A non-invasive sleep assistance improvement method and system

Abstract

The present invention discloses a non-invasive sleep assistance improvement method and system, belonging to the technical field of biomedical engineering. The method specifically includes: collecting static data and dynamic data, generating a preliminary three-dimensional grid model according to the collected data, constructing a functional connectivity matrix and a dynamic functional connectivity matrix, generating an individualized three-dimensional grid model, selecting a stimulation area according to the individualized three-dimensional grid model, generating an electrical stimulation signal, and determining stimulation parameters, simulating the neural response of the target area in combination with a machine learning model, screening out the optimal stimulation parameter combination, and applying the electrical stimulation signal according to the optimal stimulation parameter combination to assist in improving the user's sleep with the optimal stimulation parameter combination; the present invention constructs an individualized three-dimensional grid model, optimizes the stimulation parameters in real time, forms a personalized stimulation parameter combination scheme, and improves the stability of the electroencephalogram signal.

Description

Technical Field

[0001] The present invention belongs to the technical field of biomedical engineering, and specifically relates to a non-invasive sleep assistance improvement method and system. Background Art

[0002] The sleep process of humans is closely related to brain electrical activities. Brain waves are electrical signals that reflect the synchronous electrical activities of brain neurons. Their frequencies, waveforms, and stability are significantly correlated with sleep states. When the brain electrical signals show abnormal fluctuations or unstable states, they usually interfere with the normal falling asleep process and sleep quality.

[0003] To achieve stable regulation of brain electrical signals, the existing technical means are divided into two major categories: invasive and non-invasive. Although invasive methods can achieve precise intervention on nerve signals, they have problems such as complex operations and high risks. Non-invasive methods have higher safety and acceptability.

[0004] Common non-invasive techniques include using means such as sound, light stimulation, weak electric fields, and magnetic fields for guidance or external regulation. However, these methods generally have the following deficiencies: First, the feedback mechanism is imperfect, and it is impossible to accurately and real-time obtain changes in electrical signals, thus affecting the regulation effect; second, the adaptability of the regulation signals is poor, and it is difficult to effectively respond to specific abnormal electrical signals; third, it is impossible to form dynamic monitoring and real-time optimization of electrical signals, resulting in deviations in the stable regulation of electrical signals. Summary of the Invention

[0005] In view of the deficiencies of the prior art, the present invention proposes a non-invasive sleep assistance improvement method and system, which performs personalized adjustment in combination with individual brain characteristics, uses a current with a specific frequency, intensity, and waveform, and precisely stimulates the target brain area through a group of non-invasive electrode devices to achieve optimization of the stimulation depth and effect.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A non-invasive sleep assistance improvement method, comprising:

[0008] Collect static data and dynamic data, generate a preliminary three-dimensional grid model according to the collected data, construct a functional connectivity matrix and a dynamic functional connectivity matrix, generate an individualized three-dimensional grid model. The dynamic functional connectivity matrix encodes task events as time series, dynamically calculates the information transfer direction and intensity through a modulation function, and shares task modulation parameters at the population level;

[0009] Select a stimulation area according to the individualized three-dimensional grid model, generate an electrical stimulation signal, and determine stimulation parameters;

[0010] Simulate the neural response of the target area with a machine learning model, screen out the optimal combination of stimulation parameters, and use the optimal combination of stimulation parameters to assist in improving the user's sleep.

[0011] Specifically, the generation of the individualized three-dimensional grid model includes:

[0012] Construct a surface topology map of the neural signal activity area based on static data and dynamic data, and identify and segment the surface topology map of the neural signal activity area;

[0013] Calculate the morphological parameters of the neural signal activity area and generate a preliminary three-dimensional grid model;

[0014] Construct a dynamic functional connection matrix and generate an individualized three-dimensional grid model.

