Deep brain targeted stimulation method based on time-frequency interference technology

Abstract

The invention discloses a deep brain targeted stimulation method based on a time-frequency interference technology, and relates to the technical field of nerve regulation, and an application system comprises a plurality of pairs of stimulation electrode modules, a data acquisition module, a data preprocessing module, a time-frequency interference stimulation signal generation module, a central control module and a data storage module. According to the method, a multi-azimuth surrounding layout of multiple pairs of stimulation electrodes is combined with a time-frequency interference technology, so that a high-frequency electric field forms constructive interference in a deep brain target area, and meanwhile, a frequency difference is dynamically adjusted based on different organism state requirements; and a closed-loop control system of data acquisition, preprocessing, state evaluation, parameter adjustment and stimulation application is constructed. The problems that a traditional deep brain stimulation technology is low in targeting precision, poor in state guiding efficiency and insufficient in stimulation system stability are solved, the energy focusing degree is improved to 90% or above, the sleep induction time is shortened to 20-30 minutes, and the method can be widely applied to nerve disease treatment and physiological state improvement.

Description

Technical Field

[0001] This invention relates to the field of neuromodulation technology, and more specifically, to a method for deep brain targeted stimulation based on time-frequency interference technology. Background Technology

[0002] Currently, deep brain stimulation (DBS) technology has some applications in clinical and research fields. Existing techniques mostly use a single pair or a few pairs of stimulation electrodes to stimulate target areas deep in the brain. Taking deep brain stimulation (DBS) as an example, its typical approach involves implanting a single or single pair of stimulation electrodes in the cranium to apply electrical stimulation signals at a fixed frequency, thereby modulating the activity of neurons in the brain to treat certain neurological diseases (such as Parkinson's disease) or improve specific physiological states. Temporal interference (TI) electrical stimulation applies two high-frequency alternating electric fields (typically 1-2 kHz, with a frequency difference Δf of 5-100 Hz), utilizing the nonlinear superposition of electric fields in biological tissues to generate a low-frequency modulation envelope (frequency equal to Δf). High-frequency carrier waves cannot directly activate neural tissue due to the "high-frequency filtering characteristics" of neuronal membranes, but they can penetrate superficial structures such as the scalp and skull to reach deep brain tissue. Low-frequency envelopes, on the other hand, can synchronize with the neuronal membrane time constant (10-100ms) and drive neuronal depolarization through the superposition of electrotonic potentials, achieving "non-invasive deep targeted stimulation". The core formula is: E(t)=E1sin(2πf1t)+E2sin(2πf2t), where E1 and E2 are the high-frequency electric field strengths, f1 and f2 are the carrier frequencies, and the low-frequency envelope frequency Δf=|f1-f2|.

[0003] Traditional intracranial stimulation (TI) uses two pairs of symmetrical electrodes (four electrodes in total), typically placed on both sides of the head (e.g., left anterior-left posterior, right anterior-right posterior). By adjusting the electrode current intensity and phase, a high-frequency electric field is used to constructively interfere in specific deep intracranial regions (e.g., hippocampus, thalamus), strengthening the low-frequency envelope; while in superficial tissues, due to weaker electric field intensity or phase cancellation, the neuronal membrane potential does not reach the activation threshold, achieving "deep activation, superficial silencing." Typical parameters are: carrier frequency 1-2 kHz, Δf = 5-50 Hz, current intensity < 1.5 mA (humans), and a linear relationship between electrode spacing and stimulation depth (d = k × s, where k is a tissue conductivity-related constant).

[0004] The core mechanisms by which TI achieves deep brain stimulation include:

[0005] High-frequency penetration: High-frequency electric fields (>1kHz) have a low attenuation coefficient in biological tissues and can penetrate the skull to reach deep tissues (such as the hippocampus, 5-10cm deep), while low-frequency electric fields (<1kHz) are attenuated at the surface.

[0006] Depolarization effect of low-frequency envelope: The low-frequency envelope generated by interference is superimposed through electrotense potential (V-film). (E_total·σ), where σ is the tissue conductivity), which depolarizes the membrane potential of deep neurons.

