A closed loop directed memory reactivation system and method

Abstract

The application belongs to the technical field of sleep neuroscience, and discloses a closed-loop directional memory reactivation system and method. The method comprises the following steps: a daytime learning stage, in which a binding relationship between learning materials and sensory stimulation clues is established; a nighttime sleep stage, in which electroencephalogram signals are collected in real time through a wearable electroencephalogram device, sleep staging is distinguished, and when it is detected that a user enters a steady state of slow wave sleep, the bound sensory stimulation clues (such as olfactory and / or auditory clues) are triggered to be output; the electroencephalogram is continuously monitored during the stimulation, and if micro-awakening is found, the stimulation is immediately interrupted; the memory effect is verified the next day, and a comprehensive report is generated. The application also discloses a system for realizing the method. The application realizes TMR intervention by constructing a "learning-stimulation-verification" whole-process closed loop, solves the problems of fragmented process, inaccurate control and chaotic data in the prior art, and improves the reproducibility and application value of the scheme.

Description

Technical Field

[0001] This invention belongs to the field of sleep neuroscience technology, specifically relating to a closed-loop directional memory reactivation system and method. Background Technology

[0002] Targeted memory reactivation (TMR) technology aims to selectively strengthen the consolidation of specific memories by re-presenting sensory cues (such as specific smells or sounds) associated with previously learned content during sleep. Existing research has demonstrated the great potential of TMR in improving declarative memory, procedural memory, and other areas.

[0003] However, in the process of realizing this invention, the inventors discovered at least the following problems in the prior art: 1. Most TMR experiments are fragmented, from daytime learning and cue binding to nighttime stimulation and the next day's memory verification, typically relying on multiple independent devices and software operated manually. This makes the experimental process cumbersome, difficult to standardize and scale, and has poor reproducibility.

[0004] 2. Many systems rely on offline analysis or coarse online judgment based on simple thresholds to determine the slow-wave sleep (SWS) stage. This leads to inaccurate timing of stimulus triggering, and may even trigger at inappropriate sleep stages (such as REM or light sleep), affecting effectiveness or disrupting sleep. There is a lack of a stable, low-latency, real-time closed-loop control system that can be deployed in a home environment.

[0005] 3. In studies that use multiple sensory cues such as smell and hearing simultaneously, it is often difficult to achieve precise synchronous output and quantitative control of dosage of the two cues under the same triggering conditions, which affects the standardization of stimuli.

[0006] 4. Experimental data recording is often incomplete, lacking unified and structured storage of raw EEG data, precise timestamps for each stimulus, stimulus parameters, online staging labels, and any interruption events. This poses obstacles to subsequent data analysis, research compliance review, and multi-center studies.

[0007] 5. Significant differences exist among individuals in sleep structure and arousal thresholds to stimuli. Existing solutions generally lack the ability to adaptively adjust to these individual differences, and also lack mechanisms for rapid detection and immediate interruption of micro-arousals induced by stimuli, posing a risk of overstimulation or disruption of normal sleep structure. Summary of the Invention

[0008] This invention aims to at least partially solve the aforementioned technical problems. Therefore, the purpose of this invention is to provide a closed-loop targeted memory reactivation system and method, which achieves precise, safe, and personalized memory enhancement by integrating the entire process of daytime learning, nighttime closed-loop stimulation, and next-day verification. This method utilizes wearable EEG devices to collect EEG signals in real time and uses a built-in algorithm engine for online sleep staging and slow-wave sleep (SWS) steady-state discrimination. When all preset trigger conditions are met, the system automatically triggers multimodal sensory cues (such as smells and sounds) associated with the daytime learning content. During stimulation, the system continuously monitors EEG to detect micro-awakening events; once detected, stimulation is immediately interrupted, and subsequent stimulation parameters are adaptively adjusted based on feedback. The entire process data is recorded in a structured manner and can generate a comprehensive report including the learning, stimulation, and verification stages.

