Application method of adaptive neural network in individualized neuromodulation

Abstract

The application relates to an application method of an adaptive neural network in individualized neural regulation; a neural system representation model of an individual is constructed based on multi-source data including electroencephalogram (EEG), functional magnetic resonance imaging (fMRI), genetic information and psychological evaluation; key features including a neural connection mode and synchronization in the neural system of the individual are mined through a machine learning technology including a graph neural network, and a neural activity basic atlas for each individual is constructed; an adaptive regulation path planning method is introduced according to the constructed individualized neural representation model; a stimulation path for producing a regulation effect on the neural system is calculated by using the change trend of the neural activity data of the individual in space and time; the path planning is stimulation of the neural signal, and the time window, the intensity and the frequency of the stimulation are calculated based on the connectivity interaction of each functional area of the brain, so that dynamic regulation of the neural activity of the individual is realized; an adaptive activation mechanism based on neural feedback is introduced.

Description

TECHNICAL FIELD

[0001] The application relates to an application method in individualized neuromodulation, in particular to an application method of an adaptive neural network in individualized neuromodulation. BACKGROUND

[0002] At present, although the application method of an adaptive neural network in individualized neuromodulation has made significant progress, it still faces many challenges and deficiencies in the actual application process, especially in comparison with the existing technology, such as the individualized neuromodulation parameter optimization method based on a tDCS stimulation response model mentioned in Chinese invention patent 2023110055781. First, in the existing adaptive neural network method, the joint modeling of neural activity signals and psychological state data can reflect the individual differences of the nervous system, but in the fusion of multi-source data and the extraction of deep coupling features, it is often limited by the heterogeneity and diversity of data. Especially when dealing with complex neural signals such as EEG and fMRI, the network model needs to efficiently encode and decode the temporal and spatial features of these signals, which is extremely high for the algorithm. However, the current neural network architecture has not been able to fully and effectively integrate these signals and extract accurate, physiologically and psychologically meaningful features. Secondly, the adaptive neural network also faces limitations in the planning and optimization of the neuromodulation path. Traditional path planning methods often rely on functional connectivity between brain regions to optimize the distribution of stimulation, which requires accurate modeling of the functional state of each region.

[0003] However, the structure of the nervous system varies greatly among individuals, making it difficult for path planning based on a general brain connection model to perfectly adapt to all individuals. This leads to insufficient personalization of neuromodulation programs, especially when treating different patients, and cannot effectively respond to the dynamic changes in the nervous system and the differences in neural function. In addition, existing adaptive adjustment mechanisms, especially when dynamically adjusting the path of neural stimulation, have the problem of high computational resource consumption and poor real-time performance. Although adjusting through real-time data feedback can theoretically improve treatment effectiveness, in practical applications, the spatiotemporal changes in neural activity are very complex, and the efficiency and accuracy of existing computational models in real-time processing and adjusting neural stimulation programs are still not high, making it difficult to accurately capture the subtle changes in brain activity in a short period of time, resulting in a lag in the adjustment of the neural stimulation program and affecting the treatment effectiveness. Unlike this, the individualized neuromodulation method based on the tDCS stimulation response model can simulate different neuromodulation programs through finite element modeling and predict the response of the whole brain neural activity through large-scale dynamics modeling, effectively overcoming the ethical constraints and heterogeneity of neuromodulation programs. This method builds an individualized tDCS stimulation response model, optimizes the current dose parameters for each patient's individual differences, reduces uncertainty in treatment, and improves the consistency of treatment effectiveness. In contrast, existing adaptive neural network methods rely too much on the training data of the model when faced with complex physiological and psychological multi-source data, and cannot completely eliminate the heterogeneity and individual differences between samples, and cannot be simulated at a large scale, limiting the applicability and accuracy of the treatment program. In addition, individualized neuromodulation also faces ethical and safety issues in actual operation, especially in the clinical verification stage.

[0004] Existing neural network models often cannot be widely used in clinical applications without sufficient verification, which not only limits their promotion but also may pose a safety hazard to patients. The individualized neuromodulation method based on tDCS can reduce the impact of ethical constraints and perform safety verification in a virtual environment, providing strong support for subsequent clinical trials. Furthermore, existing adaptive neural network methods often rely on real-time neural activity data of individuals for neural system regulation, but in many cases, the psychological state and emotional fluctuations of patients can have a profound impact on neural activity. Although the present invention attempts to introduce psychological data such as emotional fluctuations and stress levels for deep coupling modeling, due to the complexity and variability of emotions and psychological states, it is still a difficult problem to accurately extract features that match neural activity from these non-physiological signals. SUMMARY

[0005] The application aims to provide an application method of adaptive neural network in individualized neuromodulation, thereby solving some of the problems and deficiencies pointed out in the background art.

[0006] The application solves the above technical problems by adopting the following technical scheme comprising the following steps:

[0007] S1, construction of an individualized nervous system representation model:

[0008] S1.1, based on multi-source data including electroencephalogram (EEG), functional magnetic resonance imaging (fMRI), genetic information, and psychological assessment, construct an individual's nervous system representation model;

[0009] S1.1, by machine learning technology including graph neural network, mine key features including neural connection pattern and synchronicity in individual nervous system, construct a neural activity base map for each individual;

[0010] S2, adaptive regulation path planning based on the neural representation model:

[0011] S2.1, according to the constructed individualized neural representation model, introduce an adaptive regulation path planning method; use the change trend of individual neural activity data in space and time to calculate the stimulation path that produces a regulation effect on the nervous system;

[0012] S2.2, path planning is the stimulation of neural signals, based on the connectivity interaction of each functional area of the brain, calculate the time window, intensity and frequency of stimulation, realize the dynamic regulation of individual neural activity;

[0013] S3, adaptive activation mechanism and dynamic regulation of nervous system:

[0014] S3.1, introduce an adaptive activation mechanism based on neural feedback; the adaptive activation mechanism monitors the real-time individual neural activity, and calculates the potential activation nodes and connections in the neural network through the dynamic changes of neural activity;

[0015] S3.2, by adjusting the stimulation parameters in real time, optimize the activation mode of the nervous system, so that the neural network adjusts itself and adapts to the changes of the external environment at different periods;

[0016] S4, combination of individualized neuromodulation and plasticity mechanism:

[0017] S4.1, use the adaptive neural network model to evaluate the plasticity of individual neural network in real time, and combine the activation and feedback mechanism of the nervous system to design a regulation strategy to promote the development of neural plasticity;

[0018] S5, deep coupling of neuromodulation and individual psychological state:

[0019] S5.1, develop a dynamic regulation method based on the coupling of neural activity and psychological state; model the neural signal and the psychological state including emotional fluctuations, stress levels through a deep learning model including a variational autoencoder, and adjust the neural regulation scheme according to the real-time changes of the individual psychological state.