[0015] Specifically, the construction of the dynamic functional connection matrix and the generation of the individualized three-dimensional grid model include:

[0016] Through phase synchrony analysis, calculate the temporal correlation of blood oxygen signals in different neural signal activity areas and construct a functional connection matrix;

[0017] Based on time-frequency analysis, extract the power changes in the task-related frequency bands, identify the task-induced neural oscillation patterns, encode the task events as time series through task-state-driven dynamic connection weight adjustment, dynamically calculate the information transfer direction and intensity between neural signal activity areas through a modulation function, and share the task modulation parameters at the population level to generate a dynamic connection weight matrix;

[0018] Integrate the structural features and functional connection strength, and analyze the structure-function coupling relationship;

[0019] According to the functional connection matrix, the dynamic connection weight matrix, and the structure-function coupling relationship, construct an individualized three-dimensional grid model, including: nodes, edges, and dynamic markers.

[0020] Specifically, the sharing of the task modulation parameters includes: task event-related parameters and modulation function parameters.

[0021] Specifically, the stimulation parameters include: current intensity, synchronization phase, stimulation frequency, waveform type.

[0022] Specifically, the simulation of the neural response of the target area with a machine learning model and the screening out of the optimal combination of stimulation parameters include:

[0023] Extract the features of the individualized three-dimensional grid model, including: static features and dynamic features, and establish a feature mapping relationship;

[0024] According to the feature mapping relationship, a neural response simulation model of the target area is constructed using numerical simulation or a neural network model based on physical modeling;

[0025] The machine learning model is used to optimize and screen the stimulation parameters to obtain the first combination of stimulation parameters;

[0026] The first combination of stimulation parameters is input into the neural response simulation model for simulation to simulate the neural activities in the target area under the first combination of stimulation parameters, and the simulation results are obtained, including: neuron excitability, neural synchrony, and therapeutic effect;

[0027] The simulation results are evaluated. According to the simulation evaluation results, the stimulation parameters are adjusted. If the simulation evaluation results do not meet the expected effect, the stimulation parameter combination is re-screened through the machine learning model for simulation until the optimal stimulation parameter combination is found.

[0028] Specifically, the machine learning model includes: CNN, SVM.

[0029] A non-invasive sleep assistance improvement system for implementing the non-invasive sleep assistance improvement method described above, including: a grid model establishment module, a signal generation module, an optimization and screening module, and a monitoring and feedback module;

[0030] The grid model establishment module is used to collect static data and dynamic data, generate a preliminary three-dimensional grid model according to the collected data, construct a functional connectivity matrix and a dynamic functional connectivity matrix, and generate an individualized three-dimensional grid model. The dynamic functional connectivity matrix encodes task events as time series, dynamically calculates the information transfer direction and intensity through a modulation function, and shares task modulation parameters at the population level;

[0031] The signal generation module is used to select a stimulation area according to the individualized three-dimensional grid model, generate an electrical stimulation signal, and determine the stimulation parameters;

[0032] The optimization and screening module is used to simulate the neural response of the target area in combination with the machine learning model, screen out the optimal stimulation parameter combination, and assist in improving the user's sleep with the optimal stimulation parameter combination.

[0033] Specifically, the grid model establishment module includes: a data collection unit, a preliminary grid model establishment unit, and a grid model establishment unit;

[0034] The data collection unit is used to collect static data and dynamic data;

[0035] The preliminary grid model establishment unit is used to calculate the morphological parameters of the neural signal activity area and generate a preliminary three-dimensional grid model according to the morphological parameters of the neural signal activity area;

[0036] The grid model establishment unit is used to construct a functional connectivity matrix and a dynamic functional connectivity matrix, and combine with the preliminary three-dimensional grid model to generate an individualized three-dimensional grid model.

[0037] Specifically, the grid model establishment unit includes: a functional connectivity subunit, a connection weight subunit, a coupling analysis subunit, and a grid model establishment subunit;

[0038] The functional connectivity subunit is used to calculate the temporal correlation of blood oxygen signals in different neural signal activity regions through phase synchrony analysis, and construct a functional connectivity matrix;

[0039] The connection weight subunit is used to adjust the dynamic connection weight driven by the task state, encode the task event as a time series, dynamically calculate the information transfer direction and intensity of the neural signal activity region through a modulation function, and share the task modulation parameters at the population level to generate a dynamic connection weight matrix;

[0040] The coupling analysis subunit is used to integrate the structural features and the functional connectivity strength, and analyze the structural-functional coupling relationship;

[0041] The grid model establishment subunit is used to construct an individualized three-dimensional grid model according to the functional connectivity matrix, the dynamic connection weight matrix, and the structural-functional coupling relationship.