[0009] The technical solution adopted in this invention is as follows: This invention provides a closed-loop directional memory reactivation method, comprising the following steps: During the daytime learning phase, a binding relationship is established between learning materials and at least one sensory stimulus cue, and a learning event record is generated; During the nighttime sleep stage, wearable EEG devices are used to collect the user's brain signals in real time. Based on the aforementioned EEG signals, sleep stages are determined in real time to detect whether the user has entered the target sleep stage; When the homeostatic condition of the user being in the target sleep stage is detected, sensory cues bound to the learning material are triggered and output; During the output of sensory cues, the EEG signals are continuously monitored to determine whether micro-arousal events are present. If they are present, the output of the sensory cues is interrupted. The next day, during the verification phase, the memory effect related to the learning material is tested, and a report is generated that includes learning event records, nighttime stimulus records, and memory effect test results.

[0010] Preferably, the target sleep stage is slow-wave sleep (SWS) stage of non-rapid eye movement sleep.

[0011] Furthermore, the determination of the steady-state conditions of the slow-wave sleep stage is based on a comprehensive assessment of multiple physiological indicators. These assessment conditions may include at least one of the following: the proportion of delta wave power in the EEG signal exceeds a preset power threshold; the level of electromyography signal or motor artifacts is lower than a preset artifact threshold; the time interval since the last sensory stimulus cue output is greater than or equal to a preset reentry period; and no micro-arousal events are detected within the recent safety window.

[0012] Furthermore, to achieve more precise neural modulation, the triggering step may also include an optional phase-locking mechanism. Specifically, after the steady-state condition is met, phase detection is performed on the delta band of the EEG signal, and sensory stimulus cues are only finally output when the delta band is detected to be within a preset phase interval (such as the rising phase of a slow wave). In a preferred embodiment, this phase detection is achieved by performing a Hilbert transform on the delta band signal to extract the instantaneous phase.

[0013] Furthermore, the sensory cues may include one or more of olfactory and auditory cues. When olfactory and auditory cues are used simultaneously, the system can synchronously control the olfactory output device and the audio playback device to output in coordination under the same triggering condition.

[0014] Furthermore, the present invention includes a safety and adaptive mechanism. After stimulation is interrupted due to a micro-arousal event, the system can adaptively adjust the output parameters of the next sensory stimulus cue, such as reducing the stimulus intensity, shortening the stimulus duration, or extending the return period, to adapt to individual differences among users.

[0015] Furthermore, the reports generated by this invention can be exported not only as visual document formats (such as PDF) for clinical review, but also as structured data formats (such as CSV or Excel) to support subsequent scientific research statistical analysis.

[0016] Accordingly, the present invention also provides a closed-loop directional memory reactivation system, the system comprising: Wearable EEG acquisition subsystem is used to acquire users' EEG signals in real time; The stimulus execution subsystem includes an olfactory output device and / or an audio playback device for outputting sensory stimulus cues; A processing terminal, such as a smartphone, tablet, or edge computing device, is equipped with a memory and a processor. The memory contains a computer program, which, when executed by the processor, can perform all or part of the steps of the aforementioned method.

[0017] In a preferred embodiment, the processing terminal is a mobile terminal on which a dedicated application (APP) runs. The APP integrates a daytime learning module, a nighttime sleep module, and a next-day verification module to guide the user through the entire TMR process.

[0018] In addition, this system may include a back-end management platform for centralized management of users, learning material libraries, sensory cue libraries, equipment, and all process data. This platform supports private deployment within the local area network of hospitals or research institutions to meet data security and compliance requirements.

[0019] The beneficial effects of this invention are as follows: This invention establishes a complete chain from "daytime learning - nighttime stimulation - next-day verification," achieving a strong system-level binding and automated association between learning materials and stimulus cues. This significantly improves the standardization level and experimental reproducibility of the TMR scheme. Based on a real-time SWS steady-state detection algorithm using wearable EEG, combined with a micro-awakening immediate interruption mechanism, it ensures that stimulation occurs at the optimal time and within a safe window, significantly reducing the risk of false triggering and sleep disturbance. It also enables the synchronous triggering and closed-loop control of multimodal cues such as olfaction and hearing, and quantifies and records the stimulus dosage, providing a standardized stimulus paradigm and a reliable data foundation for research.