[0020] Further, the adaptive regulation path planning method based on the neural representation model comprises:

[0021] By combining multi-source physiological data including EEG, fMRI, genomics and psychological assessment data, a neural activity feature model of each individual is constructed; the neural activity feature model reflects the structure of the nervous system and describes the dynamic changes of neural function, including the connectivity and synchronicity of neural networks; the individualized neural representation model is represented as:

[0022]

[0023] Wherein:

[0024] N model (t) represents the output of the individualized neural representation model at time t, representing the comprehensive performance of individual neural activity; f i (t) is the i-th input feature, which refers to the time series data extracted from different physiological data sources including EEG signals, fMRI image data, reflecting the characteristics of neural activity; w i is the weight coefficient of the i-th feature, indicating the importance of the feature to the output of the neural representation model; b is the bias term, adjusting the basic output of the model; n is the total number of input features, representing the dimension of the model input features.

[0025] Further, the adaptive regulation path planning method based on the neural representation model comprises:

[0026] According to the functional connectivity of different regions of the nervous system including the cerebral cortex and the basal ganglia, the stimulation path planning is carried out; the degree of functional connectivity of each region to other regions determines the contribution to the overall neural activity, and the stimulation path is allocated according to the connectivity; the formula for spatial optimization is:

[0027]

[0028] Wherein:

[0029] P spatial represents the optimized stimulation path, reflecting the spatial distribution of stimulation; is the spatial range of the brain region, representing the set of regions of the nervous system that need to be stimulated; S j(x, t) is the intensity of neural activity at spatial location x and time t, reflecting the state of neural activity in a specific region; a j is the weight coefficient of the jth region, indicating the contribution weight of the region in the overall neural activity; m is the number of neural regions that need to be regulated, indicating the number of neural regions considered; dx is the spatial integral variable, indicating the integration process of the brain spatial region.

[0030] In terms of time optimization, a stimulation time window based on individual neural activity time series is designed; the stimulation timing within the dynamic time window is selected by weighted sum of the timing characteristics of neural activity and maximization of sum; the formula for time optimization is:

[0031]

[0032] where:

[0033] P temporal (t) is the optimal stimulation path at time t, indicating the optimized stimulation scheme for the current time period; E i (t) is the time series feature of the ith neural activity, indicating the fluctuation of neural activity at time t; r i is the weight coefficient of the ith neural activity feature, indicating the influence degree of the feature on the time optimization path; n is the number of neural activities involved in regulation, indicating the dimension of the time series data considered.

[0034] Further, the adaptive regulation path planning method based on the neural representation model comprises:

[0035] Real-time monitoring and dynamic adjustment mechanism is adopted; an adaptive adjustment mechanism based on real-time data is designed, which dynamically adjusts the stimulation path according to the changes of neural activity; the real-time adjustment formula is:

[0036]

[0037] where: ΔP(t) represents the stimulation path adjustment amount at time t, indicating the stimulation direction and amplitude that need to be adjusted; γ is the adjustment coefficient, controlling the sensitivity of the adjustment amount; N model (t) is the neural activity value predicted by the individualized neural representation model, indicating the expected state of neural system activity; N current (t) is the current neural activity value obtained by real-time monitoring device, indicating the actual state of neural system activity; is the gradient operator, indicating the gradient calculation of the change of difference N model (t)N current (t).

[0038] Further, the deep coupling method of neural regulation and individual psychological state comprises:

[0039] Firstly, the neural activity signal and the psychological state data are preprocessed by the encoder part of the deep learning model; the encoder part fuses the multi-modal data into a low-dimensional representation through a nonlinear mapping, forming the coupling features of neural activity and psychological state; then the coupling features are restored by the decoder part to generate the corresponding regulatory output signal of neural activity, and adjust the neural stimulation scheme according to the current psychological state; the coupling mapping expression formula of the autoencoder is:

[0040]

[0041] Where: Z N is a low-dimensional coupling representation of neural activity and psychological state; through the encoder part of the autoencoder model, the neural activity signal and the psychological state data are mapped to a shared low-dimensional feature space, forming the coupling representation of neural and psychological state; X N is the input neural activity signal including EEG, fMRI; X P is the input psychological state data; the state data is obtained through physiological or psychological evaluation; Θ E is the weight and bias of the encoder network, which is the parameter adjusted in the network learning process; is a nonlinear mapping function used to convert neural activity and psychological state signals into a low-dimensional coupling representation.

[0042] Further, the deep coupling method of neural regulation and individual psychological state comprises:

[0043] By introducing the autoencoder, the neural activity and the psychological state are jointly modeled, and the interaction between the two is adaptively learned and recognized under the deep learning framework; the structure of the autoencoder includes an encoder, a decoder and a joint layer; the encoder maps the input neural activity signal and psychological state data to a low-dimensional representation through a nonlinear activation function; the features of neural activity and psychological state in the hidden space are learned and fused; and the joint modeling and decoding mapping expression is:

[0044] Z N = W N · X N + W P · X P + b

[0045] Where: Z N is a low-dimensional coupling representation of neural activity and psychological state; after the calculation of the encoder layer, the low-dimensional representation contains the comprehensive features of neural activity and psychological state, which is transmitted as input to the subsequent decoder part; W N is the weight matrix of the neural activity signal, which controls the influence degree of the neural activity signal; W Pis a weight matrix of the psychological state data, controlling the contribution degree of the psychological state data in the model; X N is the input neural activity signal; X P is the input psychological state data; b is a bias term for shifting the output.

