A mild cognitive impairment brain rehabilitation training device and method

Abstract

The present application relates to the field of rehabilitation training equipment, in particular to a mild cognitive impairment brain rehabilitation training device and method, the device comprises: an electroencephalogram acquisition module, a transcranial electrical stimulation module, an immersive scene presentation module, an olfactory release module, an eye tracking module, a control unit; the control unit is used for analyzing the resting state electroencephalogram of the user, calculating the cross-frequency phase-amplitude coupling strength, determining the stimulation target and frequency; real-time receiving the data of the electroencephalogram acquisition module and the eye tracking module, jointly determining whether the user is in a memory coding difficult state based on the real-time cross-frequency phase-amplitude coupling index and the eye movement index; when it is determined that the memory coding difficult state, according to the current electroencephalogram instantaneous phase, trigger the transcranial electrical stimulation module to output the phase-locked alternating current stimulation pulse train to the target brain area. The present application can realize real-time perception of the brain function state in the training process of the patient, realize dynamic self-adaptive adjustment of the cognitive training strategy, and improve the effectiveness of MCI cognitive rehabilitation.

Description

Technical Field

[0001] This invention relates to the field of rehabilitation training equipment, specifically to a brain rehabilitation training device and method for mild cognitive impairment. Background Technology

[0002] Mild cognitive impairment (MCI) is a transitional stage between normal aging and dementia, characterized by age- and education-level-disproportionate decline in memory, attention, or executive function, but without significantly affecting daily living abilities. This population represents a critical window for early intervention and delaying cognitive decline. Currently, non-pharmacological interventions primarily include computerized cognitive training, virtual reality (VR) scenario training, transcranial electrical stimulation (TES) neuromodulation, and multi-sensory stimulation.

[0003] Computerized cognitive training involves repetitive tasks such as memorization and calculation on tablets or computers. VR training provides more ecologically valid everyday scenarios, allowing patients to perform simulated tasks such as spatial navigation, shopping, and meal preparation in a virtual environment to enhance episodic memory and executive function. In terms of neuromodulation, transcranial alternating current stimulation (tACS) modulates the phase and amplitude of neuronal group oscillations by applying weak currents synchronized with endogenous brain rhythms, and has shown potential to enhance working and episodic memory in preliminary studies. Meanwhile, olfaction, due to its direct neural projections to the hippocampus-entorhinal cortex, is considered a sensory channel that can strongly trigger episodic memory, and some olfactory training products have been used for cognitive stimulation.

[0004] However, most existing cognitive training systems adjust the difficulty based solely on behavioral indicators such as the error rate and reaction time of task completion, failing to acquire real-time information about the patient's internal brain function. When patients are in a state of inattention or memory encoding failure, the system cannot identify and provide targeted intervention, resulting in a significant amount of training cycles being wasted inefficiently or even ineffectively.

[0005] Conventional tACS interventions are mostly performed in a resting state, using fixed stimulation frequencies, phases, and durations, completely disregarding the real-time changes in the patient's brain rhythms during active cognitive tasks. This open-loop stimulation may mismatch with the natural rhythms of the cognitive process, or even interfere with normal brain function, lacking the ability to provide on-demand synchronous intervention, thus limiting both safety and efficacy.

[0006] Given the shortcomings of existing technologies, there is an urgent need for a novel brain rehabilitation training device and method for mild cognitive impairment that can sense the patient's brain function status in real time during the training process, realize dynamic adaptive adjustment of cognitive training strategies, and improve the accuracy and effectiveness of MCI cognitive rehabilitation. Summary of the Invention

[0007] To address the aforementioned issues, this invention provides a brain rehabilitation training device and method for mild cognitive impairment, which is used to perceive the brain function status of patients in real time during the training process, realize dynamic adaptive adjustment of cognitive training strategies, improve the accuracy and effectiveness of MCI cognitive rehabilitation, and achieve efficient and personalized rehabilitation intervention for the cognitive function of patients with mild cognitive impairment.

[0008] To achieve the above objectives, the technical solution of the present invention is as follows: A brain rehabilitation training device for mild cognitive impairment, comprising:

[0009] The EEG acquisition module is used to acquire EEG signals from the user's frontal-temporal lobe region in real time.

[0010] The transcranial electrical stimulation module is used to deliver alternating current stimulation to the user's target brain regions;

[0011] The immersive scene presentation module is used to present users with three-dimensional daily life scenes containing memory tasks and spatialized audio;

[0012] The olfactory release module is used to release a scent that matches the scene into the user's nasal cavity when the immersive scene presentation module presents the scene;

[0013] An eye-tracking module is used to capture the user's gaze point position and pupil diameter data in real time;

[0014] The control unit, which communicates with the above modules, is used to execute the following closed-loop adaptive rehabilitation training control:

[0015] During the individualized parameter calibration phase, the user's resting-state EEG is analyzed, the individual peak frequency of the frequency band related to the memory task is extracted, the cross-frequency phase-amplitude coupling strength is calculated, and the stimulation target and frequency are determined.

[0016] During the training phase, the immersive scenario presentation module and the olfactory release module jointly present three-dimensional daily life scenes.

[0017] It receives data from the EEG acquisition module and the eye-tracking module in real time, and determines whether the user is in a state of memory encoding difficulty based on the real-time cross-frequency phase-amplitude coupling index and eye-tracking index.

[0018] When a state of memory encoding difficulty is identified, the transcranial electrical stimulation module is triggered to output a phase-locked AC stimulation pulse train to the target brain region based on the current instantaneous phase of the EEG.