[0042] Compared with the prior art, the beneficial effects of the present invention are:

[0043] 1. The present invention proposes a non-invasive sleep assistance improvement method. By constructing an individualized three-dimensional grid model, encoding and processing task events, and sharing task modulation parameters for individuals, it can more accurately capture how the connection strength between stimulation regions dynamically changes under different task conditions. And at the population level, it can improve the robustness of individualized estimation and at the same time improve the targeting accuracy.

[0044] 2. The present invention proposes a non-invasive sleep assistance improvement method. By using a machine learning model to simulate neural responses and real-time optimize stimulation parameters such as frequency, synchronous phase, intensity, waveform, etc., a personalized combination plan of stimulation parameters is formed, avoiding the one-size-fits-all method based on standard stimulation parameters in the prior art.

[0045] 3. The present invention proposes a non-invasive sleep assistance improvement method. By real-time collecting the feedback of the stimulation region and applying the neural simulation model, the stimulation parameters can be adjusted at any time to avoid over-strong or over-weak stimulation. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a flowchart of a non-invasive sleep assistance improvement method provided by the present invention;

[0047] Figure 2 The flow chart for constructing an individualized three-dimensional grid model provided by the present invention;

[0048] Figure 3 The flow chart for screening simulation of stimulation parameters provided by the present invention;

[0049] Figure 4 The architecture diagram of a non-invasive sleep assistance improvement system provided by the present invention. Detailed implementation manners

[0050] The following further elaborates on the present application with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present application, but do not limit the present application in any form. It should be noted that those of ordinary skill in the art can make several modifications and improvements without departing from the concept of the present application. These all fall within the protection scope of the present application.

[0051] In order to make the purpose, technical solutions and advantages of the present application clearer, the following further elaborates on the present application with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0052] It should be noted that if there is no conflict, the various features in the embodiments of the present application can be combined with each other, and all are within the protection scope of the present application. Additionally, although functional module division is carried out in the device schematic diagram and the logical sequence is shown in the flow chart, in some cases, the steps shown or described can be executed in a different order from the module division in the device or the sequence in the flow chart. Furthermore, the terms "first", "second", "third", etc. used in the present application do not limit the data and execution order, but only distinguish the same items or similar items with basically the same functions and effects.

[0053] The terms "first", "second", "third", etc. used in the present application do not limit the data and execution order, but only distinguish the same items or similar items with basically the same functions and effects.

[0054] Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by those skilled in the technical field to which the present application belongs. The terms used in this specification in the description of the present application are only for the purpose of describing specific implementation manners and are not used to limit the present application. The term "and / or" used in this specification includes any and all combinations of one or more of the related listed items.

[0055] Embodiment 1

[0056] Please refer to Figures 1-3 , an embodiment provided by the present invention: a non-invasive sleep assistance improvement method, including the following specific steps:

[0057] Step S1: Based on technologies such as electroencephalogram, collect the static data and dynamic data of the patient, construct an individualized three-dimensional grid model, and accurately locate the target area and its functional connection path;

[0058] High-density electroencephalogram (EEG): Use a 64-channel electrode cap with a sampling rate of 1000 Hz to record the brain electrical activity during the task state, and focus on capturing event-related potentials (ERPs) and band powers (such as alpha waves, beta waves, gamma waves);

[0059] As Figure 2 shown, the specific steps for constructing the individualized three-dimensional grid model in the step S1 are as follows:

[0060] Step S101: Construct a surface topology map of the neural signal activity area according to the collected static data and dynamic data, and identify and segment the surface topology map of the neural signal activity area, including the boundaries of key brain regions such as the prefrontal cortex, motor cortex, and hippocampus;

[0061] Step S102: Calculate the morphological parameters such as the volume, surface area, and cortical thickness of the neural signal activity area, and generate a preliminary three-dimensional grid model;

[0062] Step S103: Construct a dynamic functional connection matrix and generate an individualized three-dimensional grid model.