[0046] Further, the method for deeply coupling the individual psychological state and the neural regulation comprises:

[0047] Real-time psychological state monitoring is introduced, and the coupling features of neural activity and psychological state extracted by the variational autoencoder are combined to dynamically adjust the neural regulation scheme; while the real-time psychological state monitoring includes real-time collection of emotional fluctuation and stress level data; the data is input into a deep learning model, and the change trend of the neural activity signal is combined to generate a neural stimulation adjustment scheme:

[0048]

[0049] Where, ΔS(t) is the adjustment amount of the neural stimulation scheme at time t; represents the real-time adjustment of the neural regulation scheme according to the current neural activity and psychological state; S0 is the initial neural stimulation scheme, which is set at the beginning of treatment, including the frequency and intensity of stimulation; P(t) is the psychological state data at time t, including emotional fluctuation and stress level; Θ A is a parameter of the regulation model, which is optimized through the training of the deep learning model to control the adjustment mode of the stimulation scheme; is a nonlinear regulation function, which is used to adjust the neural stimulation scheme according to the coupling features of neural activity and psychological state.

[0050] The application method of the adaptive neural network in individualized neural regulation has the following beneficial effects:

[0051] By combining various physiological data (such as EEG, fMRI, genomics data) and psychological state evaluation data, a unique neural activity feature model for each individual is constructed, which can accurately represent the structure and function of the nervous system of each patient. Based on these individualized models, the neural regulation path can be dynamically adjusted according to the specific neural activity pattern, psychological state and environmental changes, so as to realize precise treatment.

[0052] An adaptive regulation mechanism based on real-time neural activity and psychological state data is introduced. By real-time monitoring of the change trend of neural activity and the changes of emotional fluctuation, stress level and other psychological states, the parameters such as the path, frequency and intensity of neural stimulation can be adjusted in time, so as to better cope with the dynamic changes of individual nervous system and psychological state. This real-time response capability improves the flexibility and adaptability of treatment, significantly optimizing the treatment effect.

[0053] By optimizing the activation mechanisms and dynamic regulation of the nervous system through adaptive neural networks, we can effectively promote the plasticity of the nervous system, enabling the brain to self-adjust and adapt to changes in the external environment at different times. This has important clinical implications for rehabilitation therapy, improving cognitive function, and repairing nerve damage, helping patients gradually recover and enhance their neurological function. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1 This is a flow chart of the application method of the adaptive neural network in individualized neural regulation of the present invention.

[0055] Figure 2 This is a flow chart of the adaptive control path planning method based on the neural representation model of the present invention.

[0056] Figure 3 This is a flow chart of the deep coupling method of neural regulation and individual psychological state of the present invention. DETAILED DESCRIPTION

[0057] The following is a detailed description of the specific embodiments of the present invention with reference to the accompanying drawings.

[0058] Combine Figure 1 The flowchart of the adaptive neural network in personalized neural regulation begins with the construction of a personalized neural system representation model. The core of this model is to fuse multi-source data to accurately depict the unique neural system characteristics of each individual. The construction of a personalized neural system representation model relies on different types of physiological and psychological data, especially electroencephalogram (EEG), functional magnetic resonance imaging (fMRI), genetic information, and psychological assessment data. EEG can capture the brain's electrical activity signals, reflecting the characteristics of neural activity in the cerebral cortex; fMRI can provide spatial distribution information of brain activity by measuring changes in brain blood oxygen levels, helping to depict the functional connections and collaborative working patterns of different brain regions; genetic information provides the potential genetic basis of an individual's neural activity, which is important for understanding differences in individual neural function; and psychological assessment helps collect data on the individual's psychological state in terms of emotions, cognition, behavior, etc. These data comprehensively reflect the individual's neural characteristics and psychological state from multiple dimensions, thereby providing accurate input information for subsequent personalized neural regulation.

[0059] In the construction of the nervous system representation model, the core technology is to use graph neural networks (GNN) for individual nervous system feature extraction. Graph neural networks are a special type of deep learning model that can handle graph-structured data and are widely used in social networks, recommendation systems, and other fields. In this technology, individual neural activity data is converted into a graph structure, where the nodes of the graph represent different regions of the brain (such as the cerebral cortex, basal ganglia, thalamus, etc.), and the edges represent the functional connection relationships between these regions. Through the multi-layer propagation mechanism of graph neural networks, the connection patterns and interaction relationships between these neural regions can be effectively mined, thereby constructing the neural connection network of the individual brain. The node features of each neural region not only include the neural activity intensity of that region, but also include the coordinated activity patterns of that region with other regions. During training, the graph neural network learns the interactions between different brain regions and extracts key information such as neural activity synchronization, functional coupling, and dynamic change patterns, thereby constructing a basic atlas of individual neural activity.

[0060] This atlas is not fixed, but dynamically adjusted, as neural activity has significant individual differences. By integrating EEG, fMRI, and other multi-source data, the graph neural network can fully consider the spatial and temporal differences of individual neural activity and generate a high-precision, personalized nervous system representation model. This model will provide precise neural basis atlases for the optimization of neural regulation schemes, avoiding the drawbacks of using "one-size-fits-all" models in traditional methods.

[0061] According to the individualized neural representation model constructed in the early stage, the neural stimulation path is designed and adjusted, thereby realizing the accurate dynamic regulation of individual neural activity. First, the neural representation model provides a personalized brain neural activity atlas for each individual, including important information such as the neural connection patterns, synchronization, and interaction of different brain regions. Based on this model, the spatial and temporal trends of individual neural activity can be further analyzed, i.e., how the activities of different brain regions fluctuate with the individual's psychological state or environmental changes at a specific time point and spatial location. In this way, the neural regulation path planning can flexibly adapt to the changes in individual brain activity and perform precise regulation. Specifically, the core of path planning lies in the design of the stimulation path, which involves not only the connectivity between different functional regions of the brain, but also factors such as the timeliness, intensity, and frequency of stimulation. First, according to the neural activity data provided by the individualized neural representation model, the connectivity between different functional regions of the brain is analyzed, and the coordinated action patterns of different regions are identified.

[0062] The connectivity between different brain regions varies, with some regions being the main signal processing centers and others playing a supporting role. Based on these connection patterns, the stimulation pathways of the nervous system in different spatial regions can be calculated, which determine the transmission route and intensity of the stimulation signal, ensuring that the stimulation signal can effectively act on the target region and exert the maximum regulatory effect through the synergy between brain regions. Secondly, the stimulation pathway is not only a spatial plan, but also includes the planning of the time window, intensity, and frequency of stimulation. The time window determines the timing and duration of the application of the stimulation signal, while the intensity and frequency directly affect the strength and persistence of the stimulation effect. For example, some neural regulation methods require the application of stimulation during a specific activity cycle of the brain to maximize their effect, which requires the regulation pathway to be flexible in response to the temporal fluctuations of neural activity. Therefore, in adaptive regulation pathway planning, the system adjusts the stimulation time window according to real-time changes in neural activity to ensure that the stimulation signal intervenes in the brain region at the most appropriate time. In addition, the adjustment of intensity and frequency is dynamically changed according to the activity intensity of the brain and the needs of the regulation target.