[0020] This invention automatically generates complete data assets including raw EEG data, event-level structured logs, and comprehensive reports, facilitating research review, statistical analysis, and multi-center data sharing, and meeting the compliance requirements of clinical and research settings. The built-in adaptive parameter adjustment strategy and multiple safety limiting mechanisms can adjust the stimulation protocol according to the user's individual response, improving the effectiveness, tolerability, and safety of individualized interventions. Attached Figure Description

[0021] Figure 1 This is a block diagram of the overall architecture of the closed-loop directional memory reactivation system in this embodiment of the invention.

[0022] Figure 2 This is a flowchart illustrating the daytime learning and clue binding method in an embodiment of the present invention.

[0023] Figure 3 This is a flowchart of the nighttime closed-loop stimulation method in an embodiment of the present invention.

[0024] Figure 4 This is a simplified logic block diagram of the online phasing and slow wave detection algorithm in this embodiment of the invention.

[0025] Figure 5 This is a flowchart illustrating the next-day verification and report generation method in an embodiment of the present invention. Detailed Implementation

[0026] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0027] It should be understood that, and also noted, in the embodiments, the functions / actions may appear in a different order than those shown in the figures. For example, depending on the functions / actions involved, they may actually be performed substantially concurrently, or sometimes the two figures shown consecutively may be performed in reverse order.

[0028] Reference Figure 1 The present invention provides a closed-loop directional memory reactivation system, the overall architecture of which includes a wearable EEG acquisition subsystem 101, a stimulus execution subsystem 102, a processing terminal 103, a background management platform 104, and an algorithm engine 105.

[0029] The wearable EEG acquisition subsystem 101 is typically a lightweight headband or forehead patch with at least one built-in EEG acquisition electrode for continuously acquiring the user's EEG signals at night. The subsystem transmits the data to the processing terminal 103 in real time via low-power wireless technologies such as Bluetooth. Its sampling rate, bandwidth, and input dynamic range meet the needs of sleep staging and slow wave feature extraction, and it has electrode detachment detection and power / connection status monitoring.

[0030] The stimulus execution subsystem 102 includes a multi-channel programmable olfactory output device and an audio playback device. The olfactory device can provide different odors by replacing pluggable odor sacs and can precisely control the width, frequency, and total dose of odor pulses. The audio playback device can be a speaker of the processing terminal 103 or an external speaker.

[0031] The processing terminal 103 can be a mobile device such as a smartphone or tablet. It runs a core TMR application (APP), which embeds an algorithm engine 105 and provides a user interface. The APP also integrates functional modules such as a daytime learning module, a nighttime sleep module, and a next-day verification module. Specifically, the daytime learning module establishes a binding relationship between learning materials and sensory cues; the nighttime sleep module receives the EEG signals and performs real-time sleep staging, homeostasis detection, trigger stimulation, and micro-awakening monitoring and interruption; and the next-day verification module performs memory performance tests and generates reports.

[0032] The 104 backend management platform is deployed on a server and supports web access. It is used by researchers or medical staff to centrally manage users, learning materials, sensory cues, equipment, and end-to-end data. The platform supports private deployment within the private cloud or local area network of hospitals or research institutions.

[0033] The algorithm engine 105 runs on the processing terminal 103 and is responsible for processing EEG signals in real time and making decisions.

[0034] The following is combined Figures 2 to 5 The execution flow of the method of the present invention will be described in detail.

[0035] 1. Daytime learning and clue binding method (refer to) Figure 2 ) Step S201: Researchers design the learning plan at the top level on the backend management platform 104, including setting up a learning material library (such as a vocabulary list, spatial location map, etc.), configuring an odor / sound cue library (managing odor sacs and audio files), and establishing detailed learning plans for subjects or subject groups (e.g., for 4 weeks, one new material per day, and automatically generating a review queue), and clarifying the pairing relationship between learning content and sensory cues.

[0036] Step S202: The subject logs into the APP on the processing terminal 103 and selects the learning material to be studied that day from the learning plan (such as a vocabulary list, spatial location task, procedural key sequence, etc.). Based on the learning plan, the system automatically loads pre-bound sensory cues and, while presenting the learning material, simultaneously outputs corresponding sounds and / or smells through the stimulus execution subsystem 102. For example, when memorizing the word "Rose," related audio is played simultaneously and the scent of roses is released.

