A multi-modal electroencephalogram closed-loop sound stimulation method and system based on a Bluetooth audio SoC

Abstract

The application discloses a kind of multi-modal electroencephalogram closed-loop sound stimulation method and system based on Bluetooth audio SoC, the method comprises: based on audio analog-digital conversion link, the raw PCM signal of the electroencephalogram of target patient is obtained, and the contact quality of electrode and target patient skin is determined;Carry out multi-modal data preprocessing, determine the control candidate window and trigger inhibition condition;Combined with the preset actual application scene carries out control decision, obtains the accurate control stimulation signal of corresponding scene;Based on the accurate control stimulation signal of corresponding scene carries out stimulation output, obtains the structured control event complete record data set and repeatedly executes original data acquisition, multi-modal data preprocessing and control decision, realizes multi-modal electroencephalogram closed-loop sound stimulation.The application improves the applicability in different electroencephalogram closed-loop stimulation and multi-physiological state monitoring scene.The application is a kind of multi-modal electroencephalogram closed-loop sound stimulation method and system based on bluetooth audio SoC, can be widely applied in electroencephalogram stimulation monitoring technical field.

Description

Technical Field

[0001] This invention relates to the field of brain stimulation monitoring technology, and in particular to a multimodal closed-loop sound stimulation method and system based on Bluetooth audio SoC. Background Technology

[0002] Closed-loop EEG stimulation and wearable physiological monitoring devices are evolving from discrete laboratory instruments towards low-power, wireless, wearable, and multimodal fusion technologies. Existing EEG acquisition hardware primarily employs dedicated EEG acquisition and recording systems, dedicated analog front-ends, or multi-channel wireless EEG acquisition and recording systems. For example, high-sampling-rate, multi-channel wireless EEG recording systems are already available for EEG monitoring and steady-state visual evoked potential detection. These approaches demonstrate that dedicated EEG hardware already possesses high resolution and wireless capabilities, but the overall approach still revolves around dedicated EEG analog front-ends or dedicated acquisition devices.

[0003] In the area of ​​wearable closed-loop stimulation, wearable EEG systems have been developed to detect sleep-related EEG oscillations and trigger closed-loop acoustic stimulation. This demonstrates that EEG acquisition and acoustic stimulation can form a closed loop within wearable systems. Regarding multimodal wearable physiological signal sensing and monitoring platforms, recent ear-worn or head-worn platforms have demonstrated the feasibility of integrating audio, motion, physiological, and edge processing capabilities into small wearable forms.

[0004] However, existing EEG acquisition devices typically rely on dedicated EEG acquisition and recording systems or dedicated analog front-ends, lacking a publicly available hardware path for receiving EEG signals using an analog microphone input / audio analog-to-digital conversion link via a Bluetooth audio SoC and uploading EEG samples in raw PCM format. General recording links, geared towards voice communication, speech recognition, or auditory optimization, often include high-pass filtering, DC blocking, automatic gain control, dynamic range compression, noise reduction, speech activity detection, echo cancellation, active noise reduction, equalization, resampling, and speech coding. These processes may weaken low-frequency EEG rhythms and alter amplitude, phase, or frequency response, making them unsuitable for direct quantitative EEG acquisition. Furthermore, existing wearable closed-loop acoustic stimulation systems typically focus on closed-loop algorithms and stimulation effects, with insufficient description of hardware-firmware coordination sources such as sound playback link preparation, playback output stage readiness, and sound data output timing; different literatures also have inconsistent definitions of digital trigger delay and headphone acoustic output delay. Summary of the Invention

[0005] To address the aforementioned technical problems, the present invention aims to provide a multimodal EEG closed-loop acoustic stimulation method and system based on a Bluetooth audio SoC, thereby improving its applicability in various EEG closed-loop stimulation and multi-physiological state monitoring scenarios.

[0006] The first technical solution adopted in this invention is: a multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC, comprising the following steps: The raw PCM signal of the target patient's EEG was acquired based on the audio analog-to-digital conversion link, and the contact quality between the electrodes and the target patient's skin was determined. Multimodal data preprocessing was performed on the raw PCM signals of the target patient's EEG and the contact quality between the electrode and the target patient's skin to determine the candidate window for regulation and the triggering inhibition condition; Based on the candidate window for regulation and the triggering inhibition condition, and combined with the preset actual application scenario, regulation decision is made to obtain the precise regulation stimulus signal for the corresponding scenario. Based on the precise control of stimulation signals in the corresponding scenario, stimulation output is performed to obtain a structured complete dataset of control events. The process of acquiring raw data, preprocessing multimodal data, and making control decisions is repeated to achieve multimodal EEG closed-loop acoustic stimulation.

[0007] Furthermore, the step of acquiring the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link and determining the contact quality between the electrode and the target patient's skin specifically includes: The frontal simulated EEG signal of the target patient was acquired, and the sampled, pre-shunted and quantized signal was obtained through an audio analog-to-digital conversion link to obtain the raw PCM signal of the target patient's EEG. The photoplethysmography (PPG) signal, motion posture data, and electrode contact quality-related signals of the target patient are acquired, and time-division sampling and contact quality judgment are performed to determine the contact quality between the electrode and the target patient's skin.

[0008] Furthermore, the step of acquiring the simulated frontal EEG signal of the target patient, sampling, pre-shunting, and quantizing it through an audio analog-to-digital conversion link to obtain the raw PCM signal of the target patient's EEG specifically includes: Prefrontal EEG signals of the target patient were obtained using a dual-channel reusable electrode in the forehead. The frontal EEG signal was sequentially amplified and conditioned to obtain a frontal analog EEG signal suitable for the sampling range of the audio analog-to-digital conversion link. The simulated EEG signal from the forehead is connected to the analog microphone input port of the Bluetooth audio SoC, and the signal is sampled through the audio analog-to-digital conversion link to generate the raw sampled data stream; Preliminary raw PCM of EEG is obtained by pre-splitting the raw sampled data stream after analog-to-digital conversion; The initial raw PCM EEG signal was subjected to high-resolution quantization to obtain the raw PCM signal of the target patient.