[0063] By monitoring the individual's neural activity in real time and making precise interventions on the nervous system according to its dynamic changes. This method first introduces an adaptive activation mechanism based on neural feedback, aiming to dynamically calculate the potential activation nodes and connections in the neural network by continuously monitoring the changes in the individual's neural activity. The dynamic changes of neural activity include electroencephalogram (EEG) signals, functional magnetic resonance imaging (fMRI) signals, and other data that can reflect the activity state of the brain at a specific time point and situation in real time. Through real-time feedback of these data, the system can identify the activity changes and potential activation nodes of key regions in the neural network. The "activation nodes" in the neural network refer to regions that exhibit significant neural activity at a certain moment, and these nodes are often activated in specific cognitive tasks or emotional states, and their activity plays a crucial role in the overall neural function of the individual.

[0064] These nodes and their connections can respond rapidly to external environments or internal stimuli, thereby changing the overall state of the neural network. By adjusting stimulation parameters such as frequency, intensity, duration, etc. in real time, the system can precisely optimize the activation pattern of the nervous system, ensuring that the neural network can efficiently and adaptively adjust in the face of different demands and environmental changes. For example, when the emotional state of an individual changes, the system can alleviate or enhance specific neural functions by adjusting the activation level of specific brain regions, optimizing the individual's emotional regulation or cognitive function. At the same time, the system can also adjust the spatial distribution of the brain, optimizing the coordination between different regions of the brain, so that the activation pattern of the entire nervous system is more reasonable and efficient. The optimization and dynamic adjustment of the activation pattern is an adaptive process, meaning that the nervous system can flexibly adjust to external stimuli and internal needs in different periods and situations.

[0065] By evaluating the plasticity of an individual's neural network in real time, combined with the activation and feedback mechanisms of the nervous system, a control strategy is designed to promote the development of neural plasticity. In this process, it is first necessary to understand that neural plasticity refers to the ability of the brain to change its structure and function after experiencing external stimuli or internal demand changes, including synaptic plasticity, neural circuit reorganization, and changes in connections between neurons. Traditional methods of neural regulation often focus on symptom relief or short-term effects, lacking consideration of long-term plasticity and self-adjustment mechanisms of the nervous system.

[0066] The introduction of the adaptive neural network model allows neural regulation to go beyond short-term intervention and consider the long-term adaptability and plasticity development of the nervous system. Specifically, the model monitors the activity state of the nervous system in real time, identifies areas in the neural circuit that need to be adjusted, and designs appropriate stimulation strategies based on the neural activity patterns of these areas. Through dynamic feedback to the neural network, the system can identify the potential plasticity of neural activity and adjust stimulation parameters at the appropriate time to promote long-term plasticity changes in the nervous system. This process not only intervenes in the current neural activity, but also guides the neural network to produce structural and functional adaptive changes after experiencing a certain amount of stimulation through the feedback mechanism of the system. For example, by adjusting the activation pattern of specific brain regions, stimulating synaptic strengthening or reorganization, the nervous system forms new information processing pathways. These adjustments not only improve the short-term effects of neural regulation, but also promote long-term healthy regulation of the nervous system.

[0067] Through a deep learning model including a variational autoencoder, neural signals are jointly modeled with an individual's mental state (e.g., emotional fluctuations, stress levels, etc.), aiming to capture the complex interrelationships between neural activity and mental state. In this process, neural signals (such as electroencephalogram EEG, functional magnetic resonance imaging fMRI, etc.) and mental state data (such as emotional fluctuations, stress levels, emotional labels, etc.) are input into the deep learning model, and through the training of the model, the intrinsic relationship between neural activity and mental state is automatically extracted. The role of the variational autoencoder is to convert the original complex data (neural signals and mental state data) into a highly compressed, representative feature space by learning a low-dimensional latent space, thereby realizing the deep coupling of neural activity and mental state.

[0068] This coupling model can accurately capture how neural activity is affected by changes in an individual's mental state, and can in turn reveal how mental state is manifested through changes in neural activity. For example, when an individual experiences emotional fluctuations, changes in neural activity will be reflected in specific regions of the brain, such as increased or decreased activity in the emotional regulation region, which in turn affects the functional state of the entire nervous system. Through this deep coupling modeling method, not only can the interaction between neural activity and mental state be comprehensively understood, but also more personalized inputs can be provided for neural regulation. Based on this coupling model, the system can monitor the individual's mental state changes in real time and adjust the neural regulation scheme according to the real-time data of the mental state. Specifically, when the individual's emotional fluctuations or stress levels change, the system will automatically adjust the stimulation parameters according to the neural-mental coupling characteristics. This adjustment not only includes the time window, intensity, frequency, etc. of the stimulation, but also involves changes in the target region of the stimulation and the stimulation pattern.

[0069] Embodiment 1:

[0070] In combination with the flowchart, Figure 2 A certain patient, named Xiao Zhang, is experiencing a stressful work environment and has mild anxiety and emotional fluctuations. In order to improve his nervous system function and alleviate emotional fluctuations, an individualized neural regulation scheme based on adaptive neural network is designed. The core of this scheme is to accurately depict his neural activity characteristics by constructing an individualized neural representation model, and to implement adaptive regulation path planning based on this.

[0071] First, a comprehensive assessment of Xiao Zhang's nervous system was conducted using multi-source physiological data. Specifically, his electroencephalogram (EEG), functional magnetic resonance imaging (fMRI), and genomic data were collected. Additionally, Xiao Zhang underwent a psychological assessment to quantify his stress levels and emotional fluctuations. Together, all this data helped build an individualized neural representation model. The goal of this model was to combine structural features of the nervous system and its dynamic changes, reflecting functional characteristics such as connectivity and synchronicity.