[0037] Step S203: The system records learning duration, operational compliance, and other indicators in real time, ultimately generating a "Learning Event Record" containing all key information, including user ID, timestamp, learning material ID, cue ID (scent channel, audio file), presentation parameters, etc. This record is stored locally and synchronized to the backend management platform 104.

[0038] 2. Nighttime closed-loop stimulation method (refer to...) Figure 3 ) Step S301: At night, the user puts on the wearable EEG acquisition subsystem 101 and activates "Nighttime Sleep Mode" in the APP. The system performs a self-check to confirm device connection, electrode contact quality, and battery status.

[0039] Step S302: Algorithm engine 105 begins to receive and process the EEG signal stream from wearable EEG acquisition subsystem 101 in real time. The processing flow is as follows: Figure 4 ,include: Step S401, Preprocessing: The input raw EEG signal undergoes a series of denoising and purification processes, specifically bandpass filtering, power line notch filtering, and artifact suppression. Bandpass filtering uses a bidirectional IIR or FIR filter with a passband set to 0.5-40Hz to eliminate DC drift and high-frequency EMG noise. Power line notch filtering targets 50Hz or 60Hz power line interference. Artifact suppression employs rule-based methods, such as marking segments with amplitudes exceeding ±100μV as artifacts and removing them; or using more advanced Independent Component Analysis (ICA) methods to separate and remove artifact components such as eye movements and EMG.

[0040] Step S402, Feature Extraction: Within a sliding short time window (e.g., 4 seconds, 50% overlap), the preprocessed signal is analyzed. Specifically, Fast Fourier Transform (FFT) or Wavelet Transform is used to calculate the absolute and relative power of each classical frequency band (δ: 0.5-4Hz, θ: 4-8Hz, α: 8-12Hz, β: 12-30Hz).

[0041] Step S403, Sleep Stage Determination: Input the extracted feature vector into a pre-trained lightweight machine learning model, such as the TinySleepNet model based on convolutional neural network (CNN) and long short-term memory network (LSTM), and output the sleep stage results of the current time window in real time, such as: wakefulness stage, N1 stage, N2 stage, N3 stage (SWS), REM stage.

[0042] Step S404, SWS Steady-State Gating Judgment: When the staging result is SWS (N3 stage), the system enters the steady-state gating judgment. Triggering a stimulus is only allowed if all of the following conditions are met simultaneously: δ power condition: When the relative power of the δ band (0.5-4Hz) exceeds a preset threshold (e.g., >50%), it indicates that the person has entered deep slow wave sleep.

[0043] Artifact level condition: The signal artifact level is below a preset threshold to ensure signal quality.

[0044] Return period condition: The time interval since the end of the last stimulus is greater than or equal to a preset return period T. refrac (e.g., ≥180 seconds) to prevent overly intense stimulation.

[0045] Recent Awakening Condition: No micro-awakening events were detected within the most recent safe window period (e.g., 30 seconds).

[0046] Step S405, Triggering command: If all conditions are met, the algorithm engine 105 generates a triggering command and sends it to the stimulus execution subsystem 102 via Bluetooth.

[0047] Step S406, Phase Detection: In a preferred embodiment, to maximize the TMR effect, the system performs precise phase locking after satisfying the steady-state conditions.

[0048] Perform a Hilbert transform on the delta-band signal to calculate its instantaneous phase.

[0049] Determine if the current phase is in the rising phase of a slow wave (e.g., between -π / 2 and π / 2). Only when the phase matches will a trigger signal be issued.

[0050] Step S407, Continuous Monitoring and Interruption: After the stimulus begins, the algorithm engine enters a high-alert monitoring mode.

[0051] Micro-arousal detection: Real-time monitoring of alpha wave or high beta wave power. If the power suddenly and significantly increases within a short period of time (e.g., 1-3 seconds) (e.g., exceeding 3 standard deviations of the recent baseline mean), it is identified as a micro-arousal event.

[0052] Immediate interruption: Once micro-arousal is detected, the system immediately sends a STOP command to interrupt all stimulus output and records the interruption event and reason code (e.g., reason code="arousal detected").