[0009] Furthermore, the step of acquiring the target patient's photoplethysmography (PPG) signal, motion posture data, and electrode contact quality-related signals, and performing time-division sampling and contact quality assessment to determine the contact quality between the electrode and the target patient's skin, specifically includes: Based on the MCU-driven auxiliary physiological data acquisition link, the photoplethysmography pulse signal, motion posture data and electrode contact quality-related signals of the target patient are acquired. By constructing a detection reference path and a charge release path using a common electrode, the MCU performs time-division sampling of the photoplethysmography pulse signal, motion posture data, and electrode contact quality-related signals according to a preset timing sequence. Based on the signal amplitude, trend of change, and equivalent impedance calculation results, the contact quality between the electrode and the target patient's skin is determined.

[0010] Furthermore, the step of performing multimodal data preprocessing on the raw PCM signals from the target patient's EEG and the contact quality between the electrodes and the target patient's skin to determine the candidate window for regulation and the triggering inhibition condition specifically includes: Based on the target patient's raw PCM EEG signal, the contact quality between the electrode and the target patient's skin, the photoplethysmography pulse signal, and the motion posture data, feature analysis is performed to determine the target patient's EEG characteristics, the contact stability of each EEG channel, physiological state changes, and body movement characteristics. The EEG characteristics of the target patient are used as the verification object, and the contact stability of each EEG channel, the physiological state change characteristics and body movement characteristics are used as the quality control basis. Multi-dimensional data reliability joint verification is carried out to obtain reliable EEG characteristics. Based on reliable EEG characteristics, combined with preset real-world application scenarios, scenario-based multimodal fusion state assessment is performed to obtain a comprehensive assessment result of the target physiological state. Based on the comprehensive assessment results of reliable EEG characteristics and target physiological state, and combined with the preset actual application scenarios to distinguish the data usage methods, the candidate windows for regulation and the conditions for triggering inhibition are determined.

[0011] Furthermore, the step of making a control decision based on the control candidate window and the trigger inhibition condition, combined with a preset actual application scenario, to obtain a precise control stimulus signal for the corresponding scenario, specifically includes: Based on the comprehensive assessment results of the target physiological state, the preset actual application scenarios, and the regulatory candidate windows, a regulatory strategy scheme adapted to the current scenario is determined. Based on the control strategy scheme adapted to the current scenario, the trigger verification is completed in stages according to the scenario dimension to determine the final trigger judgment result, which includes triggering, delaying and suppressing. Based on the final trigger determination result and combined with the trigger suppression condition, the sound link preheating preprocessing is performed to obtain a ready sound output link; By combining the final trigger judgment result with the ready sound output link, multi-form regulation of stimulus output is performed to obtain precise regulation stimulus signals for the corresponding scenario.

[0012] Furthermore, the step of precisely regulating stimulation signals based on the corresponding scenario to output stimulation, obtaining a structured complete dataset of regulatory events, and repeatedly executing the original data acquisition, multimodal data preprocessing, and regulatory decision-making to achieve multimodal EEG closed-loop acoustic stimulation specifically includes: Based on the precise modulation stimulation signal of the corresponding scenario, determine the corresponding output mode and output the precise modulation stimulation signal acting on the target site; Based on precise regulatory stimulation signals acting on target sites, the entire stimulation chain data is synchronously recorded, and a structured dataset of complete records of regulatory events is output. Based on a complete dataset of regulatory events, the end of a single stimulus output is marked, and the processes of acquiring raw data, preprocessing multimodal data, and making regulatory decisions are repeatedly executed to achieve multimodal EEG closed-loop acoustic stimulation.

[0013] The second technical solution adopted in this invention is: a multimodal EEG closed-loop acoustic stimulation system based on a Bluetooth audio SoC, comprising: The first module is used to acquire the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link and determine the contact quality between the electrode and the target patient's skin. The second module is used to perform multimodal data preprocessing on the raw PCM signals of the target patient's EEG and the contact quality between the electrodes and the target patient's skin, and to determine the candidate window for regulation and the triggering inhibition conditions. The third module is used to make control decisions based on the control candidate window and the trigger inhibition condition, combined with the preset actual application scenario, so as to obtain the precise control stimulus signal for the corresponding scenario. The fourth module is used to precisely control the stimulation signal based on the corresponding scenario to output stimulation, obtain a structured complete record dataset of the control event, and repeatedly execute the original data acquisition, multimodal data preprocessing and control decision-making to realize multimodal EEG closed-loop acoustic stimulation.

[0014] The beneficial effects of the method and system of this invention are as follows: This invention acquires the raw PCM signal of the target patient's EEG based on an audio analog-to-digital conversion link and determines the contact quality between the electrodes and the target patient's skin. It utilizes the analog microphone input / audio analog-to-digital conversion link of a Bluetooth audio SoC to acquire the raw PCM data of the EEG, and performs offloading before the ordinary speech processing link, which helps to preserve low-frequency EEG rhythm and phase information. Electrode contact quality detection can be used to determine the reliability of the EEG channel. Multimodal data preprocessing is performed on the raw PCM signal of the target patient's EEG and the contact quality between the electrodes and the target patient's skin to determine the control candidate window and trigger inhibition conditions. Based on the control candidate window and trigger inhibition conditions, combined with preset actual... The system makes control decisions based on application scenarios, obtaining precise control stimulation signals for the corresponding scenarios. By using a preheating mechanism to move the playback link preparation process forward to the closed-loop triggering stage, the uncertainty of link establishment at the moment of triggering can be reduced. Based on the precise control stimulation signals for the corresponding scenarios, stimulation output is performed to obtain a structured complete record dataset of control events. The system repeatedly performs original data acquisition, multimodal data preprocessing, and control decisions to achieve multimodal EEG closed-loop acoustic stimulation. Depending on the waveform parameters and electrode positions, the system can be extended to transcranial DC stimulation, transcranial AC stimulation, or peripheral nerve stimulation through the electrical stimulation simulation output stage and relay / multiplexed electrode path, thereby improving the applicability of the system in different EEG closed-loop stimulation and multi-physiological state monitoring scenarios. Attached Figure Description

[0015] Figure 1 This is a flowchart of the steps of a multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to the present invention; Figure 2 This is a structural block diagram of a multimodal EEG closed-loop acoustic stimulation system based on a Bluetooth audio SoC according to the present invention. Figure 3 This is a schematic diagram of system communication and SoC hub provided in an embodiment of the present invention.