[0072] Next, the individualized neural representation model for Xiao Zhang was constructed using the following formula:

[0073]

[0074] where N model (t) represents the output of the individualized neural representation model at time t, reflecting the overall performance of Xiao Zhang's neural activity. f i (t) is the i-th input feature, representing time series data extracted from physiological data sources such as EEG, fMRI, etc., reflecting the characteristics of neural activity. For example, specific frequency bands in the EEG signal (such as alpha waves, beta waves, etc.) can indicate Xiao Zhang's emotional and cognitive state, while fMRI image data reveals the activation of specific brain regions.

[0075] w i is the weight coefficient of the i-th feature, indicating the importance of that feature to the output of the neural representation model. Suppose after data training, the weight coefficients of these features are obtained. For example, w1 of EEG is 0.4, w2 of fMRI is 0.3, w3 of genomic data is 0.2, and w4 of psychological assessment data is 0.1. b is the bias term, used to adjust the base output of the model to adapt to the baseline neural activity characteristics of Xiao Zhang's individual. n is the total number of input features, which is determined by the dimension of the data. In the case of Xiao Zhang, suppose four main input features are used (EEG signal, fMRI data, genomic data, and psychological assessment data), then n = 4.

[0076] Example of model output calculation:

[0077] Suppose at time t, Xiao Zhang's EEG signal shows some fluctuations, fMRI images show activation in certain regions (such as the prefrontal cortex), genomic data shows that genes related to emotional regulation are active, and psychological assessment shows that he is currently at a moderate stress level. Based on these features, the calculation process of the model output is as follows:

[0078] N model (t) = w1·f1(t) + w2·f2(t) + w3·f3(t) + w4·f4(t) + b

[0079] Assuming the following values are obtained:

[0080] f1(t) = 0.8 (Eigenvalue of EEG signal, reflecting fluctuations in neural activity)

[0081] f2(t) = 1.2 (fMRI data, showing activation intensity in the prefrontal cortex)

[0082] f3(t) = 0.5 (Genomics data feature, activity level of relevant genes)

[0083] f4(t) = 0.7 (Psychological assessment results, stress level score)

[0084] Substituting the above values and weight coefficients, assuming b = 0.1:

[0085] N model (t) = 0.4 * 0.8 + 0.3 * 1.2 + 0.2 * 0.5 + 0.1 * 0.7 + 0.1

[0086] N model (t) = 0.32 + 0.36 + 0.1 + 0.07 + 0.1 = 0.95

[0087] Therefore, based on this output result, the model concludes that Xiao Zhang's current neural activity characteristics are 0.95. This value reflects Xiao Zhang's neural activity performance in the current state, providing a basis for subsequent regulatory path planning.

[0088] Next, according to this neural activity characteristic, an adaptive regulatory path planning method is used to design a personalized neural regulation scheme. According to the spatiotemporal variation trend of neural activity and the connectivity interaction between brain functional areas, the appropriate stimulation path for Xiao Zhang is calculated. Assuming that at this time, the model predicts that Xiao Zhang's emotional fluctuations are large and the activation of the prefrontal cortex is high, and needs to be adjusted in this area.

[0089] The specific regulation steps are as follows:

[0090] Time window: According to the neural activity time series characteristics of Xiao Zhang, assume that the neural regulation is performed in the time period of large emotional fluctuations (for example: within 10 minutes before and after the emotional fluctuation peak), and the time window is 10 minutes.

[0091] Stimulation frequency: Through spectral analysis of Xiao Zhang's neural activity, a suitable stimulation frequency is selected, for example 10 Hz, to intervene in the activity of the prefrontal cortex area, helping to relieve anxiety.

[0092] Stimulation intensity: The stimulation intensity is set according to the amplitude of neural activity, for example by calculating the amplitude of the EEG signal, a moderate stimulation intensity value (assuming 2 mA) is selected to ensure that the stimulation does not cause negative effects on the nervous system.

[0093] In Xiao Zhang's case, because he exhibited certain mood swings and anxiety reactions, particularly strong activity in specific brain regions (such as the prefrontal cortex), stimulation pathways needed to be planned based on the functional connectivity of different regions, such as the cerebral cortex and basal ganglia, and appropriate stimulation timing was selected through time optimization. These steps made the neuromodulation solution more efficient and able to dynamically adapt to Xiao Zhang's neural activity state.

[0094] Spatial Optimization Path Planning:

[0095] First, consider the spatial distribution of neural activity in different brain regions, especially the connectivity between the cerebral cortex and the basal ganglia. Assume that the neural activity characteristics of these brain regions have been obtained through previous EEG and fMRI data. In spatial optimization, the following formula is used to plan the stimulation path:

[0096]

[0097] Where: P spatial It represents the optimized stimulation path, reflecting the spatial stimulation distribution of the nervous system. Is the spatial range of the brain region, indicating the set of areas of the nervous system that need to be stimulated. In Xiao Zhang's case, assuming that the prefrontal cortex and basal ganglia areas need to be stimulated, then It is the combination of these two areas. j (x, t) is the intensity of neural activity at spatial location x and time t, reflecting the state of neural activity in a specific region. These intensity values ​​are obtained using EEG and fMRI data. For example, at a certain moment, the intensity of neural activity in Xiao Zhang's prefrontal cortex was 0.8, while the intensity of neural activity in the basal ganglia was 0.6.

[0098] a j is the weight coefficient for the jth region, representing the contribution of that region to overall neural activity. Assume that, based on spatial connectivity analysis, the weight for the prefrontal cortex is a1 = 0.7, and the weight for the basal ganglia is a2 = 0.3. m is the number of neural regions to be modulated. Assuming only the prefrontal cortex and basal ganglia are considered, m = 2. dx is the spatial integration variable, representing the integration process over spatial regions of the brain.

[0099] To further illustrate this, we can apply the above formula to Xiao Zhang's case. Assume that the neural activity intensity in the prefrontal region is 0.8, the neural activity intensity in the basal ganglia region is 0.6, and the known weight coefficients are 0.7 and 0.3, respectively. The above formula is used to calculate the optimized stimulation path, taking into account the contribution and neural activity intensity of each region.

[0100]

[0101] This calculation result shows that the optimized stimulation path will balance between the prefrontal cortex and basal ganglia regions, reflecting the higher contribution of the prefrontal cortex region to the overall neural activity.