[0019] The technical principles of the above solution are as follows:

[0020] Successful encoding of episodic memory highly depends on the phase-amplitude coupling (PAC) of theta rhythm to gamma oscillation in the medial temporal-prefrontal network. This cross-frequency coupling provides a precise time window for synaptic plasticity. This invention utilizes real-time EEG monitoring of this core neurophysiological indicator: when a user engages in multisensory (visual, auditory, olfactory) episodic exploration, insufficient endogenous theta-gamma coupling indicates that the neural circuitry for memory encoding is not effectively activated. At this critical juncture, the control unit directly enhances the theta rhythm synchronicity of the target network by applying alternating current stimulation highly synchronized with the endogenous theta wave phase, reorganizing and strengthening gamma nesting, thereby artificially inducing a highly efficient memory encoding state. Simultaneously, olfactory input is projected directly to the hippocampus via the olfactory bulb, deeply fused with VR spatial information and task instructions, activating the generation of multimodal episodic memory traces. Eye-tracking cues confirm the user's explicit attentional focus, ensuring the correspondence between implicit brain state decoding and explicit behavior.

[0021] The above approach has the following beneficial effects:

[0022] 1. This solution directly decodes the user's internal memory encoding neural state by real-time acquisition of frontotemporal EEG signals and calculation of cross-frequency phase-amplitude coupling strength. This overcomes the limitations of traditional systems that rely solely on behavioral accuracy, externality and lag in reaction time, and ensures that intervention decisions are truly based on objective indicators of neural activity.

[0023] 2. Unlike conventional open-loop, fixed-frequency tACS intervention, this approach triggers stimulation only when a memory encoding difficulty state is detected, and strictly locks the output according to the current EEG instantaneous phase to ensure that the applied electric field is superimposed in phase with the endogenous rhythm, reducing the interference of mismatch stimulation on normal brain rhythms and improving the targeting efficacy and biosafety of neuromodulation.

[0024] 3. This solution organically integrates VR scenario presentation, spatialized audio, and scene-related olfactory stimulation to construct a highly ecologically valid daily life scenario. Compared with single-channel computerized training, it can better activate the multimodal memory encoding network of the hippocampus-entorhinal cortex. The introduction of olfactory stimulation strengthens the contextual dependence of scenario memory and improves the transfer effect of training content.

[0025] 4. This solution introduces eye tracking as behavioral evidence of overt attention, forming a joint judgment system that complements subjective and objective factors with EEG neural indicators. This effectively reduces the probability of misjudgment caused by artifacts or random fluctuations in single neural signals, making the system's triggering decisions more robust and reliable.

[0026] Furthermore, when the control unit determines a difficult state for memory encoding, it performs the following operations:

[0027] A real-time sliding window is used to calculate the phase-amplitude coupling value between the theta band phase and the gamma band amplitude in the frontotemporal region; simultaneously, it is determined whether the gaze point is fixed on a task-related target object, and changes in pupil diameter are monitored; a state of memory encoding difficulty is identified when the following conditions are met simultaneously:

[0028] The phase-amplitude coupling value is lower than the preset individualization threshold;

[0029] The gaze point remains continuously on the same target object for more than a first preset duration;

[0030] The pupil diameter increases by more than a first preset ratio relative to the baseline state, or decreases by more than a second preset ratio.

[0031] Beneficial Effects: Existing single EEG indicators are susceptible to false positives due to artifacts from electromyography (EMG), electrooculography (EOG), or transient psychological fluctuations. This invention logically constrains implicit EEG PAC coupling attenuation (neural level), explicit fixation anchoring (visual attention level), and extreme changes in pupil diameter (autonomic arousal / fatigue level), ensuring that the system only triggers intervention when the neural coding circuit is indeed dysfunctional and the behavior and physiological state synchronously confirm this dysfunction. This triple verification mechanism significantly improves the specificity of determining memory coding difficulties and reduces ineffective or excessive stimulation of normal cognitive fluctuations.

[0032] Furthermore, the control unit is also used to control the immersive scene presentation module to visually highlight the target object currently being looked at by the user, or to reduce the complexity of the scene background, while triggering transcranial electrical stimulation.

[0033] Beneficial effects: Transcranial electrical stimulation amplifies endogenous theta-gamma coupling at the neural level, representing a "bottom-up" neural gain; while visual salience and background simplification lower the target discrimination threshold and reduce competition from irrelevant information at the perceptual level, representing a "top-down" perceptual noise reduction. The precise superposition of these two at the same time creates a synergistic amplification of the neural-perceptual dual pathways, enhancing the intervention effect.

[0034] Furthermore, the olfactory release module includes multiple odor generators, each storing different types of odor cartridges; the control unit selectively activates one or more odor generators based on the current scene type presented by the immersive scene presentation module, so that the odor matches the scene content.

[0035] Beneficial effects: By strictly semantically matching the released odor with the current VR scene content (e.g., a supermarket scene with the smell of coffee, a park scene with the smell of grass), the hippocampus is ensured to bind "visual scene—spatial location—task objective—olfactory tag" into a highly integrated episodic memory trace during the encoding stage. In subsequent recall, this matched odor can serve as a powerful retrieval cue to reactivate the entire episodic network, improving the episodic memory retrieval ability of MCI patients.