[0016] Figure 4 This is a schematic diagram of EEG acquisition and original PCM shunt provided in an embodiment of the present invention.

[0017] Figure 5 This is a schematic diagram of MCU-assisted acquisition and stimulation output provided in an embodiment of the present invention.

[0018] Figure 6 This is a schematic diagram of data acquisition and multimodal processing provided in an embodiment of the present invention.

[0019] Figure 7 This is a schematic diagram of the control decision and feedback record provided in an embodiment of the present invention.

[0020] Figure 8This is a schematic diagram comparing the EEG raw PCM acquisition and speech processing links provided in an embodiment of the present invention.

[0021] Figure 9 This is a schematic diagram of multimodal input and fusion evaluation provided in an embodiment of the present invention.

[0022] Figure 10 This is a schematic diagram of the regulatory decision and stimulus branch provided in an embodiment of the present invention.

[0023] Figure 11 This is a detailed schematic diagram of the stimulation output channel architecture provided in an embodiment of the present invention.

[0024] Figure 12 This is a schematic diagram comparing the preheating and non-preheating playback links of the sound link provided in an embodiment of the present invention. Detailed Implementation

[0025] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. The step numbers in the following embodiments are only for ease of explanation and do not limit the order of the steps. The execution order of each step in the embodiments can be adapted according to the understanding of those skilled in the art.

[0026] First, the technical terminology of this invention will be explained: 1) SoC (System on Chip): System on a chip refers to an integrated circuit that integrates a processor, memory, interface and dedicated function modules within a single chip.

[0027] 2) SPP (Serial Port Profile): Bluetooth serial port protocol, in this invention referring to the wireless data transmission channel between the Bluetooth audio SoC and the host computer, used to upload raw PCM EEG data and multimodal physiological data, and to send acquisition commands and stimulation parameters. The core is the transmission of raw data rather than being limited to a specific wireless protocol.

[0028] 3) PPG (Photoplethysmography): In this invention, it refers to acquiring pulse waveforms through photoelectric sensors to extract heart rate and blood oxygenation-related states, serving as one of the auxiliary inputs for multimodal physiological state assessment.

[0029] 4) tDCS (transcranial Direct Current Stimulation): Transcranial direct current stimulation, in this invention, refers to one of the forms of electrical stimulation that uses a programmable DAC of an MCU to generate a DC or low-frequency modulated waveform, which is then applied to the target brain region via an electrical stimulation analog output stage and multiplexed electrode paths.

[0030] 5) tACS (transcranial Alternating Current Stimulation): Transcranial alternating current stimulation, in this invention, refers to one of the forms of electrical stimulation that uses the programmable DAC of the MCU to generate AC or pulse modulation waveforms, which are then applied to the target brain region via the electrical stimulation analog output stage and multiplexed electrode path.

[0031] 6) IAF (Individual Alpha Frequency): In this invention, it refers to the estimation of the peak frequency of the alpha band of an individual subject through spectral analysis of the raw PCM of the EEG, which is used as a personalized trigger phase or frequency reference for closed-loop sound stimulation.

[0032] 7) ADC (Analog-to-Digital Converter): An analog-to-digital converter used to convert analog voltage signals into digital sample values. In this invention, the audio analog-to-digital conversion link refers to the digitization link within the Bluetooth audio SoC that faces the analog microphone input.

[0033] 8) PCM (Pulse Code Modulation): Pulse code modulation data. In this invention, it refers to the raw digital sample stream formed after audio analog-to-digital conversion, rather than audio features after speech coding or auditory optimization.

[0034] 9) MCU (Microcontroller Unit): A microcontroller used for multimodal physiological data acquisition, electrode contact quality detection, ADC time-division sampling, relay control, and programmable DAC waveform output.

[0035] 10) DAC (Digital-to-Analog Converter): A digital-to-analog converter used to convert digital waveforms into analog voltage waveforms. In this invention, it can be used for sound output as well as as an output source for programmable electrical stimulation waveforms.

[0036] 11) IMU (Inertial Measurement Unit): An inertial measurement unit used to collect acceleration and angular velocity data, reflecting the wearer's status, body motion, and attitude changes.

[0037] 12) bit: binary bit, used in this invention to describe the effective digital bit width of the raw PCM sample of EEG.

[0038] 13) dB (decibel): In this invention, dB is used to represent the gain of an analog link.

[0039] 14) uV / count, nV / count: These represent the microvolt or nanovolt voltage change corresponding to each digital count, respectively, and are used to describe the quantization step or equivalent input resolution scale.

[0040] This invention provides a multimodal EEG closed-loop acoustic stimulation method and system based on a Bluetooth audio SoC. The method enables EEG signals to be input via an analog microphone / audio analog-to-digital conversion link to form raw PCM data, which is then streamed and uploaded before ordinary speech processing. Simultaneously, through sound link preheating, multimodal low-power acquisition, electrode contact quality detection, relay / multiplexed electrode paths, and programmable electrical stimulation extension, the system can realize EEG acquisition, acoustic stimulation triggering, physiological state monitoring, and state coordination between the acquisition link and the stimulation link on wearable hardware.