[0102] Time-optimized path planning:

[0103] In addition to spatial optimization, the temporal sequence characteristics of neural activity, especially the timing of stimulation, need to be considered. In Xiao Zhang's case, his emotional fluctuations change over time, so the time window of stimulation needs to be optimized according to the time sequence characteristics of neural activity. The following formula is used for time optimization:

[0104]

[0105] Where:

[0106] P temporal (t) is the optimal stimulation path at time t, representing the optimized stimulation scheme for the current time period. E i (t) is the time sequence characteristic of the i-th neural activity, representing the fluctuation of neural activity at time t. Assuming that through the joint analysis of EEG and fMRI, the time sequence characteristic of the prefrontal cortex region E1(t) = 0.7 and the time sequence characteristic of the basal ganglia region E2(t) = 0.5 are obtained, reflecting the fluctuation intensity of Xiao Zhang's current emotional fluctuations.r i is the weight coefficient of the i-th neural activity characteristic, representing the influence degree of this characteristic on the time-optimized path. Assuming that according to the connectivity of the prefrontal cortex and basal ganglia, the weight coefficient of the prefrontal cortex region r1 = 0.6, and the weight coefficient of the basal ganglia region r2 = 0.4. n is the number of neural activities involved in regulation, representing the dimension of the time sequence data considered. Assuming that only the prefrontal cortex and basal ganglia regions are considered, therefore n = 2.

[0107] Apply the above formula to Xiao Zhang's case to calculate the optimal stimulation time path. First, substitute the known time sequence characteristics and weight coefficients:

[0108]

[0109] By maximizing this weighted sum result, the best time window for stimulation is determined. This means that in this time period, the stimulation will exert a higher weight on Xiao Zhang's prefrontal cortex region, thereby alleviating his emotional fluctuations.

[0110] Through these spatial and temporal optimization calculations, a personalized neural control pathway was derived. This pathway not only spatially allocates information based on Xiao Zhang's neural activity characteristics but also dynamically adjusts based on his mood swings and the temporal characteristics of his neural activity. Ultimately, based on these calculations, the neural control system can precisely adjust the timing and area of ​​stimulation to achieve the optimal neural control effect.

[0111] In this embodiment, since Xiao Zhang's nervous system will fluctuate during the treatment process, especially when his emotions fluctuate greatly, it is necessary to design an adaptive adjustment mechanism based on real-time data to dynamically adjust the stimulation path according to changes in neural activity. This process relies on the formula:

[0112]

[0113] in:

[0114] ΔP(t) represents the amount of stimulation path adjustment at time t, indicating the stimulation direction and amplitude that need to be adjusted. Through real-time monitoring and adjustment, the stimulation parameters can be fine-tuned to ensure that Xiao Zhang's neural activity remains at an ideal level. γ is the adjustment coefficient, which controls the sensitivity of the adjustment amount. Assume that in Xiao Zhang's personalized control plan, γ = 0.8, indicating that a more sensitive adjustment is desired within a certain range to cope with the changes in neural activity caused by Xiao Zhang's emotional fluctuations. N model (t) is the neural activity value predicted by the individualized neural representation model, which represents the expected neural system activity state. Assume that the model predicts that Xiao Zhang’s expected neural activity value at a certain moment is N model (t) = 0.75, representing the intensity of neural activity in the prefrontal area under the expectation state.

[0115] N current (t) is the current neural activity value obtained by the real-time monitoring device, which represents the actual state of nervous system activity. Based on Xiao Zhang’s real-time EEG monitoring data, it is assumed that the current neural activity value is N current (t) = 0.6, indicating that the current neural activity intensity is lower than the expected state. is the gradient operator, which represents the difference N model (t)N current The gradient of the change in (t) is calculated to measure the degree to which the neural activity state deviates from the expected state.

[0116] In Xiao Zhang’s case, assume that real-time data is obtained at a certain moment (e.g., the 10th minute during treatment). According to real-time monitoring, the current neural activity value is N current (10) = 0.6, and according to the prediction of the individualized neural representation model, the expected neural activity value is N model (10)=0.75.

[0117] Compute the difference:

[0118] ΔN = N model (10)N current (10) = 0.75 0.6 = 0.15

[0119] Next, compute the gradient of this difference:

[0120]

[0121] Assuming that the rate of change of the difference ΔN is relatively stable within this time window, and assuming that the gradient is approximately 1 (i.e., the rate of change per unit time is constant).

[0122] According to the formula:

[0123]

[0124] This means that at the 10th minute, an adjustment to the stimulation path is needed, with an adjustment amount ΔP(10) = 0.12, i.e., adjusting the stimulation intensity or time window to make up for the gap between neural activity and expectations.

[0125] Dynamic adjustment of the stimulation path: According to the above calculation, the adjustment amount ΔP(10) = 0.12 indicates that the intensity of the stimulation needs to be increased or the duration of the stimulation needs to be extended to make Zhang's neural activity closer to the desired state. Assuming that the basic intensity of the stimulation is 0.5, after real-time adjustment, the stimulation intensity can be increased to:

[0126] New stimulation intensity = 0.5 + 0.12 = 0.62

[0127] In addition, the stimulation time window will also be extended accordingly. Assuming that the stimulation time window is originally 10 minutes, the time window will be extended to 11 minutes after adjustment to better adjust the neural activity.

[0128] Throughout the treatment process, Zhang's real-time neural activity will be continuously monitored, and the stimulation path will be dynamically adjusted according to the above formula. For example, in the next few minutes, if Zhang's neural activity continues to be lower than the desired state, the same adjustment mechanism will be used to update the stimulation intensity and time window in real time. After each adjustment, the stimulation parameters will continuously approach the optimal value, ensuring the long-term effectiveness of neural regulation.

[0129] Example 2:

[0130] In combination Figure 3and psychological state as two independent systems, but this example method couples both to provide more accurate and customized regulation tailored to individual changes.

[0131] First, the neural activity signals and psychological state data are preprocessed through the encoder part of the deep learning model. In the case of Xiao Zhang, the neural activity signals include EEG signals and fMRI image data, while the psychological state data is obtained through physiological or psychological assessments, such as emotional fluctuations, stress levels, anxiety, etc.