[0036] Furthermore, the target brain regions include the left dorsolateral prefrontal cortex or the entorhinal cortex projection area; the individualized parameter calibration stage includes comparing the resting-state theta-gamma phase-amplitude coupling strength of different EEG channels, and selecting the channel with the lowest coupling strength as the stimulation target.

[0037] Beneficial effects: Information flow in neural networks is constrained by the transmission efficiency of the weakest node. This scheme strengthens the rhythmic synchronization of this node, reduces its information transmission impedance, and enables the gamma oscillations generated by the hippocampus to be transmitted more effectively to the prefrontal cortex. The theta instructions from the prefrontal cortex can also be transmitted more smoothly to the hippocampus, promoting efficient bidirectional information interaction throughout the memory network.

[0038] Furthermore, the AC stimulation frequency output by the transcranial electrical stimulation module is equal to the individual theta peak frequency determined during the individualized parameter calibration stage, and the stimulation intensity is adaptively adjusted within a preset range.

[0039] Beneficial effects: Precisely setting the stimulation frequency to an individualized theta peak frequency (rather than a fixed frequency such as 6Hz) aligns with the "resonance principle" of neural entrainment. When the applied electric field frequency is perfectly synchronized with the intrinsic rhythm, the neuronal population achieves maximum synchronization response with minimal external energy input.

[0040] Furthermore, when the transcranial electrical stimulation is triggered, the control unit extracts the instantaneous phase of the target channel theta band based on the latest EEG data at the moment of determining the difficult state, and calculates the target trigger phase by combining it with the inherent delay of the system, so that the AC stimulation output by the transcranial electrical stimulation module is locked with the target trigger phase.

[0041] Beneficial Effects: Physical delays (typically tens to hundreds of milliseconds) exist in all stages of EEG acquisition, filtering calculation, communication transmission, and stimulator response. If triggered directly by the "current real-time phase," the intrinsic rhythm has already undergone phase shift by the time the electrical stimulation reaches the target brain region, causing the actual superimposed phase to deviate from the expected one. This approach precisely measures the inherent delay of the system and reverse-calculates the target trigger phase, ensuring that the electric field actually acts on cortical neurons precisely aligned with the peak or trough of the intrinsic theta wave, achieving precise in-phase intervention at the cortical level.

[0042] Furthermore, the EEG acquisition module uses a 32-channel active electrode cap, with the electrode positions arranged according to the international 10-20 system.

[0043] Beneficial effects: The 32-channel layout in the frontal and temporal lobe regions (covering standard sites such as F3 / F4, F7 / F8, and T7 / T8) provides a sufficiently dense sampling grid, accurately distinguishing PAC differences between adjacent target points such as the left DLPFC and the entorhinal cortex projection area. The internationally recognized 10-20 system is a standardized spatial reference framework in the field of EEG, ensuring that EEG data collected from different patients and on different training days originate from anatomically consistent locations. This standardized foundation ensures the traceability and retest reliability of the PAC thresholds and target coordinates established during the individualized calibration phase, and also provides a unified input format for subsequent optimization of the judgment algorithm based on large sample data.

[0044] Furthermore, the control unit is also used to adaptively adjust the stimulation intensity or stimulation pulse duration output by the transcranial electrical stimulation module according to the degree of deviation between the phase-amplitude coupling value and the preset individualized threshold; the greater the deviation, the higher the stimulation intensity or the longer the stimulation pulse duration controlled by the control unit.

[0045] Beneficial effects: Fixing high intensity may lead to neuroadaptive inhibition or discomfort in patients with mild deviations, while fixing low intensity is ineffective for patients with severe deviations. This protocol, based on real-time dynamic deviation output, maintains a low-intensity comfort zone for patients with mild deviations and provides more effective interventions in a timely manner for patients with severe deviations, thus balancing safety, comfort, and effectiveness.

[0046] A brain rehabilitation training method for mild cognitive impairment includes the following steps:

[0047] Step A, Individualized parameter calibration: Collect the user's resting-state EEG, extract the individual peak frequency of the frontotemporal lobe region in the theta band, calculate the phase-amplitude coupling strength between the theta phase and the gamma amplitude, and determine the stimulation target, stimulation frequency and difficult state judgment threshold in this way.

[0048] Step B, Multi-sensory Contextual Memory Training: Using virtual reality, a three-dimensional daily life scene containing memory tasks is presented to the user, and a scent matching the scene content is released simultaneously while the scene is presented, allowing the user to freely observe and perform memory encoding.

[0049] Step C, Real-time State Analysis: Continuously collect the user's EEG and eye movement signals, calculate the theta-gamma phase-amplitude coupling value at the stimulation target point in real time, and monitor the fixation point position and pupil diameter. When the phase-amplitude coupling value is lower than the difficult state determination threshold, and the fixation point dwell time and pupil diameter change meet the preset conditions, it is determined to be a difficult state of memory encoding.

[0050] Step D, Closed-loop phase-synchronous electrical stimulation intervention: Once the memory encoding difficulty state is determined, the instantaneous phase of the current theta frequency band is extracted, and after delay compensation, the transcranial electrical stimulation module is triggered to output phase-locked AC electrical stimulation to the stimulation target at the individual peak frequency;

[0051] Step E, Multi-cycle Iteration and Evaluation: Repeat steps B to D. After completing a predetermined number of memory training tasks, collect resting-state EEG data again for evaluation and generate a training effect report. Attached Figure Description

[0052] Figure 1 This is a three-dimensional structural schematic diagram of the brain rehabilitation training device for mild cognitive impairment according to the present invention;

[0053] Figure 2 This is a schematic diagram of the partial internal structure of the anterior nasal support in the mild cognitive impairment brain rehabilitation training device of the present invention.