[0063] The specific steps of the step S103 are as follows:

[0064] Step S1031: Through phase synchrony analysis, calculate the temporal correlation of the blood oxygen signals in different neural signal activity areas, and construct a functional connection matrix;

[0065] In this embodiment, collect the blood oxygenation level-dependent (BOLD) signal data of the brain and perform preprocessing, including: time correction, head motion correction, spatial normalization, and filtering, etc. Perform time-frequency analysis on the preprocessed blood oxygen signal, convert the blood oxygen signal in the time domain to the time-frequency domain, extract the phase information of each brain region signal at different frequencies, calculate the instantaneous phase of the blood oxygen signal of each brain region at each time point and specific frequency, and for each pair of brain regions, calculate the phase synchrony index between them. The phase synchrony index includes the phase locking value (PLV), coherence, etc. According to the calculated phase synchrony index, construct a functional connection matrix, where the rows and columns of the matrix correspond to different brain regions respectively, and the matrix elements represent the functional connection strength between brain regions, that is, the value of the phase synchrony index;

[0066] Step S1032: Based on time-frequency analysis, extract the power changes in task-related frequency bands (such as theta wave 4 - 8 Hz, gamma wave 30 - 80 Hz), identify the task-induced neural oscillation patterns, encode task events as time series through task-state-driven dynamic connection weight adjustment, dynamically calculate the information transfer direction and intensity between brain regions through a modulation function, and share task modulation parameters at the population level to generate a dynamic connection weight matrix;

[0067] The shared task modulation parameters include: task time-related parameters, including stimulus start time parameters, response time parameters, etc., and modulation function parameters. The sharing method is to divide the parameters into different levels through a hierarchical Bayesian model. The task modulation parameters are divided into individual-level parameters and population-level parameters. The population-level parameters are common parameters shared by all individuals, and the individual-level parameters are related to the specific characteristics of each individual but are constrained by the population-level parameters.

[0068] In this embodiment, according to the results of power changes, classify the neural oscillation patterns in different brain regions and frequency bands. For example, if the theta wave power in a certain brain region increases significantly during task execution, it may indicate that this brain region is involved in cognitive processes such as task-related memory and attention; while the increase in gamma wave power may be related to the integration and processing speed of information;

[0069] Traditional dynamic causal modeling is based on prior knowledge and the results of neural oscillation pattern analysis. Although it can describe the effective connections between brain regions, it has certain limitations in dealing with task-related connection changes;

[0070] This application improves traditional dynamic causal modeling by encoding task-related events as time series, perceiving the process and time characteristics of the task, using a modulation function (such as the Sigmoid function) to achieve dynamic adjustment of connection weights, and sharing task modulation parameters among different individuals at the population level. Although there are individual differences, there are also some common task modulation patterns. By sharing these parameters, the information of the population can be used to provide additional constraints and support for the parameter estimation of each individual;

[0071] Benefits: 1) It can more accurately capture how the connection strength between brain regions in the brain network dynamically changes under different task conditions; 2) And at the population level, it can improve the robustness of individual estimation. For a single individual, its data will be affected by factors such as noise and measurement errors, resulting in inaccurate parameter estimation. By sharing parameters at the population level and integrating the information of the population into the individual parameter estimation, the uncertainty of individual data can be reduced, and a more reliable and stable individual parameter estimation result can be obtained.

[0072] For example: In an experiment involving multiple subjects, when different subjects perform memory tasks, the task modulation patterns between memory-related brain areas in the brain have certain similarities. When there is noise in the data of an individual, his individual parameter estimate will fluctuate greatly. However, with the help of the task modulation parameters shared by the group, his individual parameter estimate can be corrected and optimized to make it closer to the true value.