[0007] Dynamic amplification of ion channels: After depolarization reaches the threshold, voltage-gated sodium channels open, triggering action potentials; while superficial neurons, due to insufficient electric field strength, are dominated by potassium channels that hyperpolarize the membrane potential, thus avoiding activation.

[0008] However, traditional two-pair electrodes have obvious targeting limitations. In a single interference region, two pairs of electrodes can only form 1-2 fixed interference focal points, which cannot cover complex brain regions (such as the basal ganglia subnucleus).

[0009] Tissue heterogeneity interference: Differences in conductivity between tissues such as the skull and cerebrospinal fluid (e.g., cerebrospinal fluid σ = 1.654 S / m, skull σ = 0.008 S / m) cause distortion of the electric field distribution, making the actual focal point easily deviate from the target (error > 2 mm).

[0010] Insufficient spatial resolution: The low-frequency envelope has a Gaussian-like distribution with a half-width at half-maximum (HWHM) of about 1 cm, making it difficult to focus on millimeter-level subnuclei (such as the substantia nigra pars compacta), which may activate adjacent neural nuclei and cause side effects (such as increased tremor).

[0011] Furthermore, traditional fixed-frequency stimulation cannot match the dynamic needs of biological rhythms. In sleep induction, single-frequency stimulation (e.g., 50Hz) struggles to synchronize the thalamus-cortex network, with an average induction time of 60-90 minutes and delta waves (1-4Hz) accounting for less than 30%, easily leading to sleep fragmentation. In inflammation control, non-specific electrical stimulation only indirectly affects the immune response through generalized neural regulation, with an inflammation resolution cycle as long as 7-10 days and unstable effects. Meanwhile, existing open-loop stimulation systems have slow response times, easily leading to effect decay or overstimulation. Summary of the Invention

[0012] To overcome the aforementioned deficiencies of the prior art, this invention provides a deep brain targeted stimulation method based on time-frequency interferometry to solve the problems of low targeting accuracy, poor state guidance efficiency, and insufficient stability of the stimulation system in traditional deep brain stimulation techniques.

[0013] The technical solution adopted by this invention to solve its technical problem is: a deep brain targeted stimulation method based on time-frequency interferometry, the system of which includes a multi-pair stimulation electrode module, a data acquisition module, a data preprocessing module, a time-frequency interferometry stimulation signal generation module, a central control module, and a data storage module; the multi-pair stimulation electrode module contains n pairs (n≥1, expandable to thousands of pairs) of stimulation electrode units, each pair of electrode units being unidirectionally connected to the time-frequency interferometry stimulation signal generation module through an independent signal transmission line; the data acquisition module communicates bidirectionally with the data preprocessing module through a data bus; the data preprocessing module is connected to the central control module through an internal bus and communicates synchronously with the data storage module; the time-frequency interferometry stimulation signal generation module and the central control module are connected through a bidirectional communication interface.

[0014] This method comprises three stages: data acquisition and preprocessing, stimulation parameter optimization and signal generation, and targeted stimulation and feedback adjustment. The specific steps are as follows:

[0015] Data acquisition and preprocessing stage

[0016] S1: The data acquisition module acquires EEG, fMRI-BOLD signals and tissue impedance at a sampling rate of 1kHz, with a sampling period of 10ms, comprehensively acquiring neurophysiological signals, cerebral blood flow signals and tissue impedance data, providing a basis for subsequent analysis.

[0017] S2: The data preprocessing module removes eye movement / electromyography artifacts through wavelet threshold denoising and ICA independent component analysis, and outliers are removed using the 3σ criterion. Wavelet threshold denoising can use db4 wavelet basis-5 layer decomposition or sym8 wavelet basis-4 layer decomposition and adaptive Kalman filtering, with a signal-to-noise ratio difference ≤5%. Outlier processing can use the 3σ principle-linear interpolation, or box plot method and cubic spline interpolation, with an error supplement of <2%, ensuring the accuracy and reliability of the data.

[0018] S3: The processed data is transmitted to the central control module via the internal bus and simultaneously stored in the data storage module for subsequent data retrieval and analysis, while also enabling data backup and management.