[0053] Step S408, Adaptive Adjustment: If an interruption occurs shortly after a stimulus (e.g., 10 seconds), the system will trigger an adaptive adjustment strategy. For example, it may automatically reduce the volume envelope by 10% (volume = volume * 0.9) or extend the return period by 20% (T) for the user in subsequent stimuli. refrac =T refrac *1.2). Furthermore, the adaptive adjustment strategy of this invention also includes a long-term learning mechanism across nights. For example, the system records and analyzes stimulus-response data for multiple consecutive nights. If it detects that a user experiences early microarousing after stimulation for three consecutive nights, the system will trigger a stronger adjustment, such as fixing the user's volume limit by one level (e.g., -5dB) and extending the pulse width interval of olfactory stimulation (e.g., +2 seconds). This adjustment rule can be based on a preset rule engine or can use an individual response model, such as dynamically adjusting the stimulation parameters by weighted averaging the microarousing response rates of the past three nights, thereby achieving a deeper level of individualized calibration.

[0054] Step S303: Throughout the night, the system will continuously execute the above closed-loop process, while ensuring the total number of stimulations (N) is maintained. max The total dose should not exceed the preset safety limit (e.g., a maximum of 40 times per night). When the system determines that the user has entered REM or a waking state, all stimulation triggering will be paused.

[0055] Step S304: After the entire night session, the system generates a detailed "Stimulus Event Log" (JSON or CSV format), recording the timestamp, cue ID, stimulus parameters, online stage tag, interruption flag, and reason for each trigger. The system also saves the original EEG file (EDF format) and uploads it to the backend management platform 104.

[0056] 3. Next-day verification and report generation method (refer to...) Figure 5 ) Step S501: The next day, the user activates the "Next Day Verification Mode" in the APP.

[0057] Step S502: The system automatically generates corresponding memory test questions (such as word recall, image recognition, spatial location, etc.) based on the learning materials from the previous day.

[0058] Step S503: After the user completes the test, the system automatically calculates their memory performance indicators (such as accuracy, reaction time, spatial error, etc.).

[0059] Step S504: The system integrates three parts of data—daytime learning records, nighttime stimulation logs (including total stimulation dose, SWS percentage, etc.), and the next day's validation results—to automatically generate a "learning-stimulation-validation triple report." This report can be exported not only as a visual PDF for clinical review and presentation, but also as a structured CSV or Excel file, facilitating subsequent research statistics and data analysis by researchers.

[0060] 4. Clinical / Research Applications This section uses a clinical study aimed at improving declarative memory in patients with mild cognitive impairment (MCI) through the system of this invention as an example to illustrate the specific application of the invention.

[0061] (A) Experimental Design Study type: A randomized, double-blind, sham-controlled clinical study.

[0062] Sample size: Sixty MCI patients who met the diagnostic criteria were recruited and randomly assigned in a 1:1 ratio to the TMR intervention group (N=30) and the sham stimulation control group (N=30) using a computer-generated random sequence.

[0063] Inclusion criteria: (1) Age 60-80 years; (2) Meeting the core clinical diagnostic criteria of MCI, with a Montreal Cognitive Assessment (MoCA) score between 18-26; (3) Having stable nighttime sleep habits; (4) Having basically normal hearing and olfactory function.

[0064] Exclusion criteria: (1) having Alzheimer's disease or other types of dementia; (2) having a severe sleep disorder (such as severe sleep apnea syndrome); (3) currently taking medications that affect the central nervous system or sleep structure; (4) being allergic to irritating odors.

[0065] Blinding setup: Subjects, researchers responsible for memory assessment, and data analysts are all blinded to the group assignments. In this invention's system, the odor sacs and sound files are configured by a non-blinded research coordinator according to a random sequence, making them visually indistinguishable.

[0066] Control group setup: During the nighttime SWS steady state, the control group received neutral sound (such as weak white noise) and odor (such as very low concentration of diluted ethanol, which is almost odorless) stimuli unrelated to the daytime learning materials. The timing, frequency and duration of the stimulation were completely matched with those of the intervention group.

[0067] Primary endpoint: Change in accuracy on the free recall test of vocabulary 24 hours after intervention, compared with baseline.

[0068] Secondary endpoints: vocabulary recognition test accuracy, total sleep time, percentage of sleep-wake cycles (SWS), number of nighttime awakenings, and subject subjective experience questionnaires.