[0054] Figure 3 This is a schematic diagram of the larynx structure in the mild cognitive impairment brain rehabilitation training device of the present invention;

[0055] Figure 4 This is a flowchart of the brain rehabilitation training method for mild cognitive impairment according to the present invention.

[0056] The reference numerals in the accompanying drawings include: 1. Electrode helmet cap; 2. VR display; 3. Nose support; 4. Throat passage; 301. Odor port; 302. Tesla valve; 303. Air passage; 304. Slider; 305. Spring; 306. First air outlet; 307. Transfer channel; 308. Second air outlet; 309. Venting channel. Detailed Implementation

[0057] The technical solution 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0058] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0059] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0060] The following detailed description illustrates the specific implementation method:

[0061] Example 1:

[0062] As attached Figure 1-3 As shown: A brain rehabilitation training device for mild cognitive impairment, mainly comprising:

[0063] The EEG acquisition module is used to acquire EEG signals from the user's frontotemporal lobe region in real time. Specifically, the EEG acquisition module includes an electrode helmet cap 1, on the inner side of which is integrated a 32-channel active electrode cap. The electrodes within the 32-channel active electrode cap are arranged according to the international 10-20 system, focusing on covering the user's frontotemporal lobe region, including channels Fp1, Fp2, F3, F4, F7, F8, T7, T8, and Cz (existing technology, which will not be elaborated further). The reference electrode is a bilateral mastoid average reference, grounded in the forehead.

[0064] The transcranial electrical stimulation (TCS) module is used to deliver alternating current stimulation to the user's target brain regions. Specifically, the TCS module contains eight independently configurable stimulation electrodes, all distributed inside the electrode helmet cap 1. The stimulation electrodes are 2cm diameter Ag / AgCl cup-shaped electrodes filled with conductive gel. In this embodiment, F3 (left dorsolateral prefrontal cortex) and F4 (right dorsolateral prefrontal cortex) are selected as the main stimulation targets, and the reference electrode is placed 1cm behind Cz. The programmable current source can output sinusoidal alternating current with a frequency range of 4-8Hz and an intensity of 0-2mA, with an intensity step of 0.05mA, and is equipped with both hardware and software overcurrent protection.

[0065] An immersive scenario presentation module is used to present users with three-dimensional everyday life scenes containing memory tasks and spatialized audio; specifically, the immersive scenario presentation module includes a head-mounted VR display 2 and a controller (not shown in the figure), combined with an attached... Figure 1 As shown, the VR display 2 is integrated on the front of the electrode helmet 1, and bone conduction headphones are integrated inside the VR display 2.

[0066] The olfactory release module is used to release an odor matching the scene into the user's nasal cavity when the immersive scene presentation module presents the scene. Specifically, the olfactory release module includes a front nasal bracket 3, a silent fan, and several miniature solenoid valves. The front nasal bracket 3 is installed on the front side of the electrode helmet cap 1 (below the VR display 2, close to the user's nostrils, as shown in the attached image). Figure 1 As shown), the nasal support 3 integrates multiple scent generators, each containing a scent cartridge (such as coffee, lemon, floral, woody, or mint). For details, refer to the attached document. Figure 2 As shown, an air passage 303 is provided inside the nasal support 3. The air passage 303 is connected to an air venting channel 309 and several odor holes 301. The odor holes 301 are located at the top center of the nasal support 3 (corresponding to the user's nostrils). The air venting channel 309 is located at the bottom of the nasal support 3. A slider 304 is also slidably connected inside the air passage 303. A spring 305 is provided between the slider 304 and the inner wall of the air passage 303. The two ends of the spring 305 abut against the slider 304 and the inner wall of the air passage 303, respectively. The end of the air passage 303 away from the spring 305 is connected to the output end of a silent fan. A transfer mechanism is provided inside the slider 304. The transfer channel 307 is connected at one end to the ventilation channel 303 (near the output end of the silent fan). The transfer channel 307 is also connected to a first air outlet 306 and several second air outlets 308. When the slider 304 moves to a first preset value, each of the second air outlets 308 is connected to the odor hole 301. Preferably, each odor hole 301 is connected to a Tesla valve 302. The gas is discharged from the ventilation channel 303 to the odor hole 301 in the direction of high resistance of the Tesla valve 302. When the slider 304 moves to a second preset value, the first air outlet 306 is connected to the venting channel 309.

[0067] Combined with appendix Figure 3As shown, several branch channels are connected between the ventilation duct 303 and the output end of the silent fan. Each branch channel is connected to a throat 4. Each throat 4 corresponds to and is connected to a scent box. Miniature solenoid valves are installed on the connection path between the throat 4 and the scent box. When any miniature solenoid valve is opened, the airflow input by the silent fan passes through the throat 4. The airflow will generate negative pressure near the throat 4, introducing the scent in the corresponding scent box into the branch channel and into the ventilation duct 303 with the airflow. The airflow accumulates in the ventilation duct 303 and pushes the slider 304 to move against the elastic force of the spring 305 until the slider 304 moves to the first preset value. Then, each scent hole 301 is connected to the second air outlet 308. The airflow with scent escapes from the scent hole 301 after being depressurized by each Tesla valve 302. When it is necessary to stop the odor release or switch the odor, the original residual odor in the ventilation duct 303 needs to be vented. At this time, close the micro solenoid valve or switch to another micro solenoid valve, and increase the power of the silent fan to increase the air pressure in the ventilation duct 303 until the slider 304 moves to the second preset value. The residual odor in the ventilation duct 303 and its connecting path will be discharged from the venting channel 309. With the continuous output of the silent fan, the airflow in the ventilation duct 303 will be continuously renewed. Then the normal power of the silent fan will be restored to reduce the interference of residual odor.