[0132] Suppose Xiao Zhang's EEG signals show fluctuations in the prefrontal region during the treatment process, and at the same time, according to psychological assessments, his stress level is 5 / 10 (moderate stress). These signals will be input into the encoder part of the variational autoencoder for nonlinear mapping and fusion into a low-dimensional representation. This process is represented by the following formula:

[0133]

[0134] Where:

[0135] Z N is the low-dimensional coupled representation of neural activity and psychological state, reflecting the integrated characteristics of neural activity and psychological state. Through the encoder part, neural signals and psychological state data are mapped into a shared low-dimensional feature space to form a coupled representation. N is the input neural activity signal, including EEG and fMRI signals. For example, Xiao Zhang's EEG data shows fluctuations in the prefrontal region, where X N an example value is [0.65, 0.72, 0.68, 0.71] (reflecting prefrontal neural activity at different time points). X P is the input psychological state data, such as Xiao Zhang's emotional state of anxiety, stress level of 5, and emotional fluctuation of 3 / 10. Suppose X P value is [0.5, 0.6, 0.4] (representing emotion, stress, and anxiety). Θ E is the weight and bias of the encoder network, which is optimized during model training to map neural activity and psychological state signals to a low-dimensional coupled space. Suppose Θ E initial value is randomly set, and after training, the model will automatically optimize these parameters.

[0136] Low-dimensional coupled representation generation:

[0137] Through the encoder part of the variational autoencoder, the neural activity signal XN and psychological state data X P is mapped to a low-dimensional space to form a coupled representation Z N This coupling representation reflects the state of the nervous system at a specific moment and combines the individual’s psychological state. For example, after being processed by the deep learning model, Xiao Zhang’s coupling representation Z N is a four-dimensional vector:

[0138] Z N =[0.78,0.65,0.80,0.72]

[0139] This vector characterizes the coupling characteristics between neural activity (such as prefrontal cortex fluctuations) and psychological states (such as stress and anxiety).

[0140] Decoding and stimulation control output:

[0141] Next, couple the feature Z N The decoder part of the transform autoencoder is restored to generate a control output signal corresponding to the neural activity. The goal of the decoder part is to represent the low-dimensional coupling Z N Generate a corresponding stimulation scheme to effectively regulate the nervous system. Assume that the control signal output by the decoder is:

[0142]

[0143] in:

[0144] S stimulus is the stimulus signal output by the decoder, which indicates the intensity, frequency, and time parameters of the neural stimulation to be applied. Assume that the stimulus signal output by the decoder is S stimulus =[0.6,1.2,0.5], where the first value represents the stimulus intensity, the second value represents the stimulus frequency, and the third value represents the stimulus duration.

[0145] is the nonlinear mapping function of the decoder, which represents Z according to the coupling N and the decoder weights Θ D Map the low-dimensional representation back to the stimulus signal space. Assume Θ D It is optimized during the training process and can accurately map the low-dimensional representation to the stimulus signal. stimulus Each component of reflects a different regulation of the activity of the Xiaozhang nerve. For example, a stimulation intensity of 0.6 means that the stimulation current intensity is relatively moderate, and the frequency is 1.2Hz, which means that the frequency of the applied nerve stimulation signal is 1.2Hz and the duration is 0.5 seconds.

[0146] According to the current psychological state (e.g. emotional fluctuation, stress, etc.), the system will further adjust the neural stimulation scheme. Assuming that Xiao Zhang's stress level rises to 7 / 10 during the treatment process, and the emotional fluctuation increases, the system will adjust the stimulation parameters based on the real-time psychological state. By inputting real-time data into the variational autoencoder model, the coupling features of neural activity and psychological state will be updated, so as to adjust the output stimulation scheme. For example, when the psychological state changes, the output stimulation intensity, frequency and duration will be adjusted accordingly to better adapt to Xiao Zhang's current physiological and psychological state.

[0147] In the personalized neuroregulation treatment process of Xiao Zhang, this embodiment deeply explores how to jointly model neural activity signals and psychological state data through variational autoencoder under the deep learning framework to achieve accurate individualized regulation. In this process, Xiao Zhang's neural activity and psychological state will be mapped through the encoder part of the deep neural network, and these data will be learned and fused in the low-dimensional space through joint modeling, so as to realize dynamic neuroregulation.

[0148] Firstly, by introducing the variational autoencoder, the neural activity and psychological state data of Xiao Zhang are jointly modeled. The encoder part maps the input neural activity signal X N and psychological state data X P to a low-dimensional representation Z N through a nonlinear activation function. This low-dimensional representation not only contains the features of neural activity, but also fuses the features of psychological state, reflecting the interaction between the two. Assuming that in the treatment process of Xiao Zhang, the input neural activity signal includes its EEG waveform data and fMRI image data, and the psychological state data is obtained through psychological assessment. For example, Xiao Zhang's emotional fluctuation score is 6 / 10, anxiety level is 4 / 10, and stress is 7 / 10. These data will be input into the encoder for joint modeling together with the neural activity signal.

[0149] The joint modeling process of the variational autoencoder is expressed by the following formula:

[0150] Z N = W N ·X N +W P ·X P +b

[0151] Wherein:

[0152] Z N is the low-dimensional coupling representation of neural activity and psychological state, which is the comprehensive features of neural activity and psychological state, representing the state of the nervous system while considering the individual's psychological state. W Nis the weight matrix of neural activity signals, controlling the influence degree of neural activity signals in the low-dimensional representation. Assume W N is in the range of [0.5, 1.0], representing the importance of neural activity signals in the model. W P is the weight matrix of psychological state data, controlling the contribution degree of psychological state data in the low-dimensional representation. Assume W P is in the range of [0.3, 0.7], representing the influence degree of psychological state data on the neural regulation model. X N is the input neural activity signal, assuming that Xiao Zhang's EEG signal X N = [0.65, 0.72, 0.68, 0.70] at a certain moment, reflecting the neural activity of the prefrontal region. X P is the input psychological state data, assuming that Xiao Zhang's psychological state data is X P = [0.6, 0.5, 0.7], representing the levels of emotion, stress, and anxiety, respectively. b is the bias term, used to adjust the base value of the output, assuming b = 0.1.

[0153] Through these inputs, the variational autoencoder model can learn and output a low-dimensional coupled representation Z N , thereby integrating Xiao Zhang's neural activity and psychological state into a comprehensive feature.