[0069] (B) Population characteristics The baseline data of the 60 enrolled participants were comparable, with a mean age of 68.5 ± 5.2 years, 55% being female, and a mean MoCA score of 22.1 ± 2.8. There were no statistically significant differences between the two groups in terms of age, sex, years of education, and baseline memory level (p > 0.05).

[0070] (C) Intervention parameters Learning materials: A list of 40 standardized, low-frequency two-character Chinese words.

[0071] Stimulating materials and channels: TMR intervention group: During the daytime, while learning 40 words, a rose scent (output by channel 1 of stimulus execution subsystem 102) and a specific, gentle piano piece (10 seconds) were presented simultaneously. At night, during SWS steady state, the system triggered channel 1 to output a rose scent pulse and played the same piano piece.

[0072] Sham stimulus control group: No additional sensory cues were presented when learning 40 words during the day. At night, in SWS steady state, the system triggered channel 2 to output a pulse of diluted ethanol odor and played a white noise clip.

[0073] Nighttime stimulation parameters: Maximum number of stimulations (N) max ): 50 times / night.

[0074] Turnaround period (T) refrac ): Minimum 120 seconds.

[0075] Phase-locking strategy: Enable slow-wave rise phase-locking triggering, with the trigger window set to the interval between -π / 2 and π / 2 of the delta wave phase.

[0076] Adaptive strategy: Enable micro-wake-up interrupt and parameter adaptive adjustment functions.

[0077] (D) Summary of Results Primary endpoint: After the intervention, the TMR intervention group showed an average improvement of 25.4% in free recall accuracy from baseline (SD=8.1%), while the sham stimulation control group showed an average improvement of 5.2% (SD=3.5%). The difference between the two groups was highly statistically significant (t-test, p<0.001), indicating that the TMR intervention of this invention can significantly improve declarative memory in MCI patients.

[0078] Safety: Analysis of the sleep structure data recorded by the system showed no statistically significant differences in total sleep time, SWS percentage, and number of nighttime awakenings between the two groups before and after the intervention (p>0.05), indicating that the closed-loop stimulation protocol of this invention did not have a negative impact on the patients' macroscopic sleep structure. No subjects reported stimulation-related discomfort or adverse events.

[0079] Tolerance: The completion rate of the entire intervention period was as high as 95%. The subjects generally reported that the device was comfortable to wear and the nighttime stimulation did not cause any discomfort, indicating good tolerance.

[0080] 5. Security and Adaptive Strategies To ensure the security and effectiveness of this invention in various application scenarios, the system integrates a comprehensive set of security and adaptive strategies, which can be summarized in the following aspects: (1) Rapid detection and immediate interruption of micro-awakening: As mentioned above, this is the core real-time security mechanism to ensure the user's sleep quality. The algorithm engine performs highly sensitive monitoring of EEG signals during stimulation to ensure that any stimulation that may interfere with sleep can be stopped immediately.

[0081] (2) Individualized calibration and adaptation: The system not only makes immediate adjustments after a single stimulus, but also has the ability to learn across nights. By analyzing the user's response to the stimulus (arousal threshold) on the first night or the first few nights, the system can adaptively calibrate the initial stimulus parameters (such as volume and odor pulse width) and minimum stimulus interval for each user, and dynamically optimize based on long-term feedback.

[0082] (3) Dual Software and Hardware Limitations: The system has built-in, insurmountable safety boundaries, including software-level parameter limitations and hardware-level protection. Specifically, this includes the maximum number of stimulations per night (N). max ), the maximum total dose for each sensory cue (Dose-total) max ), and the shortest turnaround period (T) refrac All of them have hard upper limits to prevent overstimulation due to algorithm or configuration errors.

[0083] (4) Device Status Watchdog: The system continuously monitors the status of the hardware device. This includes the electrode contact quality of the EEG acquisition subsystem, the device battery level, and the stability of the Bluetooth connection with the stimulation execution subsystem. Once an anomaly such as a disconnection, low battery, or poor signal quality is detected, the system will immediately pause closed-loop stimulation and enter a safe shutdown mode until the status returns to normal.

[0084] (5) Abnormal events and information prompts: When the system detects the above-mentioned abnormal events (such as electrode detachment, poor signal quality, etc.), it will issue clear information prompts or alarms to users or researchers through the APP interface to ensure that the problem can be detected and dealt with in a timely manner.