[0068] The eye-tracking module is used to capture the user's gaze point position and pupil diameter data in real time. Specifically, the eye-tracking module uses a near-infrared eye-tracking camera, which is integrated into the VR display 2 and can provide data such as gaze point coordinates and pupil diameter.

[0069] The control unit, which communicates with the above modules, is used to execute the following closed-loop adaptive rehabilitation training control:

[0070] During the individualized parameter calibration phase, the user's resting-state EEG was analyzed to extract the individual peak frequencies of the relevant frequency bands for the memory task. The cross-frequency phase-amplitude coupling (PAC) strength was calculated to determine the stimulation target and frequency. Specifically, the user was in a comfortable sitting position, wearing an electrode helmet 1, a VR display 2, and other devices. First, 5 minutes of resting-state, eye-closed EEG data were collected. EEG signals from the frontotemporal lobe channels F3, F4, Fz, and T7 were extracted, and a 4-8 Hz frequency band was filtered using a 4th-order Butterworth bandpass filter. The Welch power spectral density estimation method (Hanning window, 2-second window length, 50% overlap) was used to calculate the power spectrum of each channel. The frequency point corresponding to the maximum power spectral density within the 4-8 Hz range was searched as the theta peak frequency of that channel. For example, if the peak frequency of the F3 channel is 6.5 Hz, this value is recorded as the individual's theta peak frequency. .

[0071] For a given F3 channel signal, calculate the theta frequency band (in terms of frequency band). The coupling strength between the instantaneous phase centered at a frequency (with a bandwidth of ±1 Hz) and the amplitude envelope of the gamma band (30-80 Hz). This embodiment uses the modulation index (MI) method proposed by Canolty et al., and the specific steps are as follows:

[0072] Sub-step 1.1: Using the raw EEG signals to... A bandpass filter centered at the center (bandwidth ±1Hz, 4th order Butterworth) is used to obtain the theta band signal. Then, a 30-80Hz bandpass filter is applied to the original signal to obtain the gamma band signal. .

[0073] Sub-step 1.2: For Perform a Hilbert transform, take the phase angle of the complex result, and obtain the instantaneous phase sequence. value range .

[0074] Sub-step 1.3: For Perform a Hilbert transform, take the amplitude of the complex result, and obtain the gamma amplitude envelope sequence. .

[0075] Sub-step 1.4: Divide the phase interval Divide into N=18 boxes (20° per box), for each box Collect all that satisfy The time t corresponding to the time when the box falls into the box Value, calculate the contents of the box The mean, denoted as .

[0076] Sub-step 1.5: Transfer each box... The values ​​form an 18-dimensional vector, and the normalized vector length of this vector is calculated as the modulation index (MI). The calculation formula is:

[0077]

[0078]

[0079] Where N=18 (total number of boxes). It is the base of the natural logarithm. It is the imaginary unit. For the first The center phase value (in radians) of each box. MI is the modulus of the complex value P, and the denominator is the average of all bin mean values. Theoretically, the MI value ranges from 0 to a positive real number, and a larger value indicates a stronger phase-amplitude coupling (PAC).

[0080] The calculated MI values ​​for each channel were compared with the MI value distribution of the same channel in a database of normal elderly individuals (age-matched, cognitively normal). Channels with MI values ​​significantly lower than the normal mean minus one standard deviation were selected as potential stimulation targets. In this embodiment, channel F3 was selected, with a resting-state MI value of [missing value]. Set the threshold for determining the difficult state to 1. 70%, that is: .

[0081] For example, if the individual's F3 channel =0.85, then =0.595, rounded to 0.60.

[0082] The channel with the lowest MI value and meeting the above-mentioned condition of being significantly lower than normal was identified as the target channel for closed-loop stimulation. In this embodiment, it was identified as F3, corresponding to the left dorsolateral prefrontal cortex. The determined individual theta peak frequency was... (6.5Hz in this example) This is set as the fixed frequency for subsequent transcranial electrical stimulation. The above individualized parameters (target channel F3, stimulation frequency 6.5Hz, difficult-state PAC threshold 0.60) are written into the control unit's configuration file for use in subsequent real-time phases.

[0083] During the training phase, the immersive scenario presentation module and the olfactory release module work together to present a three-dimensional everyday scene. For example, a table is presented in VR, with four items on it: a kettle, a coffee cup, a fruit basket, and a lamp. Each item is pre-defined as an independent area of ​​interest (AOI). The control unit simultaneously triggers the olfactory release module: it turns on the "coffee" scent generator and blows air for 2 seconds, allowing the user to smell the aroma of coffee; then it turns on the "fruit" scent generator corresponding to the fruit basket. About 1 second later, the user hears a spatialized instruction in VR: "Please carefully observe and remember the names, colors, and positions of all the items on the table. You will be tested on recall later."

[0084] Users freely observe and explore the scene by turning their heads and looking around, acquiring a memory encoding experience that integrates visual, auditory, and olfactory senses. Each task encoding phase lasts 30 seconds.