[0041] Reference Figure 1 This invention provides a multimodal EEG closed-loop acoustic stimulation method based on a Bluetooth audio SoC, the method comprising the following steps: S100: Acquire the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link, and determine the contact quality between the electrode and the target patient's skin; S110. Acquire the frontal simulated EEG signal of the target patient, and perform sampling, pre-shunting and quantization processing through the audio analog-to-digital conversion link to obtain the raw PCM signal of the target patient's EEG. Specifically, the frontal EEG signal of the target patient is acquired through a dual-channel multiplexed electrode in the frontal region; the frontal EEG signal is then amplified and conditioned sequentially to obtain a frontal analog EEG signal suitable for the sampling range of the audio analog-to-digital conversion link; the frontal analog EEG signal is connected to the analog microphone input port of the Bluetooth audio SoC, and the signal is sampled through the audio analog-to-digital conversion link to generate a raw sampled data stream; the raw sampled data stream after analog-to-digital conversion is pre-splitting to obtain a preliminary raw PCM of the EEG; the preliminary raw PCM of the EEG is then subjected to high-resolution quantization processing to obtain the raw PCM signal of the target patient's EEG.

[0042] In this embodiment, this stage includes two pathways: raw PCM acquisition from EEG and auxiliary physiological data acquisition. For example... Figure 4 As shown, in the EEG link, after the user wears the frontal dual-channel multiplexed electrode, the frontal EEG signal is first amplified and conditioned by analog, so that the weak EEG potential is adjusted to a range suitable for the audio analog-to-digital conversion link to sample, and then connected to the analog microphone input / audio analog-to-digital conversion link of the Bluetooth audio SoC.

[0043] like Figure 3As shown, typical recording or voice communication links are usually geared towards human voice communication, call noise reduction, or auditory optimization. Subsequent processing may include DC blocking, high-pass filtering, automatic gain control, dynamic range compression, environmental noise reduction, voice activity detection, acoustic echo cancellation, active noise reduction, equalization processing, resampling, and speech coding. While these processes are beneficial for voice communication, they may have adverse effects on EEG signals.

[0044] Specifically, the low-frequency rhythms, slow-varying trends, phase relationships, and amplitude ratios of EEG signals, which are of interest, may be weakened by DC blocking or high-pass filtering; automatic gain control and dynamic range compression may alter the true ratio of EEG amplitudes across different time periods; noise reduction, speech activity detection, echo cancellation, and active noise reduction may misclassify non-speech low-frequency EEG components as noise or background components; resampling and speech coding may alter waveform details and phase relationships. Therefore, ordinary recording links are not suitable as the primary acquisition link for closed-loop EEG modulation.

[0045] The link used in this invention does not process EEG as a regular recording signal. Instead, after the raw sampling data is generated by the analog microphone input / audio analog-to-digital conversion link of the Bluetooth audio SoC, the raw EEG PCM is diverted before regular speech processing. This raw EEG PCM bypasses the processing steps for speech communication and auditory optimization and is transmitted to a host computer or equivalent processing unit via a wireless link, making it more suitable for preserving and analyzing low-frequency EEG rhythms, phase, and amplitude changes.

[0046] Scalp EEG signals are typically in the microvolt range, with common rhythmic components and event-related waveforms ranging from several to tens of microvolts. To reproduce the amplitude variations and phase relationships of EEG rhythms after digitization, the equivalent quantization step size at the electrode input should be significantly smaller than the amplitude of the EEG feature being analyzed. According to the uniform quantization model, the maximum quantization error is no more than half a quantization step size, and the equivalent root-mean-square quantization noise is approximately the quantization step size divided by the square root of 12. Therefore, for EEG features in the 5-10 µV range, if the equivalent quantization step size at the electrode input is below 1 µV, the quantization error is already below the main amplitude variation range of the target signal; if it is further reduced to below 0.1 µV, a finer amplitude representation margin can be provided for weaker EEG fluctuations. In one implementation, the raw EEG PCM is carried in a 32-bit integer container, where the effective EEG sample is interpreted in 24-bit increments. Based on a 1.2 V reference voltage, the theoretical quantization step size at the audio analog-to-digital converter is approximately 0.143 uV / count. After further considering a theoretical link gain of approximately 34 dB and a preamplifier amplification of approximately 140x, the theoretical equivalent input quantization step size at the electrode input is calculated to be approximately 0.020 nV / count. This theoretical equivalent input quantization step size is significantly smaller than the aforementioned 0.1-1 uV level requirement for EEG acquisition quantization, demonstrating that the original PCM link in this solution possesses the digital sample representation capability suitable for EEG rhythm analysis.

[0047] S120. Acquire the photoplethysmography (PPG) signal, motion posture data, and electrode contact quality-related signals of the target patient, and perform time-division sampling and contact quality judgment to determine the contact quality between the electrode and the target patient's skin.

[0048] Specifically, based on the MCU carrying the auxiliary physiological data acquisition link, the photoplethysmography (PPG) signal, motion posture data, and electrode contact quality related signals of the target patient are acquired. A detection reference path and charge release path are constructed through a common electrode. The MCU performs time-division sampling of the PPG signal, motion posture data, and electrode contact quality related signals according to a preset time sequence. Based on the signal amplitude, change trend, and equivalent impedance calculation results, the contact quality between the electrode and the target patient's skin is determined.

[0049] In this embodiment, as Figure 5 As shown, the auxiliary physiological data acquisition link is completed by the MCU. The MCU collects data such as photoplethysmography (PPG), motion posture, and electrode contact quality. Among them, PPG is used to obtain heart rate and blood oxygenation status, motion posture is used to obtain information on body movement, posture changes, and wearing disturbances, and electrode contact quality is used to determine the reliability of contact between the forehead electrode and the skin.

[0050] During electrode contact quality detection, the system forms a detection reference or charge release path through a common electrode terminal, and the MCU samples the detection response signals of each electrode channel in a time-division manner according to a preset timing sequence. Based on the amplitude, trend of change, or equivalent impedance result of the response signal of each channel, the contact quality between the corresponding electrode and the skin can be determined, and this contact quality result will be used for subsequent data reliability assessment.

[0051] S200: Perform multimodal data preprocessing on the raw PCM signals of the target patient's EEG and the contact quality between the electrode and the target patient's skin to determine the candidate window for regulation and the triggering inhibition condition; In this embodiment, as Figure 6 As shown, the system receives raw PCM data from EEG, heart rate / blood oxygenation data, motion posture data, and electrode contact quality data, and converts them into state information tailored to different application scenarios. The processing goal is not simply to display multiple data streams, but to identify the state changes reflected by various physiological signals within the current time period, assess data reliability, and determine suitability for closed-loop regulation.