[0019] Stimulus parameter optimization and signal generation stage

[0020] S4: The central control module analyzes data based on a deep learning model (LSTM+CNN) and outputs stimulus parameter instructions, including target region coordinates, frequency difference Δf, amplitude (0.5-1.2mA), and pulse width (200μs). The frequency difference Δf can be flexibly adjusted for different application scenarios, such as Δf = 4-8Hz for cognitive enhancement, Δf = 13-30Hz for movement disorder treatment, Δf = 0.5-2Hz for epilepsy control, Δf = 0.5-2Hz for sleep induction, and Δf = 10-15Hz for inflammation control, achieving precise matching of "stimulus parameters - target state." Furthermore, the central control module can also add a multimodal AI model, inputting EEG, fMRI, and inflammatory factor concentration data, and outputting an adaptively adjusted frequency difference to further improve the accuracy of parameter optimization.

[0021] S5: The time-frequency interference stimulation signal generation module receives instructions and generates a high-frequency signal through direct digital synthesis (DDS). The high-frequency signal can be a sine wave, a square wave, or a sawtooth wave. After filtering out harmonics, the frequency difference error of the square wave is <1%, and the energy utilization rate of the sawtooth wave is improved by 15%. The generated high-frequency signal is amplified (efficiency >90%) and then sent to the electrode unit through an independent transmission line to provide a suitable signal source for subsequent targeted stimulation.

[0022] Targeted Stimulation and Feedback Adjustment Phase

[0023] S6: Multiple pairs of electrodes release high-frequency electric fields to form a focused low-frequency envelope in the target area, inducing action potentials. The activation effect can be verified by monitoring the N170 / P300 waveform via EEG, ensuring that the stimulation acts on the target area and produces the expected neural response.

[0024] S7: The central control module receives feedback signals in real time. If the electric field strength is less than the threshold of 0.1V / m, it adjusts the phase difference (step size 0.1°). The closed-loop control cycle is ≤50ms, ensuring stable and dynamically optimized stimulation effects, avoiding the effect decay or overstimulation problems that easily occur in open-loop stimulation of existing technologies. Furthermore, the stimulation electrode unit uses a titanium alloy substrate with an iridium oxide coating, or a platinum-iridium alloy with a titanium nitride coating, or a pure gold-diamond-like coating. Its biocompatibility meets ISO 10993 standards, and signal transmission loss is ≤2%, reducing foreign body reactions between electrodes and tissues and improving signal transmission quality. Different array layouts are used for different brain regions. Deep nuclei use a dodecahedral array with an electrode spacing of 500-800μm, covering a target area of ​​±2mm. Cortical regions use a 4×4 grid array with an electrode spacing of 1-2mm, adapting to the anatomical structures of different brain regions and improving stimulation accuracy. Each pair of electrode units is implanted with the assistance of a miniature robotic arm or a stereotactic brain scanner. Based on preoperative MRI / CT positioning, the error is ≤0.1mm. It supports dynamic adjustment of angle (±15°) and depth (±0.5mm) after surgery, further ensuring the accuracy and flexibility of electrode implantation position.

[0025] Independent signal transmission lines use an insulating layer (such as polyimide) + conductive core wire (such as platinum or copper), with a diameter ≤100μm and a tensile strength >50MPa to prevent breakage after implantation; each transmission line integrates an independent shielding layer with a crosstalk rate <-80dB, reducing signal interference between electrodes and ensuring the stability and accuracy of signal transmission.

[0026] The beneficial effects of this invention are:

[0027] Significantly improved stimulation precision: Multiple electrode pairs employ a "multi-directional surrounding layout," with the phase and frequency of each pair dynamically adjusted by a central control module. This allows the high-frequency electric field to generate constructive interference in the target area, while non-target areas experience "energy troughs" due to phase cancellation. Based on the electric field superposition formula, by optimizing the phase difference, the superposition gain of the electric field intensity in the target area is >20dB, while the intensity in the surrounding area is reduced by more than 60%. The energy focusing efficiency is increased from 30%-50% in existing technologies to over 90%, and the energy density in the target area is 10-20 times higher than that in the surrounding tissue (in animal experiments, the stimulation energy density in the hippocampus reached 1.2V / m, while it was only 0.1V / m in the cortex). The activation rate of non-target neurons is reduced from 30% to below 5% (in the Parkinson's model, 38% of the traditional stimulation group experienced dyskinesia, while only 4% of the group using this invention experienced dyskinesia), significantly reducing the incidence of side effects.