[0085] The system receives data in real time from the EEG acquisition module and the eye-tracking module, and all data streams are timestamped to maintain synchronization. Based on real-time cross-frequency phase-amplitude coupling (PAC) indicators and eye-tracking indicators, it jointly determines whether the user is in a state of memory encoding difficulty. Specifically, when the control unit determines a state of memory encoding difficulty, it performs the following operations:

[0086] A real-time sliding window is used to calculate the phase-amplitude coupling (PAC) value between the theta band phase and the gamma band amplitude in the frontotemporal region. A common average reference is used for rereference across all EEG channels, and a 1-second sliding time window (100ms step) is employed for real-time processing. Within each window, the PAC value of the F3 channel is rapidly calculated. To meet real-time requirements, Hilbert transform is used instead of wavelets: the theta component is obtained by 4-8Hz bandpass filtering of the data within the window. The phase is obtained by taking the Hilbert transform. Filter the frequency range to 30-80Hz, and use the Hilbert transform amplitude as the gamma amplitude envelope; then select the phase interval... Divide the container into 18 boxes and count the number of items that fall into each box. The mean value is calculated using the formula in sub-step 1.5 above and denoted as PAC. This calculation is performed once every 100ms during each sliding window update to obtain the PAC value at the current time.

[0087] Simultaneously determine whether the gaze point is fixed on a task-related target object and monitor changes in pupil diameter; specifically, determine whether the gaze point falls within the area of ​​interest (AOI) of any target object, record the duration of continuous gaze, and calculate the percentage change in current pupil diameter relative to the median pupil diameter 10 seconds before the start of the task.

[0088] A state is considered to be in a difficult state of memorization encoding when all of the following conditions are met:

[0089] The phase-amplitude coupling value is below a preset individualization threshold; for example, PAC < 0.6. This condition indicates that the user's current theta-gamma cross-frequency coupling strength in the prefrontal-temporal network is significantly lower than its resting-state baseline level, reflecting insufficient synchronization of key brain networks required for memory encoding.

[0090] The duration of continuous fixation on the same target object exceeds the first preset duration; the duration of fixation on an object is >1.8 seconds. This data can be determined based on preliminary experimental data: the average fixation duration of healthy elderly people on a single object in similar scene coding tasks is about 800-1200ms, and continuous fixation exceeding 1800ms usually indicates cognitive processing difficulties rather than normal deep coding.

[0091] An increase in pupil diameter relative to a baseline exceeding a first preset percentage, or a decrease exceeding a second preset percentage, is considered a negative pupil diameter. For example, an increase of >20% or a decrease of >15%. An abnormal increase in pupil diameter (>+20%) typically reflects excessive cognitive load or mental exertion; an abnormal decrease in pupil diameter (<-15%) may reflect inattention, avoidance processing, or decreased activity of the noradrenergic system due to fatigue.

[0092] Once all three conditions are met simultaneously, it is determined that a "memory encoding difficulty state" has occurred.

[0093] When a state of memory encoding difficulty is identified, the transcranial electrical stimulation (TCS) module is triggered to output a phase-locked alternating current (AC) stimulation pulse train to the target brain region based on the current instantaneous EEG phase. Specifically, the target brain region includes the left dorsolateral prefrontal cortex or the entorhinal cortex projection area. The individualized parameter calibration stage includes comparing the resting-state theta-gamma phase-amplitude coupling strength of different EEG channels and selecting the channel with the lowest coupling strength as the stimulation target. Preferably, the AC stimulation frequency output by the TCS module is equal to the individual theta peak frequency determined in the individualized parameter calibration stage, and the stimulation intensity is adaptively adjusted within a preset range.

[0094] Preferably, when triggering transcranial electrical stimulation, the control unit extracts the instantaneous phase of the target channel's theta band based on the latest EEG data at the moment of determining the difficult state, and calculates the target trigger phase by combining it with the system's inherent delay, thereby locking the AC stimulation output by the transcranial electrical stimulation module with the target trigger phase. Preferably, the control unit is also used to adaptively adjust the stimulation intensity or stimulation pulse train duration output by the transcranial electrical stimulation module according to the degree of deviation between the phase-amplitude coupling value and the preset individualized threshold; the greater the deviation, the higher the stimulation intensity or the longer the stimulation pulse train duration controlled by the control unit.

[0095] Example: Once a difficult state is determined at time t0, the control and computation unit extracts the F3 channel theta band signal of the latest 1-second EEG data and calculates the instantaneous phase using Hilbert transform. To compensate for signal transmission and algorithm delays (e.g., a measured average delay of 27ms), the predicted target trigger phase is set... for +ω·Δt, where ω=2π×6.5Hz and Δt is 30ms.

[0096] Then, a trigger command is immediately sent to the transcranial electrical stimulation module. The command includes the following parameters:

[0097] Output channels: F3 (stimulation electrode) and behind Cz (reference electrode)

[0098] Stimulus waveform: sinusoidal alternating current

[0099] frequency: =6.5Hz

[0100] Initial phase:

[0101] Intensity: Initial value =1.0mA

[0102] Duration: 8 complete theta cycles = 8 / 6.5Hz ≈ 1.23 seconds.

[0103] The initial current intensity is set to 1.0 mA. If a difficult state is triggered twice consecutively in the same task, it will automatically increase to 1.2 mA. If there is still no improvement, it will increase to a maximum of 1.5 mA (the preset safety limit in this embodiment). If the user experiences facial muscle twitching or actively reports discomfort at any time, the operator can stop the stimulation output via an external button. During stimulation, the EEG signal may contain artifacts. During this 1.23-second period, the system will stop PAC calculation and difficult state judgment, and instead use the last effective state before stimulation.