[0052] S210. Based on the target patient's raw PCM EEG signal, the contact quality between the electrode and the target patient's skin, the photoplethysmography pulse signal, and the motion posture data, feature analysis is performed to determine the target patient's EEG characteristics, the contact stability of each EEG channel, physiological state change characteristics, and body movement characteristics. S220. Using the EEG characteristics of the target patient as the verification object, and using the contact stability of each EEG channel, physiological state change characteristics and body movement characteristics as the quality control basis, a multi-dimensional data reliability joint verification is carried out to obtain reliable EEG characteristics. In this embodiment, during EEG data processing, the host computer or equivalent processing unit can perform filtering, notch filtering, spectrum analysis, rhythm analysis, individualized alpha frequency estimation, phase estimation, and event window extraction on the raw PCM EEG data. The EEG data is primarily used to reflect neural rhythms, phase relationships, target frequency band energy changes, and EEG features that can be used for closed-loop triggering.

[0053] Heart rate / blood oxygenation data are used to help describe changes in overall physiological state, such as physiological stability, changes in arousal or tension levels, respiratory fluctuations, fatigue trends, or state transitions. This type of data can serve as a reference signal for arousal, body movement, respiratory fluctuations, or changes in physiological state, and is suitable as an auxiliary physiological basis for judging the current timing of regulation and interpreting changes in electroencephalogram (EEG).

[0054] Motor posture data is used to identify rolling over, head lifting, posture changes, wear perturbations, motion artifacts, and activity status. If IMU data indicates significant body movement or wear perturbations, the system can mark that time period as having low EEG data reliability, reducing its impact on closed-loop trigger judgment, or suspend trigger judgment to avoid mistaking motion artifacts for genuine EEG rhythm changes.

[0055] Electrode contact quality data is used to determine the reliability of the EEG channel. If the contact quality is poor, the system can mark the corresponding EEG segment, suppress closed-loop triggering, or prompt for re-wearing; if the contact quality is stable, the time segment can be allowed to enter subsequent EEG rhythm analysis, target state assessment, and modulation candidate window screening.

[0056] S230. Based on reliable EEG characteristics, combined with preset real-world application scenarios, scenario-based multimodal fusion state assessment is performed to obtain a comprehensive assessment result of the target physiological state. In this embodiment, during the multimodal state assessment process, the system can select different data combinations according to the specific application scenario. For example, in a sleep scenario, more attention can be paid to EEG rhythm, heart rate / blood oxygen stability, and body movement changes; in an emotion recognition or stress detection scenario, more attention can be paid to changes in EEG rhythm, heart rate fluctuations, blood oxygen-related states, and postural activities; in a fatigue monitoring scenario, a comprehensive judgment can be made by combining EEG rhythm, reduced body movement, postural changes, and changes in physiological stability.

[0057] S240. Based on the comprehensive evaluation results of reliable EEG characteristics and target physiological state, and combined with the preset actual application scenarios, distinguish the data usage methods and determine the regulatory candidate window and trigger inhibition conditions.

[0058] In this embodiment, the output of the multimodal data processing stage includes target physiological state estimation, EEG phase or rhythm state, data reliability score, modulation candidate window, and trigger inhibition condition. For closed-loop sound modulation scenarios based solely on EEG, heart rate / blood oxygen and IMU can serve as auxiliary recordings or quality references; for multimodal state monitoring scenarios, these auxiliary data can directly participate in state judgment and candidate modulation window selection.

[0059] S300: Based on the candidate window for regulation and the triggering inhibition condition, and combined with the preset actual application scenario, regulation decision is made to obtain the precise regulation stimulus signal for the corresponding scenario. In this embodiment, as Figure 7 As shown, the control method is selected based on the data processing results and the target application scenario. For different scenarios such as sleep, emotion recognition, stress detection, fatigue monitoring, and closed-loop regulation of EEG rhythms, the system can adopt different feature combinations, triggering conditions, and stimulation strategies to transform multimodal state assessment results into specific control actions. The core of this stage lies in transforming multiple physiological states into specific stimulation strategies, rather than outputting a fixed single stimulus form.

[0060] S310. Based on the comprehensive assessment results of the target physiological state, the preset actual application scenario, and the control candidate window, determine the control strategy scheme that is suitable for the current scenario. S320. Based on the control strategy scheme adapted to the current scenario, complete the trigger verification by scenario dimension and determine the final trigger judgment result, which includes triggering, delaying and suppressing. In this embodiment, for closed-loop sound modulation scenarios primarily based on EEG, the system can trigger sound stimulation according to EEG rhythm, individualized alpha frequency, phase state, or event markers. For scenarios requiring multimodal auxiliary judgment, the system can further combine heart rate / blood oxygenation, IMU, and electrode contact quality results to confirm, delay, or suppress the triggering timing.

[0061] S330. Based on the final trigger determination result and combined with the trigger suppression condition, perform sound link preheating preprocessing to obtain a ready sound output link; In this embodiment, as Figure 12 As shown, sound link preheating is part of the output preprocessing in the control decision-making stage. Before the closed-loop trigger or the target state meets the candidate conditions, the system pre-prepares the sound playback link, putting the sound output in a ready-to-output state. When the trigger conditions are met, the system mainly updates the stimulus parameters and starts the sound output, thereby reducing the timing uncertainty caused by temporarily establishing the sound link at the moment of triggering.

[0062] S340: Combining the final trigger judgment result with the ready sound output link, perform multi-form control stimulus output to obtain the precise control stimulus signal for the corresponding scenario.

[0063] In this embodiment, electrical stimulation extension is used as one of the control methods. The MCU generates configurable DC, AC, pulse, or envelope modulated waveforms via a DAC. These waveforms are output through an electrical stimulation analog output stage and connected to the corresponding stimulation sites via relay / multiplexed electrode paths. By adjusting the DC component, AC frequency, waveform amplitude, duration, envelope form, and electrode position, it can be extended to implementations such as transcranial DC stimulation, transcranial AC stimulation, or peripheral nerve stimulation.