[0028] Significantly improved efficiency in inducing biological states: Specific frequency difference stimulation protocols are designed to meet the needs of different biological states. During sleep induction, Δf = 0.5-2Hz (δ wave band) is used. Multiple pairs of electrodes form a low-frequency envelope in the thalamic nuclei, directly driving thalamic neurons to generate synchronous firing, promoting the release of GABAergic inhibitory neurotransmitters, accelerating sleep onset, and shortening the induction time to 20-30 minutes (50%-67% improvement over existing technologies). The proportion of δ waves is stably maintained above 50%, and the sleep cycle is extended by 2-3 hours (in a rat model, total sleep time increased from 4.2 hours to 6.5 hours). During inflammation control, a Δf = 10-15 Hz (β wave band) is used to target and stimulate the dorsal nucleus of the vagus nerve. This upregulates anti-inflammatory factors (such as IL-10) and inhibits the release of pro-inflammatory factors (such as TNF-α) through the "neuro-immune axis." Based on the dose-response relationship between frequency difference and vagal nerve activation, the inflammation resolution cycle can be shortened to 3-5 days (40%-50% improvement over existing technologies).

[0029] The stimulation system boasts high stability and safety: a closed-loop control system is constructed, encompassing data acquisition, preprocessing, status assessment, parameter adjustment, and stimulation application. The delay from data acquisition to parameter adjustment is ≤50ms, representing an 8-fold improvement in response speed compared to open-loop systems, ensuring stable stimulation effects (error <±5% in long-term experiments). The stimulation electrode unit utilizes biocompatible materials, passing ISO 10993-5 cytotoxicity testing, extending the implantation period to 2 years (traditional materials show corrosion after 6 months), reducing the risk of secondary surgery. Simultaneously, strict control ensures brain tissue temperature drift ≤0.5℃ and inflammatory factor levels ≤1.2 times the normal range, preventing irreversible neurological damage and guaranteeing the safety of the stimulation process. Attached Figure Description

[0030] The structures, proportions, sizes, etc. illustrated in this specification are only for the purpose of assisting those skilled in the art in understanding and reading the content disclosed herein, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0031] Figure 1 This is a schematic diagram illustrating the technical principle of the present invention;

[0032] Figure 2 Schematic diagram of multiple electrode placement for mice and patients;

[0033] Figure 3 This is a schematic diagram illustrating the practical application scenarios of dodecahedral arrays and mesh arrays;

[0034] Figure 4This is a block diagram of a deep brain targeted stimulation system based on time-frequency interferometry. Detailed Implementation

[0035] The following specific embodiments illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. 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.

[0036] Refer to the instruction manual appendix Figure 1-4 This embodiment describes a deep brain targeted stimulation method based on time-frequency interferometry. The applied system includes a multi-pair stimulation electrode module, a data acquisition module, a data preprocessing module, a time-frequency interferometry stimulation signal generation module, a central control module, and a data storage module. The multi-pair stimulation electrode module contains n pairs (n≥1, expandable to thousands of pairs) of stimulation electrode units, each pair of electrode units being unidirectionally connected to the time-frequency interferometry stimulation signal generation module via an independent signal transmission line. The data acquisition module communicates bidirectionally with the data preprocessing module via a data bus. The data preprocessing module is connected to the central control module via an internal bus and communicates synchronously with the data storage module. The time-frequency interferometry stimulation signal generation module is connected to the central control module via a bidirectional communication interface.

[0037] This method comprises three stages: data acquisition and preprocessing, stimulation parameter optimization and signal generation, and targeted stimulation and feedback adjustment. The specific steps are as follows:

[0038] Data acquisition and preprocessing stage

[0039] S1: The data acquisition module acquires EEG, fMRI-BOLD signals and tissue impedance at a sampling rate of 1kHz, with a sampling period of 10ms, comprehensively acquiring neurophysiological signals, cerebral blood flow signals and tissue impedance data, providing a basis for subsequent analysis.