[0104] Preferably, the control unit is also used to control the immersive scene presentation module to visually highlight the target object currently being viewed by the user, or to reduce the complexity of the scene background, while triggering transcranial electrical stimulation. In VR, the control unit adds a bright yellow outline to the edge of the object currently being viewed by the user and reduces the visual transparency of unrelated objects in the background by 20% for 2 seconds. This operation reduces the visual search burden and guides the user to reallocate their attention to the target object.

[0105] Example 2:

[0106] As attached Figure 4 As shown, the difference from Example 1 is that a brain rehabilitation training method for mild cognitive impairment is also provided, including the following steps:

[0107] Step A: Individualized Parameter Calibration: Resting-state EEG data is collected from the user. Individual peak frequencies in the theta band (4-8 Hz) are extracted from the frontotemporal lobe regions such as F3, F4, Fz, and T7. The phase-amplitude coupling strength between the theta phase and gamma amplitude is calculated to determine the stimulation target, stimulation frequency, and difficult state threshold. For example, the peak frequency obtained from the F3 channel is 6.5 Hz. Subsequently, the phase-amplitude coupling (PAC) strength between the theta phase and gamma amplitude (30-80 Hz) in the resting state of this channel is calculated. This embodiment uses the method proposed by Canolty: the signal is filtered using complex Morlet wavelets to obtain the instantaneous phase in the theta band. and gamma band amplitude envelope ;by Divide the entire phase period into 18 equal boxes (each box 20°) with the edge as the boundary, and calculate the value of each box. The mean of the distribution is calculated, and then the vector length of this mean (the deviation from the uniform distribution) is taken as the PAC value. In this example, channel F3 is selected, which has an individual theta peak of 6.5Hz and the lowest PAC value. This frequency is set as the fixed frequency for subsequent stimuli.

[0108] Step B, Multi-sensory Contextual Memory Training: Using virtual reality, a three-dimensional daily life scene containing memory tasks is presented to the user, and a scent matching the scene content is released simultaneously while the scene is presented, allowing the user to freely observe and perform memory encoding.

[0109] Step C, Real-time Status Analysis: Continuously collect the user's EEG and eye movement signals, calculate the theta-gamma phase-amplitude coupling (PAC) value at the stimulation target point in real time, and monitor the fixation point position and pupil diameter. When the phase-gamma coupling value is lower than the difficult state determination threshold, and the fixation point dwell time and pupil diameter change meet preset conditions, it is determined to be a memory encoding difficult state. The control unit pre-stores the user's individual difficult state determination criteria: PAC < 0.6 (determined based on 70% of the individual's resting state PAC value of 0.85 in Step 1), fixation time on an object > 1.8 seconds, and (pupil diameter increase > 20% or decrease > 15%). Once all three conditions are met simultaneously, it is determined that a "memory encoding difficult state" has occurred, and immediately proceed to Step D. If not met, continue collecting data and sliding window progression until the encoding stage ends.

[0110] Step D: Closed-loop phase-synchronized electrical stimulation intervention: Once a memory encoding difficulty state is identified, the F3 channel theta band signal of the latest 1-second EEG data is extracted, and the instantaneous phase φ(t0) is calculated using Hilbert transform. Immediately afterwards, a trigger command is sent to the transcranial electrical stimulation module, outputting a continuous sine wave at the next sampling point with a frequency of 6.5 Hz and an initial phase equal to the predicted phase, lasting for 8 theta cycles (approximately 1.23 seconds). The initial current intensity is set to 1.0 mA. If a difficulty state is triggered twice consecutively in the same task, it automatically increases to 1.2 mA; if there is still no improvement, it increases to a maximum of 1.5 mA; if discomfort such as facial muscle twitching occurs, the experimenter can stop or reduce the current.

[0111] Step E, Multi-cycle Iteration and Evaluation: Repeat steps B to D. After completing a predetermined number of memory training tasks, collect resting-state EEG again for evaluation and generate a training effect report. A single training sequence contains multiple memory tasks in different scenarios (kitchen, living room, bedroom, study, etc.), each task coded for 30 seconds, followed by a testing phase: the user uses a VR controller to identify objects in the scene and drag them back to their original positions. The system records behavioral data such as the number of correctly remembered items and positional accuracy. One round of training lasts approximately 20 minutes. After training, collect resting-state EEG and task-oriented PAC indicators for another 5 minutes and compare them with pre-training data. The control unit generates a detailed neurobehavioral report in the background, including theta-gamma PAC increase amplitude, changes in the number of difficult-state triggers, and memory accuracy trends, for clinical personnel to review. Subjects can exit or view historical trends through the main menu.