[0064] S400, based on the precise control of stimulation signals in the corresponding scenario, outputs stimulation, obtains a structured complete record dataset of control events, and repeatedly executes the original data acquisition, multimodal data preprocessing, and control decision-making to achieve multimodal EEG closed-loop acoustic stimulation.

[0065] S410. Based on the precise control stimulus signal of the corresponding scenario, determine the corresponding output mode and output the precise control stimulus signal acting on the target site. In this embodiment, the system outputs sound stimulation, programmable electrical stimulation, or, in application scenarios requiring synchronized control, coordinated sound and electrical stimulation. Sound stimulation can be output through the sound output unit; electrical stimulation can be output through the electrical stimulation simulation output stage and multiplexed electrode path; synchronized control records the correspondence between different stimulation forms within the same session.

[0066] S420: Based on the precise regulatory stimulus signal acting on the target site, synchronously record the entire stimulus chain data and output a structured complete record dataset of regulatory events. In this embodiment, as Figure 11 As shown, during the output process, the system simultaneously records the stimulus event, stimulus type, stimulus time, stimulus parameters, triggering basis, EEG segments before and after stimulation, heart rate / blood oxygenation status, IMU status, electrode contact quality, and session configuration. These records describe the multimodal state background during a single modulation event and provide a complete basis for subsequent experimental review.

[0067] S430: Based on the complete recording dataset of regulatory events, mark the end of a single stimulus output, and repeatedly execute the original data acquisition, multimodal data preprocessing, and regulatory decision-making to achieve multimodal EEG closed-loop acoustic stimulation.

[0068] In this embodiment, after a stimulus output ends, the system returns to the data acquisition and multimodal data processing stage to continue forming the next closed loop. Thus, this invention forms a continuous technical path of "raw data acquisition - multimodal state analysis - scenario-based control decision-making - stimulus output and feedback recording," highlighting both the core acquisition method of Bluetooth audio SoC raw PCM EEG acquisition and the specific applications of heart rate / blood oxygen, IMU, and contact quality data in various scenarios.

[0069] The user wears a dual-channel multiplexed electrode and a sound output unit on the forehead. After the system is powered on via a button, the Bluetooth audio SoC initializes the analog microphone input, wireless communication, and sound output resources. The MCU enters the preparation state for auxiliary physiological data acquisition and impedance / contact quality detection. After establishing a connection with the host computer, the device begins to acquire data such as raw EEG PCM, photoplethysmography pulse, motor posture, and electrode contact quality.

[0070] like Figure 8 As shown, in the data acquisition embodiment, the EEG signal is amplified by an analog amplifier circuit and input to two analog microphone input / audio analog-to-digital conversion channels. Before ordinary speech processing, a raw EEG PCM is generated and transmitted to the host computer via a wireless link. This link bypasses the DC blocking, high-pass, automatic gain control, noise reduction, and speech coding processes in ordinary recording links that may affect low-frequency EEG rhythms.

[0071] In the embodiment of assisted physiological data acquisition, photoplethysmography (PPG), motor posture, and electrode contact quality are acquired by the MCU and combined with the raw PCM data from the EEG to form unified conversational data. During electrode contact quality detection, the system forms a detection reference or charge release path through a common electrode terminal, and the MCU samples the detection response signals of each electrode channel in a time-division manner according to a preset timing sequence to obtain contact quality information for different channels.

[0072] like Figure 9 As shown in the multimodal state monitoring and closed-loop control embodiment, the host computer performs rhythm, phase, or target EEG feature analysis on the raw PCM EEG data. Simultaneously, heart rate / blood oxygenation-related states are used to assist in judging physiological stability, changes in tension or relaxation, arousal changes, fatigue trends, or respiratory fluctuations. IMU motion posture is used to identify body movement, posture changes, activity states, and motion artifacts. Based on this, the system can form the target application state, data reliability, and candidate control opportunities.

[0073] In the closed-loop sound modulation embodiment, the system can primarily determine whether to trigger sound stimulation based on EEG rhythm, individualized alpha frequency, alpha phase, or other event conditions. Sound link preheating, as a pre-processing step before output, can complete playback link preparation before the closed-loop triggering or target state meets candidate conditions. Upon triggering, the system enters the stimulation parameter update and sound output process.

[0074] In the extended electrical stimulation embodiment, the MCU can generate a programmable digital waveform according to preset parameters or host computer configuration, convert it into an analog waveform via a DAC, and then condition the output through the analog output stage of electrical stimulation before outputting it to the corresponding stimulation site via a relay / multiplexed electrode path. This extension can be used for transcranial DC stimulation or transcranial AC stimulation research at the forehead electrode position, and can also be used as a peripheral nerve stimulation extension after changing the stimulation electrode position and output parameters.

[0075] Therefore. Figure 10 and Figure 11 As shown, this embodiment's system completes the acquisition of raw PCM EEG data, multimodal state analysis, control decision-making, sound link warm-up, stimulation output, and feedback recording in a staged closed-loop manner. Heart rate / blood oxygen and IMU data are no longer merely supplementary acquisition data, but are used to assist in judging the target physiological state, data reliability, and motion artifacts. Electrode contact quality detection serves as an auxiliary input for judging data reliability, while electrical stimulation extension focuses on programmable DAC waveforms, electrical stimulation analog output stages, and multiplexed electrode paths. Compared with common 24-bit EEG acquisition links, the advantage of this solution should not be summarized as a single analog-to-digital conversion with absolutely higher accuracy, but rather as using the same low-power Bluetooth audio SoC for raw PCM EEG acquisition, wireless transmission, sound stimulation, and wearable multimodal monitoring.

[0076] In summary, the embodiments of the present invention have the following distinguishing technical features compared with the prior art: 1) The analog microphone input / audio analog-to-digital conversion channel of the Bluetooth audio SoC is reused as the frontal EEG acquisition entry point, and the EEG PCM data is sent using the raw sampled data upload mechanism, instead of entering the ordinary speech coding or auditory optimization link.