[0040] S2: The data preprocessing module removes eye movement / electromyography artifacts through wavelet threshold denoising and ICA independent component analysis, and outliers are removed using the 3σ criterion. Wavelet threshold denoising can use db4 wavelet basis-5 layer decomposition or sym8 wavelet basis-4 layer decomposition and adaptive Kalman filtering, with a signal-to-noise ratio difference ≤5%. Outlier processing can use the 3σ principle-linear interpolation, or box plot method and cubic spline interpolation, with an error supplement of <2%, ensuring the accuracy and reliability of the data.

[0041] S3: The processed data is transmitted to the central control module via the internal bus and simultaneously stored in the data storage module for subsequent data retrieval and analysis, while also enabling data backup and management.

[0042] Stimulus parameter optimization and signal generation stage

[0043] S4: The central control module analyzes data based on a deep learning model (LSTM+CNN) and outputs stimulus parameter instructions, including target region coordinates, frequency difference Δf, amplitude (0.5-1.2mA), and pulse width (200μs). The frequency difference Δf can be flexibly adjusted for different application scenarios, such as Δf = 4-8Hz for cognitive enhancement, Δf = 13-30Hz for movement disorder treatment, Δf = 0.5-2Hz for epilepsy control, Δf = 0.5-2Hz for sleep induction, and Δf = 10-15Hz for inflammation control, achieving precise matching of "stimulus parameters - target state." Furthermore, the central control module can also add a multimodal AI model, inputting EEG, fMRI, and inflammatory factor concentration data, and outputting an adaptively adjusted frequency difference to further improve the accuracy of parameter optimization.

[0044] S5: The time-frequency interference stimulation signal generation module receives instructions and generates a high-frequency signal through direct digital synthesis (DDS). The high-frequency signal can be a sine wave, a square wave, or a sawtooth wave. After filtering out harmonics, the frequency difference error of the square wave is <1%, and the energy utilization rate of the sawtooth wave is improved by 15%. The generated high-frequency signal is amplified (efficiency >90%) and then sent to the electrode unit through an independent transmission line to provide a suitable signal source for subsequent targeted stimulation.

[0045] Targeted Stimulation and Feedback Adjustment Phase

[0046] S6: Multiple pairs of electrodes release high-frequency electric fields to form a focused low-frequency envelope in the target area, inducing action potentials. The activation effect can be verified by monitoring the N170 / P300 waveform via EEG, ensuring that the stimulation acts on the target area and produces the expected neural response.

[0047] S7: The central control module receives feedback signals in real time. If the electric field strength is less than the threshold of 0.1V / m, it adjusts the phase difference (step size 0.1°). The closed-loop control cycle is ≤50ms, ensuring stable and dynamically optimized stimulation effects, avoiding the effect decay or overstimulation problems that easily occur in open-loop stimulation of existing technologies. Furthermore, the stimulation electrode unit uses a titanium alloy substrate with an iridium oxide coating, or a platinum-iridium alloy with a titanium nitride coating, or a pure gold-diamond-like coating. Its biocompatibility meets ISO 10993 standards, and signal transmission loss is ≤2%, reducing foreign body reactions between electrodes and tissues and improving signal transmission quality. Different array layouts are used for different brain regions. Deep nuclei use a dodecahedral array with an electrode spacing of 500-800μm, covering a target area of ​​±2mm. Cortical regions use a 4×4 grid array with an electrode spacing of 1-2mm, adapting to the anatomical structures of different brain regions and improving stimulation accuracy. Each pair of electrode units is implanted with the assistance of a miniature robotic arm or a stereotactic brain scanner. Based on preoperative MRI / CT positioning, the error is ≤0.1mm. It supports dynamic adjustment of angle (±15°) and depth (±0.5mm) after surgery, further ensuring the accuracy and flexibility of electrode implantation position.

[0048] Independent signal transmission lines use an insulating layer (such as polyimide) + conductive core wire (such as platinum or copper), with a diameter ≤100μm and a tensile strength >50MPa to prevent breakage after implantation; each transmission line integrates an independent shielding layer with a crosstalk rate <-80dB, reducing signal interference between electrodes and ensuring the stability and accuracy of signal transmission.

