need for conductive cream. The electrode status self-test module realizes real-time detection of the contact impedance between the electrode and the scalp based on the AC constant current source injection method. The module monitors the contact impedance value of each electrode in real time, and triggers a contact abnormality reminder when the contact impedance is >10kΩ. The integrated signal acquisition unit is used to synchronously acquire EEG, EOG, and EMG multimodal physiological signals. The unit has a built-in low-noise instrumentation amplifier, a programmable bandpass filter circuit, a 50Hz power frequency notch filter circuit, and a 24-bit high-precision A / D converter. The monitoring controller body is the core control unit, which integrates a brain-computer interface system for monitoring sleep disorders. The wireless transmission and local storage module adopts the Bluetooth transmission protocol and supports symmetric encryption algorithm to encrypt the collected data throughout the entire link. The module has built-in non-volatile flash memory, which can completely store more than 10 days of complete monitoring data in the absence of network. After the network is restored, it will automatically verify and retransmit the missing data. The power management module uses a lithium polymer battery and has a built-in four-fold protection circuit for overcharge, over-discharge, overcurrent, and short circuit. The battery capacity is ≥500mAh to meet the continuous monitoring requirements of ≥8 hours.

2. A brain-computer interface monitoring system for sleep disorder monitoring, characterized in that: The monitoring system includes the brain-computer interface detection cap of claim 1, and the carrier is the monitoring controller body (2). The monitoring system includes an edge computing layer, a cloud intelligence layer, an application interaction layer, and a closed-loop intervention layer. Each layer realizes bidirectional data flow through a standardized encrypted interface. The edge computing layer is a local terminal processing module used to receive encrypted data uploaded by the detection cap, and to complete decryption, adaptive artifact removal, signal standardization preprocessing, and local data backup. The cloud-based intelligent layer is the MianShu AI Cloud Platform, which includes a data management submodule, an AI sleep automatic segmentation submodule, an insomnia accurate classification submodule, and a diagnosis and treatment decision support submodule. The AI ​​sleep automatic segmentation submodule is based on a Transformer-CNN fusion model to realize automatic sleep cycle segmentation, and the insomnia accurate classification submodule is based on a graph neural network (GNN) to build a multimodal fusion classification model. The application interaction layer includes a doctor-side management module and a patient-side display module, which are used for monitoring report visualization, treatment plan interaction, and full-cycle follow-up management. The closed-loop intervention layer is a neuromodulation adaptation module, used to generate personalized transcranial magnetic stimulation (rTMS) treatment parameters based on insomnia classification results, and to complete the objective quantitative evaluation of treatment efficacy and iterative optimization of the treatment plan.

3. The brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The adaptive artifact removal module of the edge computing layer achieves artifact separation and removal based on the FastICA independent component analysis algorithm. Its core decomposition formula is: ; in For multi-channel observation signal matrix, It is a mixed matrix. Given the source signal matrix, the module uses the FastICA algorithm to separate artifact components such as eye movement, electromyography, and power frequency interference. After removing artifacts, a pure EEG signal is reconstructed. The algorithm's convergence threshold is 1e-6, and the maximum number of iterations is 1000.

4. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The core attention mechanism of the AI ​​sleep auto-segmentation submodule in the cloud-based intelligent layer satisfies the formula: ; in For query matrix, For the key matrix, For value matrices, The feature vector dimension is used; the model first extracts the temporal-frequency domain local features of the preprocessed EEG signal through CNN, then captures the long-term sleep cycle dependent features through the Transformer attention mechanism, and finally divides the sleep signal into 5 stages according to the AASM sleep stage standard: wakefulness, N1 stage, N2 stage, N3 deep sleep stage, and REM rapid eye movement stage. The core node feature update formula for the insomnia precision sub-module of the cloud-based intelligent layer to construct the multimodal fusion sub-model is as follows: ; in For the first Layered brain network nodes eigenvectors, For nodes The set of adjacent nodes, For neighborhood feature aggregation function, , For the first Learnable weights and biases of layers Using a nonlinear activation function, the model constructs a brain functional connectivity network with EEG channels as nodes, integrates sleep stage characteristics, EEG power spectrum characteristics, and clinical scale characteristics, and mines neurophysiological biomarkers of insomnia to achieve objective and accurate classification of insomnia subtypes.

5. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The insomnia precise subtyping submodule of the cloud-based intelligent layer has a built-in multimodal feature fusion unit, which achieves adaptive fusion of multi-source features based on an attention weighting mechanism. The fusion formula is as follows: ; For the first Each single-modal feature vector includes EEG frequency domain features, sleep structure features, and clinical scale features. For the first Adaptive attention weights for each modality, and satisfying The weight calculation formula is: ; in This is the mapping function for a multilayer perceptron.

6. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The AI ​​sleep auto-switching submodule of the cloud-based intelligent layer has a built-in EEG feature extraction unit that extracts the power spectral density (PSD) feature of the EEG signal based on the Welch method. The calculation formula is as follows: ; in Number of signal segments The energy of the window function, For the Hanning window function, For the first segment time domain signal, For Fast Fourier Transform, the absolute and relative energy characteristics of five sleep-related frequency bands—δ (0.5~4Hz), θ (4~8Hz), α (8~13Hz), β (13~30Hz), and standard sleep σ = 12~14Hz—are extracted based on the PSD results.

7. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The closed-loop intervention layer has a built-in personalized parameter generation unit. Based on the patient's insomnia subtype and EEG characteristics, the personalized parameter generation unit outputs core parameters such as stimulation target, stimulation frequency, stimulation intensity, stimulation pulse number, and treatment course. After treatment, it automatically compares the changes in sleep structure parameters and EEG characteristic markers before and after treatment to complete an objective quantitative evaluation of the efficacy. Based on the evaluation results, it iteratively optimizes the parameter model.

8. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The data management submodule follows medical data security standards, performs full lifecycle access control and anonymization of patient data, and the anonymization process meets irreversible anonymization rules, which can be directly connected to the hospital's electronic medical record system.

9. A brain-computer interface monitoring system for sleep disorder monitoring according to claim 2, characterized in that: The patient-side module of the application interaction layer supports functions such as viewing sleep reports, sleep quality scoring, health education, and follow-up reminders. The doctor-side module supports functions such as batch patient management, report review, issuance of treatment plans, and multi-center scientific research data statistics. It can generate sleep monitoring reports and subtype diagnosis reports that conform to clinical standards.

Images