[0112] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A brain rehabilitation training device for mild cognitive impairment, characterized in that, include: The EEG acquisition module is used to acquire EEG signals from the user's frontal-temporal lobe region in real time. The transcranial electrical stimulation module is used to deliver alternating current stimulation to the user's target brain regions; The immersive scene presentation module is used to present users with three-dimensional daily life scenes containing memory tasks and spatialized audio; The olfactory release module is used to release a scent that matches the scene into the user's nasal cavity when the immersive scene presentation module presents the scene; The eye-tracking module is used to capture the user's eye movement metrics in real time, including fixation point position and pupil diameter data. The control unit is communicatively connected to the above modules and is used to execute the following closed-loop adaptive rehabilitation training control: During the individualized parameter calibration phase, the control unit analyzes the user's resting-state EEG, extracts the individual peak frequency of the frequency band related to the memory task, calculates the cross-frequency phase-amplitude coupling strength, and determines the stimulation target and frequency. During the training phase, the control unit controls the immersive scene presentation module and the olfactory release module to jointly present three-dimensional daily life scenes. It receives data from the EEG acquisition module and the eye-tracking module in real time, and determines whether the user is in a state of memory encoding difficulty based on the real-time cross-frequency phase-amplitude coupling index and eye-tracking index. When a state of memory encoding difficulty is identified, the transcranial electrical stimulation module is triggered to output a phase-locked AC stimulation pulse train to the target brain region based on the current instantaneous phase of the EEG.

2. The brain rehabilitation training device for mild cognitive impairment according to claim 1, characterized in that, When the control unit determines that the memory encoding is difficult, it performs the following operations: A real-time sliding window is used to calculate the phase-amplitude coupling value between the theta band phase and the gamma band amplitude in the frontotemporal region; simultaneously, it is determined whether the fixation point is stationary on a task-related target object, and changes in pupil diameter data are monitored; a state of memory encoding difficulty is identified when the following conditions are met simultaneously: The phase-amplitude coupling value is lower than the preset individualization threshold; The gaze point remains continuously on the same target object for more than a first preset duration; The pupil diameter increases by more than a first preset ratio relative to the baseline state, or decreases by more than a second preset ratio.

3. The brain rehabilitation training device for mild cognitive impairment according to claim 2, characterized in that, The control unit is also used to control the immersive scene presentation module to visually highlight the target object that the user is currently looking at, or to reduce the complexity of the scene background, while triggering transcranial electrical stimulation.

4. The brain rehabilitation training device for mild cognitive impairment according to claim 3, characterized in that, The olfactory release module contains several odor generators, each containing different types of odor cartridges. The control unit selectively activates one or more odor generators based on the current scene type presented by the immersive scene presentation module, so that the odor matches the scene content.

5. The brain rehabilitation training device for mild cognitive impairment according to claim 4, characterized in that, The target brain regions include the left dorsolateral prefrontal cortex or the entorhinal cortex projection area; the individualized parameter calibration stage includes comparing the resting-state theta-gamma phase-amplitude coupling strength of different EEG channels, and selecting the channel with the lowest coupling strength as the stimulation target.

6. The brain rehabilitation training device for mild cognitive impairment according to claim 5, characterized in that, The AC stimulation frequency output by the transcranial electrical stimulation module is equal to the individual theta peak frequency determined during the individualized parameter calibration stage, and the stimulation intensity is adaptively adjusted within a preset range.

7. The brain rehabilitation training device for mild cognitive impairment according to claim 6, characterized in that, When transcranial electrical stimulation is triggered, the control unit extracts the instantaneous phase of the target channel theta band based on the latest EEG data at the moment of determining the difficult state, and calculates the target trigger phase by combining it with the inherent delay of the system, so that the AC stimulation output by the transcranial electrical stimulation module is locked with the target trigger phase.

8. The brain rehabilitation training device for mild cognitive impairment according to claim 7, characterized in that, The EEG acquisition module uses a 32-channel active electrode cap, and the electrode positions are arranged according to the international 10-20 system.

9. The brain rehabilitation training device for mild cognitive impairment according to claim 8, characterized in that, The control unit is also used to adaptively adjust the stimulation intensity or duration of the stimulation pulse train output by the transcranial electrical stimulation module according to the degree of deviation between the phase-amplitude coupling value and the preset individualized threshold. The greater the deviation, the higher the stimulation intensity controlled by the control unit or the longer the duration of the stimulation pulse train.

10. A method for brain rehabilitation training for mild cognitive impairment, performed using the brain rehabilitation training device for mild cognitive impairment as described in any one of claims 1-9, characterized in that, Includes the following steps: Step A, Individualized parameter calibration: Collect the user's resting-state EEG, extract the individual peak frequency of the frontotemporal lobe region in the theta band, calculate the phase-amplitude coupling strength between the theta phase and the gamma amplitude, and determine the stimulation target, stimulation frequency and difficult state judgment threshold in this way. Step B, Multi-sensory Contextual Memory Training: Using virtual reality, a three-dimensional daily life scene containing memory tasks is presented to the user, and a scent matching the scene content is released simultaneously while the scene is presented, allowing the user to freely observe and perform memory encoding. Step C, Real-time State Analysis: Continuously collect the user's EEG and eye movement signals, calculate the theta-gamma phase-amplitude coupling value at the stimulation target point in real time, and monitor the fixation point position and pupil diameter. When the phase-amplitude coupling value is lower than the difficult state determination threshold, and the fixation point dwell time and pupil diameter change meet the preset conditions, it is determined to be a difficult state of memory encoding. Step D, Closed-loop phase-synchronous electrical stimulation intervention: Once the memory encoding difficulty state is determined, the instantaneous phase of the current theta frequency band is extracted, and after delay compensation, the transcranial electrical stimulation module is triggered to output phase-locked AC electrical stimulation to the stimulation target at the individual peak frequency; Step E, Multi-cycle Iteration and Evaluation: Repeat steps B to D. After completing a predetermined number of memory training tasks, collect resting-state EEG data again for evaluation and generate a training effect report.

Images