[0049] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A deep brain targeted stimulation method based on time-frequency interferometry, characterized in that, The system includes a multi-pair stimulation electrode module (10), a data acquisition module (20), a data preprocessing module (30), a time-frequency interference stimulation signal generation module (40), a central control module (50), and a data storage module (60). The multi-pair stimulation electrode module (10) contains n pairs (n≥1, expandable to thousands of pairs) of stimulation electrode units (11), and each pair of electrode units (11) is unidirectionally connected to the time-frequency interference stimulation signal generation module (40) through an independent signal transmission line. The data acquisition module (20) communicates bidirectionally with the data preprocessing module (30) through a data bus (31). The data preprocessing module (30) is connected to the central control module (50) through an internal bus (32) and communicates synchronously with the data storage module (60). The time-frequency interference stimulation signal generation module (40) is connected to the central control module (50) through a bidirectional communication interface (51). The method includes the following steps: S1: The data acquisition module (20) acquires EEG, fMRI-BOLD signals and tissue impedance at a sampling rate of 1kHz, with a sampling period of 10ms. S2: The data preprocessing module (30) removes eye movement / electromyography artifacts by wavelet thresholding and ICA independent component analysis, and removes outliers by using the 3σ criterion. S3: The processed data is transmitted to the central control module (50) via the internal bus (32) and simultaneously stored in the data storage module (60); S4: The central control module (50) analyzes data based on a deep learning model and outputs stimulus parameter instructions, including target area coordinates, frequency difference Δf, amplitude and pulse width; S5: The time-frequency interference stimulation signal generation module (40) receives the instruction, generates a high-frequency signal through direct digital synthesis, and sends it to the electrode unit (11) through an independent transmission line after power amplification. S6: Multiple pairs of electrodes (11) release high-frequency electric fields to form a focused low-frequency envelope in the target area, inducing action potentials; S7: The central control module (50) receives feedback signals in real time. If the electric field strength is less than the threshold of 0.1V / m, the phase difference is adjusted and the closed-loop control cycle is less than 50ms.

2. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that... For deep nuclei, the stimulation electrode units (11) are arranged in a symmetrical array with an electrode spacing of 500-1500μm, covering the target area within ±2mm; for the dermal area, a 4×4 grid array is used with an electrode spacing of 1-10mm.

3. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, The stimulation electrode unit (11) adopts a titanium alloy substrate with an iridium oxide coating, a platinum-iridium alloy, conductive carbon or conductive polymer and other conductive coatings, and its biocompatibility meets the ISO 10993 standard, with signal transmission loss ≤2%.

4. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, The independent signal transmission line adopts an insulation layer + conductive core wire with a diameter ≤100μm and a tensile strength >50MPa. Each transmission line integrates an independent shielding layer with a crosstalk rate <-80dB.

5. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, The wavelet threshold denoising uses db4 wavelet basis-5 decomposition or sym8 wavelet basis-4 decomposition and adaptive Kalman filtering, with a signal-to-noise ratio difference of ≤5%; outlier processing uses 3σ principle-linear interpolation, or box plot method and cubic spline interpolation, with a supplementary error of <2%.

6. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, The high-frequency signal generated by the time-frequency interference stimulation signal generation module (40) is a sine wave, or a square wave or a sawtooth wave. After filtering out harmonics, the frequency difference error of the square wave is <1%, and the energy utilization rate of the sawtooth wave is increased by 15%.

7. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, During cognitive enhancement, the frequency difference Δf = 4-8 Hz; during motor disorder treatment, Δf = 13-30 Hz; during epilepsy control, Δf = 0.5-2 Hz; during sleep induction, Δf = 0.5-2 Hz; and during inflammation control, Δf = 10-15 Hz.

8. The deep brain targeted stimulation method based on time-frequency interferometry according to claim 1, characterized in that, The central control module (50) can be equipped with a multimodal AI model, which can input EEG, fMRI, and inflammatory factor concentration data and output adaptively adjusted frequency difference.

Images