[0077] 2) The EEG data is shunted at the original PCM node formed by audio analog-to-digital conversion, and processed by bypass DC blocking, high-pass, automatic gain, dynamic range compression, noise reduction, speech activity detection, echo cancellation, active noise reduction, equalization, resampling and speech coding, which may change the low-frequency rhythm of EEG; and 24 bits of valid EEG samples are carried in a 32-bit integer container.

[0078] 3) The system combines EEG rhythm / phase, heart rate / blood oxygenation status, IMU movement posture, and electrode contact quality for multimodal physiological state assessment, data reliability judgment, regulation candidate window screening, and trigger inhibition condition generation. The system can select different feature combinations and judgment criteria according to the target application scenario (such as sleep monitoring, emotion recognition, stress detection, fatigue monitoring, EEG rhythm closed-loop regulation, etc.).

[0079] 4) Set up a sound link warm-up step to complete the sound playback link preparation in advance during the closed-loop waiting-to-trigger stage, so that the main execution parameters are updated and the output is started when triggered, and the digital trigger response record is included in the host computer session data.

[0080] 5) By configuring the MCU's programmable DAC stimulation output, electrical stimulation analog output stage, relay / multiplexed electrode path, and electrode position, it can be extended to different implementation forms such as transcranial DC stimulation, transcranial AC stimulation, or peripheral nerve stimulation; and the EEG, motor posture, photoplethysmography, electrode contact quality, stimulation type, stimulation time, and triggering basis can be collected, uploaded, and recorded in the same low-power wireless session.

[0081] Therefore, the embodiments of the present invention have the following advantages compared with the prior art: 1) Compared with existing solutions that use dedicated EEG acquisition and recording systems, the embodiments of this invention utilize the analog microphone input / audio analog-to-digital conversion link and raw PCM data link of the Bluetooth audio SoC to acquire EEG data. This reduces reliance on dedicated EEG acquisition hardware and forms an acquisition path suitable for wearable low-power devices. This invention does not present the prior art as a precedent of "Bluetooth audio chip acquiring raw PCM EEG data," but rather compares it with existing dedicated EEG acquisition routes to highlight the innovative raw PCM acquisition of this invention.

[0082] 2) From the perspective of data integrity and quantization capability, this embodiment of the invention does not use a conventional recording link to acquire EEG, but instead bypasses the process to obtain the raw PCM, reducing the impact of DC blocking, high-pass filtering, automatic gain control, noise reduction, and speech coding on the low-frequency EEG rhythm, phase, and amplitude relationships. Scalp EEG is typically a microvolt-level signal. The uniform quantization model shows that the smaller the quantization step size is than the target EEG characteristic amplitude, the smaller the impact of quantization error on waveform reproduction and rhythm analysis. In one embodiment, a 32-bit integer transmission container retains 24 bits of valid EEG samples. Combined with a 1.2 V reference voltage, approximately 34 dB theoretical link gain, and approximately 140x amplification in the pre-amplifier stage, a theoretical equivalent resolution scale of approximately 0.143 uV / count at the audio analog-to-digital converter and approximately 0.020 nV / count at the electrode input stage can be formed. This quantization capability is on the same order of magnitude as a common 24-bit EEG acquisition link, which demonstrates that the raw PCM link has a sample representation capability suitable for EEG analysis.

[0083] 3) Compared with existing wearable EEG closed-loop acoustic stimulation systems, this invention further clarifies the acoustic link preheating mechanism, moving steps such as acoustic playback link preparation, parameter configuration, and playback output readiness forward from the trigger critical path, reducing the delay and timing uncertainty from triggering to acoustic output; simultaneously, it incorporates heart rate / blood oxygen, IMU, and electrode contact quality into multimodal physiological state assessment, data reliability judgment, and trigger inhibition, enabling the system to adapt to various application scenarios such as sleep monitoring, emotion recognition, stress detection, fatigue monitoring, and EEG rhythm closed-loop regulation. The core contribution of this invention lies in unifying the original PCM sample expression capability, Bluetooth audio SoC wireless SPP link, acoustic stimulation output, programmable electrical stimulation extension, and multimodal state recording into a single wearable closed-loop system.

[0084] Reference Figure 2 A multimodal EEG closed-loop acoustic stimulation system based on a Bluetooth audio SoC, comprising: The first module 201 is used to acquire the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link and determine the contact quality between the electrode and the target patient's skin. The second module 202 is used to perform multimodal data preprocessing on the raw PCM signals of the target patient's EEG and the contact quality between the electrodes and the target patient's skin, and to determine the candidate window for regulation and the triggering inhibition conditions. The third module 203 is used to make control decisions based on the control candidate window and the trigger inhibition condition, combined with the preset actual application scenario, so as to obtain the precise control stimulus signal for the corresponding scenario. The fourth module 204 is used to precisely control the stimulation signal based on the corresponding scenario to output stimulation, obtain a structured complete record dataset of the control event, and repeatedly execute the original data acquisition, multimodal data preprocessing and control decision-making to realize multimodal EEG closed-loop acoustic stimulation.

[0085] The content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0086] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this is not intended to limit the scope of the embodiments of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the embodiments of the present application.

Claims

1. A multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC, characterized in that, Includes the following steps: The raw PCM signal of the target patient's EEG was acquired based on the audio analog-to-digital conversion link, and the contact quality between the electrodes and the target patient's skin was determined. Multimodal data preprocessing was performed on the raw PCM signals of the target patient's EEG and the contact quality between the electrode and the target patient's skin to determine the candidate window for regulation and the triggering inhibition condition; Based on the candidate window for regulation and the triggering inhibition condition, and combined with the preset actual application scenario, regulation decision is made to obtain the precise regulation stimulus signal for the corresponding scenario. Based on the precise control of stimulation signals in the corresponding scenario, stimulation output is performed to obtain a structured complete dataset of control events. The process of acquiring raw data, preprocessing multimodal data, and making control decisions is repeated to achieve multimodal EEG closed-loop acoustic stimulation.

2. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 1, characterized in that, The step of acquiring the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link and determining the contact quality between the electrodes and the target patient's skin specifically includes: The frontal simulated EEG signal of the target patient was acquired, and the sampled, pre-shunted and quantized signal was obtained through an audio analog-to-digital conversion link to obtain the raw PCM signal of the target patient's EEG. The photoplethysmography (PPG) signal, motion posture data, and electrode contact quality-related signals of the target patient are acquired, and time-division sampling and contact quality judgment are performed to determine the contact quality between the electrode and the target patient's skin.

3. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 2, characterized in that, The step of acquiring the simulated frontal EEG signal of the target patient, sampling, pre-shunting, and quantizing it through an audio analog-to-digital conversion link to obtain the raw PCM signal of the target patient's EEG, specifically includes: Prefrontal EEG signals of the target patient were obtained using a dual-channel reusable electrode in the forehead. The frontal EEG signal was sequentially amplified and conditioned to obtain a frontal analog EEG signal suitable for the sampling range of the audio analog-to-digital conversion link. The simulated EEG signal from the forehead is connected to the analog microphone input port of the Bluetooth audio SoC, and the signal is sampled through the audio analog-to-digital conversion link to generate the raw sampled data stream; Preliminary raw PCM of EEG is obtained by pre-splitting the raw sampled data stream after analog-to-digital conversion; The initial raw PCM EEG signal was subjected to high-resolution quantization to obtain the raw PCM signal of the target patient.

4. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 3, characterized in that, The step of acquiring the photoplethysmography (PPG) signal, motion posture data, and electrode contact quality-related signals of the target patient, and performing time-division sampling and contact quality judgment to determine the contact quality between the electrode and the target patient's skin, specifically includes: Based on the MCU-driven auxiliary physiological data acquisition link, the photoplethysmography pulse signal, motion posture data and electrode contact quality-related signals of the target patient are acquired. By constructing a detection reference path and a charge release path using a common electrode, the MCU performs time-division sampling of the photoplethysmography pulse signal, motion posture data, and electrode contact quality-related signals according to a preset timing sequence. Based on the signal amplitude, trend of change, and equivalent impedance calculation results, the contact quality between the electrode and the target patient's skin is determined.

5. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 4, characterized in that, The step of performing multimodal data preprocessing on the raw PCM signals from the target patient's EEG and the contact quality between the electrodes and the target patient's skin to determine the candidate windows for regulation and the triggering inhibition conditions specifically includes: Based on the target patient's raw PCM EEG signal, the contact quality between the electrode and the target patient's skin, the photoplethysmography pulse signal, and the motion posture data, feature analysis is performed to determine the target patient's EEG characteristics, the contact stability of each EEG channel, physiological state changes, and body movement characteristics. The EEG characteristics of the target patient are used as the verification object, and the contact stability of each EEG channel, the physiological state change characteristics and body movement characteristics are used as the quality control basis. Multi-dimensional data reliability joint verification is carried out to obtain reliable EEG characteristics. Based on reliable EEG characteristics, combined with preset real-world application scenarios, scenario-based multimodal fusion state assessment is performed to obtain a comprehensive assessment result of the target physiological state. Based on the comprehensive assessment results of reliable EEG characteristics and target physiological state, and combined with the preset actual application scenarios to distinguish the data usage methods, the candidate windows for regulation and the conditions for triggering inhibition are determined.

6. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 5, characterized in that, The step of making a control decision based on the control candidate window and the trigger inhibition condition, combined with a preset actual application scenario, to obtain a precise control stimulus signal for the corresponding scenario, specifically includes: Based on the comprehensive assessment results of the target physiological state, the preset actual application scenarios, and the regulatory candidate windows, a regulatory strategy scheme adapted to the current scenario is determined. Based on the control strategy scheme adapted to the current scenario, the trigger verification is completed in stages according to the scenario dimension to determine the final trigger judgment result, which includes triggering, delaying and suppressing. Based on the final trigger determination result and combined with the trigger suppression condition, the sound link preheating preprocessing is performed to obtain a ready sound output link; By combining the final trigger judgment result with the ready sound output link, multi-form regulation of stimulus output is performed to obtain precise regulation stimulus signals for the corresponding scenario.

7. The multimodal EEG closed-loop acoustic stimulation method based on Bluetooth audio SoC according to claim 6, characterized in that, The step of precisely modulating stimulation signals based on the corresponding scenario to output stimulation results in a structured dataset of complete records of modulated events. This process involves repeatedly performing raw data acquisition, multimodal data preprocessing, and modulation decision-making to achieve multimodal EEG closed-loop acoustic stimulation. Specifically, this includes: Based on the precise modulation stimulation signal of the corresponding scenario, determine the corresponding output mode and output the precise modulation stimulation signal acting on the target site; Based on precise regulatory stimulation signals acting on target sites, the entire stimulation chain data is synchronously recorded, and a structured dataset of complete records of regulatory events is output. Based on a complete dataset of regulatory events, the end of a single stimulus output is marked, and the processes of acquiring raw data, preprocessing multimodal data, and making regulatory decisions are repeatedly executed to achieve multimodal EEG closed-loop acoustic stimulation.

8. A multimodal EEG closed-loop acoustic stimulation system based on a Bluetooth audio SoC, characterized in that, Includes the following modules: The first module is used to acquire the raw PCM signal of the target patient's EEG based on the audio analog-to-digital conversion link and determine the contact quality between the electrode and the target patient's skin. The second module is used to perform multimodal data preprocessing on the raw PCM signals of the target patient's EEG and the contact quality between the electrodes and the target patient's skin, and to determine the candidate window for regulation and the triggering inhibition conditions. The third module is used to make control decisions based on the control candidate window and the trigger inhibition condition, combined with the preset actual application scenario, so as to obtain the precise control stimulus signal for the corresponding scenario. The fourth module is used to precisely control the stimulation signal based on the corresponding scenario to output stimulation, obtain a structured complete record dataset of the control event, and repeatedly execute the original data acquisition, multimodal data preprocessing and control decision-making to realize multimodal EEG closed-loop acoustic stimulation.

Images