[0073] A dynamic causal model containing multiple brain region improvements is constructed. Each brain region in the model is a node. The connection between nodes represents the information transmission path between brain regions. The direction of information transmission between brain regions is determined according to the positive and negative values of the connection parameters in the model. A positive connection parameter indicates that the previous brain region has a promoting effect on the next brain region (information is transmitted from the former to the latter), while a negative connection parameter indicates an inhibitory effect. The absolute value of the connection parameter reflects the intensity of information transmission. The larger the absolute value, the stronger the information transmission between brain regions. With brain regions as rows and columns, a dynamic connection weight matrix is constructed according to the information transmission intensity between brain regions calculated by the model. The elements in the matrix represent the connection weights between corresponding brain regions, that is, the information transmission intensity. Since neural activity changes dynamically over time, the above analysis process is repeated at different time points to update the connection weight matrix, thereby obtaining a connection weight matrix sequence that reflects the dynamic changes of the brain network.

[0074] Step S1033: Integrate structural features (such as the volume of the neural signal activity area) and functional connection strength to analyze the structure-function coupling relationship;

[0075] In this embodiment, the structure-function coupling relationship is analyzed by an improved Bayesian network model for target priority sorting. In brain network research, the structural characteristic variables are: connection strength, connection length and brain volume between brain regions, etc. The functional connection strength is obtained by calculating the functional connection matrix and the dynamic connection weight matrix;

[0076] Step S1034: Based on the functional connectivity matrix, the dynamic connectivity weight matrix and the structure-function coupling relationship, an individualized three-dimensional grid model is constructed, which includes: nodes, edges and dynamic markers. The nodes are the segmented brain regions, and the node size indicates their functional connectivity; the edges are the connection strengths between brain regions, and the edge width and color gradient (such as red-strong, blue-weak) indicate the connection weights; the dynamic markers indicate that the brain regions that are significantly activated in the task state are highlighted with flashing.

[0077] In this embodiment, the traditional brain network topology map only describes the functional connections. The present application quantifies the interactive relationship between the morphological characteristics of brain regions and the functional connections through a Bayesian network, providing a dual basis for the selection of individualized stimulation targets.

[0078] Step S2: selecting a stimulation area, generating an electrical stimulation signal, and determining stimulation parameters according to the individualized three-dimensional grid model;

[0079] The stimulation parameters in step S2 include: current intensity, synchronization phase, stimulation frequency, and waveform type;

[0080] The current intensity directly determines the intensity of the stimulation and its impact on neural activity. For transcranial alternating current stimulation (tACS), the current intensity is usually between 1 mA and 3 mA;

[0081] The stimulation frequency determines the activation mode and pattern of neurons. The stimulation frequencies are divided into: low-frequency stimulation (1 Hz - 10 Hz), which is often used to promote neural plasticity or increase the activation degree of brain regions, and is suitable for the treatment of chronic neurological diseases such as depression and cognitive impairment; high-frequency stimulation (20 Hz - 100 Hz and above), which is used to enhance the synchronous activity of neurons and is suitable for the treatment of movement disorders (such as Parkinson's disease) or the improvement of anxiety symptoms; and frequency modulation, which uses a mixed stimulation of specific frequencies. For example, the combination of low-frequency and high-frequency stimulation can produce a more complex neural response pattern and is used for the multi-dimensional regulation of symptom improvement;

[0082] The waveform types are divided into: sine wave, which regulates the neural activity of brain regions through a stable current change and is used to enhance neural plasticity; square wave, which produces a stronger pulse effect and triggers more significant neural excitation, and is suitable for treatments that require rapid and short-term effects; triangular wave, which has a gradual change effect, can slowly activate neurons and effectively avoid over-stimulation, and is suitable for the treatment of some chronic diseases such as depression or chronic anxiety;

[0083] Step S3: Combine a machine learning model to simulate the neural response of the target area, screen the optimal combination of stimulation parameters, and use the optimal combination of stimulation parameters to assist in improving the user's sleep;

[0084] As Figure 3 shown, the specific steps of step S3 are:

[0085] Step S301: Extract the features of the individualized three-dimensional grid model, including: static features and dynamic features, and establish a feature mapping relationship;