[0085] Through the above multi-level and comprehensive strategy combination, this invention constructs an intelligent and reliable closed-loop intervention system, ensuring its safety and stability in clinical research and future home applications.

[0086] In summary, the system and method disclosed in this invention, through their integrated design, precise closed-loop control, standardized data management, and individualized security strategies, effectively solve many pain points of existing TMR technology in practical applications, providing a complete and reliable technical solution for targeted memory reactivation technology to move from the laboratory to clinical applications and home scenarios.

[0087] This invention is not limited to the above-described optional embodiments. Anyone can derive other various forms of products under the guidance of this invention. However, regardless of any changes made in their shape or structure, any technical solution that falls within the scope of the claims of this invention shall be protected by this invention.

Claims

1. A closed-loop directional memory reactivation method, characterized in that, Includes the following steps: During the daytime learning phase, a binding relationship is established between learning materials and at least one sensory stimulus cue, and a learning event record is generated; During the nighttime sleep stage, wearable EEG devices are used to collect the user's brain signals in real time. Based on the aforementioned EEG signals, sleep stages are determined in real time to detect whether the user has entered the target sleep stage; When the homeostatic condition of the user being in the target sleep stage is detected, sensory cues bound to the learning material are triggered and output; During the output of sensory cues, the EEG signals are continuously monitored to determine whether micro-arousal events are present. If they are present, the output of the sensory cues is interrupted. The next day, during the verification phase, the memory effect related to the learning material is tested, and a report is generated that includes learning event records, nighttime stimulus records, and memory effect test results.

2. The method according to claim 1, characterized in that, The target sleep stage is the slow-wave sleep stage of non-rapid eye movement (NREM) sleep.

3. The method according to claim 2, characterized in that, The determination of the homeostatic conditions of the slow-wave sleep stage is based on an assessment of at least one of the following conditions: The power ratio of delta waves in the EEG signal exceeds the preset power threshold; The electromyographic signal or motor artifact level is below the preset artifact threshold; The time interval between the last sensory stimulus output and the last output is greater than or equal to the preset return period; The time interval since the last detected micro-awakening event is greater than the preset safety interval.

4. The method according to claim 1 or 2, characterized in that, The triggering step also includes: After the steady-state condition is met, phase detection is performed on the delta band of the EEG signal; The sensory stimulus cue is only output when the delta band is detected to be in the slow rising phase.

5. The method according to claim 4, characterized in that, The phase detection is achieved by performing a Hilbert transform on the electroencephalogram (EEG) signal to extract the instantaneous phase.

6. The method according to claim 1, characterized in that, The sensory cues include at least one of olfactory cues and auditory cues; When the sensory stimulus cues include both olfactory and auditory cues, the triggering step includes: under the same triggering conditions, synchronously controlling the olfactory output device and the audio playback device to output olfactory and auditory cues that are bound to the learning material.

7. The method according to claim 1, characterized in that, Following the step of interrupting the output of sensory cues, the method further includes: Based on the micro-awakening event, the stimulation parameters of the next sensory stimulus cue output are adaptively adjusted, including stimulus intensity, stimulus duration, or return period.

8. The method according to claim 1, characterized in that, The method further includes: Throughout the nighttime sleep phase, the total number of sensory stimulus outputs and / or the total output dose are limited to not exceed a preset upper limit.

9. A closed-loop directional memory reactivation system, characterized in that, include: Wearable EEG acquisition subsystem is used to acquire users' EEG signals in real time; A stimulus execution subsystem is used to output at least one sensory stimulus cue according to control instructions; A processing terminal is configured with a memory and a processor, wherein the memory stores a computer program, and the processor, when running the computer program, implements the method as described in any one of claims 1 to 8.

10. The system according to claim 9, characterized in that, The processing terminal is a mobile terminal, on which an application runs, and the application integrates the following functional modules: The daytime learning module is used to establish a binding relationship between learning materials and sensory stimulus cues; The nighttime sleep module is used to receive the EEG signals and perform real-time sleep stage determination, homeostasis condition detection, trigger stimulation, and micro-awakening monitoring and interruption. The next-day verification module is used to perform memory effect tests and generate reports.

Images