[0086] Specifically, the static features are the features of the functional connection matrix, including: structural connection features, node attribute features, and network topology features, etc.; the dynamic features are the features of the dynamic connection weight matrix, including: the time-varying characteristics of functional connections and state transition features, etc.; the disease labels include: clinical diagnosis information and biomarker information; for high-dimensional features, dimensionality reduction processing is required, calculate the correlation between brain network features and disease labels, screen out the features significantly related to the disease, and then establish a disease-brain network feature mapping model through a machine learning algorithm to obtain the feature mapping relationship;

[0087] Step S302: According to the feature mapping relationship, use numerical simulation or a neural response simulation model based on physical modeling (such as a model based on biological neurons, a population neuron model, etc.) to construct a neural response simulation model for the target area;

[0088] The core of the model is to be able to reflect the influence of different stimulation parameters (frequency, intensity, waveform, etc.) on neuron activity and simulate the excitation and inhibition processes of neurons;

[0089] Step S303: Use a machine learning model to optimize and screen the stimulation parameters to obtain the first stimulation parameter combination;

[0090] The machine learning model in Step S303 includes: CNN, SVM, etc.;

[0091] Step S304: Input the first stimulation parameter combination into the neural response simulation model for simulation, simulate the neural activity in the target brain area under the first stimulation parameter combination, and obtain the simulation results, including: neuron excitation degree, neural synchrony, and therapeutic effect;

[0092] Step S305: Evaluate the simulation results. According to the simulation evaluation results, further adjust the stimulation parameters. If the simulation evaluation results do not meet the expected effect, return to Step S303, and re-screen the stimulation parameter combination through the machine learning model until the optimal stimulation parameter combination is found.

[0093] Embodiment 2

[0094] Please refer to Figure 4 , another embodiment provided by the present invention: A non-invasive sleep assistance improvement system, including: a grid model establishment module, a signal generation module, an optimization and screening module, and a monitoring and feedback module;

[0095] The grid model establishment module is used to collect static data and dynamic data, generate a preliminary three-dimensional grid model according to the collected data, construct a functional connection matrix and a dynamic functional connection matrix, generate an individualized three-dimensional grid model. The dynamic functional connection matrix encodes task events as time series, dynamically calculates the information transmission direction and intensity through modulation functions, and shares task modulation parameters at the population level;

[0096] The signal generation module is used to select a stimulation area according to the individualized three-dimensional grid model, generate an electrical stimulation signal, and determine the stimulation parameters;

[0097] The optimization and screening module is used to combine a machine learning model to simulate the neural response of the target area, screen out the optimal stimulation parameter combination, and use the optimal stimulation parameter combination to assist in improving the user's sleep.

[0098] The grid model establishment module includes: a data acquisition unit, a preliminary grid model establishment unit, and a grid model establishment unit;

[0099] The data acquisition unit is used to acquire static data and dynamic data;

[0100] The preliminary grid model establishment unit is used to calculate the morphological parameters of the neural signal activity region and generate a preliminary three-dimensional grid model according to the morphological parameters of the neural signal activity region;

[0101] The grid model establishment unit is used to construct a functional connection matrix and a dynamic functional connection matrix, and combine with the preliminary three-dimensional grid model to generate an individualized three-dimensional grid model.

[0102] The grid model establishment unit includes: a functional connection subunit, a connection weight subunit, a coupling analysis subunit, and a grid model establishment subunit;

[0103] The functional connection subunit is used to calculate the temporal correlation of the blood oxygen signals in different neural signal activity regions through phase synchrony analysis and construct a functional connection matrix;

[0104] The connection weight subunit is used to adjust the dynamic connection weight driven by the task state, encode the task event as a time series, dynamically calculate the information transfer direction and intensity of the neural signal activity region through a modulation function, and share the task modulation parameters at the population level to generate a dynamic connection weight matrix;

[0105] The coupling analysis subunit is used to integrate the structural features and the functional connection strength and analyze the structure-function coupling relationship;

[0106] The grid model establishment subunit is used to construct an individualized three-dimensional grid model according to the functional connection matrix, the dynamic connection weight matrix, and the structure-function coupling relationship.

[0107] In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the corresponding technical solutions in the prior art in terms of implementation principles are not described in detail to avoid excessive elaboration.

[0108] As described above in the specific embodiments, the purpose, technical solutions, and beneficial effects of the present invention have been further described in detail. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A non-invasive sleep assistance improvement system, characterized in that, Including: a grid model establishment module, a signal generation module, an optimization and screening module, and a monitoring and feedback module; The grid model establishment module collects static data and dynamic data based on electroencephalogram technology, generates a preliminary three-dimensional grid model according to the collected data, constructs a functional connectivity matrix and a dynamic functional connectivity matrix, and generates an individualized three-dimensional grid model. The dynamic functional connectivity matrix encodes task events as time series, dynamically calculates the information transfer direction and intensity through a modulation function, and shares task modulation parameters at the population level; The signal generation module is used to select a stimulation area according to the individualized three-dimensional grid model, generate an electrical stimulation signal, and determine stimulation parameters; The optimization and screening module is used to simulate the neural response of the target area by combining a machine learning model and screen out the optimal combination of stimulation parameters; The grid model establishment module includes: a data acquisition unit, a preliminary grid model establishment unit, and a grid model establishment unit; The data acquisition unit collects static data and dynamic data based on electroencephalogram technology; The preliminary grid model establishment unit is used to calculate the morphological parameters of the neural signal activity area and generate a preliminary three-dimensional grid model according to the morphological parameters of the neural signal activity area; The grid model establishment unit includes: a functional connectivity subunit, a connection weight subunit, a coupling analysis subunit, and a grid model establishment subunit; The functional connectivity subunit is used to calculate the temporal correlation of blood oxygen signals in different neural signal activity areas through phase synchrony analysis and construct a functional connectivity matrix; The connection weight subunit, based on time-frequency analysis, extracts the power changes in the task-related frequency band, identifies the neural oscillation patterns induced by the task, encodes the task events as time series through task-state-driven dynamic connection weight adjustment, dynamically calculates the information transfer direction and intensity between neural signal activity areas through a modulation function, and shares task modulation parameters at the population level to generate a dynamic connection weight matrix; The coupling analysis subunit is used to integrate the structural features and functional connection strength and analyze the structure-function coupling relationship; The grid model establishment subunit is used to construct an individualized three-dimensional grid model according to the functional connectivity matrix, the dynamic connection weight matrix, and the structure-function coupling relationship.

2. The non-invasive sleep assistance improvement system according to claim 1, wherein, The individualized three-dimensional grid model includes: nodes, edges, and dynamic markers.

3. The non-invasive sleep assistance improvement system according to claim 2, wherein, The shared task modulation parameters include: task event-related parameters and modulation function parameters.

4. The non-invasive sleep assistance improvement system according to claim 3, wherein The stimulation parameters include: current intensity, synchronous phase, stimulation frequency, waveform type.

5. The non-invasive sleep assistance improvement system according to claim 4, wherein The simulating the neural response of the target area by combining a machine learning model and screening out the optimal combination of stimulation parameters includes: extracting the features of the individualized three-dimensional grid model, including: static features and dynamic features, and establishing a feature mapping relationship; according to the feature mapping relationship, constructing a neural response simulation model of the target area by using numerical simulation or a neural network model based on physical modeling; using a machine learning model to optimize and screen the stimulation parameters to obtain a first combination of stimulation parameters, and the machine learning model includes: CNN, SVM; Input the first set of stimulation parameters into the neural response simulation model for simulation, simulate the neural activities in the target area under the first set of stimulation parameters, and obtain the simulation results, including: neuron excitation level, neural synchrony, and therapeutic effect; Evaluate the simulation results, adjust the stimulation parameters according to the simulation evaluation results. If the simulation evaluation results do not meet the expected effect, re-screen the set of stimulation parameters through a machine learning model and conduct simulation until the optimal set of stimulation parameters is found.

Images