[0154] According to the above formula, the low-dimensional coupled representation Z N of Xiao Zhang is calculated. Assuming that the weight matrix W N = [0.6, 0.8, 0.7, 0.75] of neural activity signals and the weight matrix W P = [0.5, 0.6, 0.7] of psychological state data, the contribution of each dimension can be calculated: Z N = [0.6·0.65+0.8·0.72+0.7·0.68+0.75·0.70]+[0.5·0.6+0.6·0.5+0.7·0.7]+0.1 After calculation, we get:

[0155] Z N = [0.39+0.576+0.476+0.525]+[0.3+0.3+0.49]+0.1 = [1.966]+[1.09]+0.1

[0156] = 3.156

[0157] Therefore, the low-dimensional coupled representation Z N is 3.156, representing the comprehensive state of Xiao Zhang's neural activity and psychological state at that moment. Once the low-dimensional representation Z N, which will be passed as input to the decoder part of the variational autoencoder. The task of the decoder is to output the corresponding neural stimulation protocol from the low-dimensional representation to modulate the neural activity of Xiao Zhang. The decoder will output the low-dimensional representation Z N to generate control signals for stimulation intensity, frequency, and timing. Assuming the output of the decoder is:

[0158]

[0159] where S stimulus is the stimulation signal generated by the decoder, representing the intensity, frequency, and duration of neural stimulation needed to be applied. For example, the stimulation signal output by the decoder is:

[0160] S stimulus = [0.65, 1.0, 0.8]

[0161] where:

[0162] 0.65 represents the stimulation intensity; 1.0 represents the stimulation frequency (Hz); and 0.8 represents the stimulation duration (seconds).

[0163] In this way, the system can generate precise neural modulation protocols based on Xiao Zhang's current neural activity and psychological state.

[0164] During the treatment process, Xiao Zhang's neural activity and psychological state will change constantly, and the system needs to dynamically adjust the neural stimulation protocol based on real-time data. For example, assume that during the treatment process, Xiao Zhang's emotional fluctuations and stress levels change, with stress levels rising from 7 / 10 to 8 / 10 and emotional fluctuations falling from 6 / 10 to 5 / 10. At this time, the system will input new psychological state data and adjust the stimulation protocol based on real-time neural activity.

[0165] Assume that the psychological state data obtained through real-time monitoring is:

[0166] X P = [0.65, 0.75, 0.8]

[0167] At this time, the system will again calculate the low-dimensional coupling representation Z N and output the adjusted stimulation signal S N based on the new Z stimulus , ensuring that the neural modulation always adapts to Xiao Zhang's actual needs. By jointly modeling neural activity and psychological state through a variational autoencoder and generating a low-dimensional coupling representation based on this, personalized neural stimulation protocols can be accurately provided based on Xiao Zhang's neural and psychological state. This method not only improves the accuracy of neural modulation, but also allows for adaptive adjustment based on real-time changes in psychological state, ensuring maximum treatment effectiveness.

[0168] This embodiment introduces real-time mental state monitoring and dynamically adjusts the coupling features of neural activity and mental state using a deep learning model. During treatment, Xiaozhang's emotional fluctuations, stress levels, and other mental state data are collected in real time and input into the deep learning model, which generates real-time adjustment plans for neural stimulation based on changes in neural activity signals. This dynamic adjustment process is calculated by a formula and optimizes the neural stimulation plan at each moment to maximize treatment effectiveness.

[0169] First, Xiaozhang's emotional fluctuations and stress level data are collected through real-time mental state monitoring devices such as emotion monitoring bracelets, stress sensors, etc. Assume that at a certain moment, Xiaozhang's emotional fluctuations are 6 / 10 and stress levels are 7 / 10. These data are combined with neural activity signals (such as EEG data) and input into the deep learning model. Through this model, the coupling features of neural activity and mental state can be extracted, and real-time adjustments to the neural regulation plan can be made based on these features.

[0170] In this process, the adjustment amount ΔS(t) of the neural stimulation plan is calculated by the following formula:

[0171]

[0172] Where:

[0173] ΔS(t) represents the adjustment amount of the neural stimulation plan at time t, indicating how the initial stimulation plan S0 needs to be adjusted based on the current neural activity and mental state data. N (t) is the coupling feature of neural activity and mental state, representing the low-dimensional representation extracted by the variational autoencoder, which combines Xiaozhang's neural activity and the interaction of mental state. S0 is the initial neural stimulation plan, usually set at the beginning of treatment, containing parameters such as stimulation frequency, intensity, etc. Assume that the initial plan S0 = [0.7, 1.0, 0.5], where:

[0174] 0.7 represents the stimulation intensity (unit: mA);

[0175] 1.0 represents the stimulation frequency (unit: Hz);

[0176] 0.5 represents the stimulation duration (unit: seconds).

[0177] P(t) is the mental state data at time t, including emotional fluctuations and stress levels. Assume that at a certain moment, Xiaozhang's mental state data is P(t) = [0.6, 0.7], representing emotional fluctuations and stress levels, respectively. A is the parameter of the regulation model, which is optimized through the deep learning training process and controls the adjustment method of the neural stimulation plan. Assume thatA contains several weight parameters that control the influence of different input features (neural activity and mental state data) on the adjustment of the stimulus.

[0178] After real-time mental state monitoring, Xiao Zhang's current mental state data is P(t) = [0.6, 0.7], with emotional fluctuation and stress level being 6 / 10 and 7 / 10 respectively. At this time, Xiao Zhang's neural activity signals are collected through an EEG device, assuming X N (t) = [0.72, 0.68, 0.75, 0.70], representing the neural activity of regions such as the prefrontal cortex. These data are processed through a variational autoencoder to obtain a low-dimensional coupling representation Z N (t).

[0179] Assume that the coupling representation extracted from the variational autoencoder model is:

[0180] Z N (t) = W N ·X N +W P ·P(t) + b

[0181] Assume W N = [0.6, 0.7, 0.8, 0.9], W P = [0.5, 0.6], and b = 0.2. The coupling representation Z N (t) can be calculated as:

[0182] Z N (t) = [0.6·0.72 + 0.7·0.68 + 0.8·0.75 + 0.9·0.70] + [0.5·0.6 + 0.6·0.7] + 0.2

[0183] The calculation is:

[0184] Z N (t) = [0.432 + 0.476 + 0.6 + 0.63] + [0.3 + 0.42] + 0.2 = 2.138

[0185] This is the low-dimensional coupling representation Z N (t) = 2.138 at the current time.

[0186] With the calculated low-dimensional coupling representation Z N (t), it is input into the regulation function for real-time adjustment:

[0187]

[0188] Assume that the regulation function is obtained according to the parameters Θ of the deep learning model after trainingA = [1.2, 0.9, 1.1] adjustment, the generated stimulation adjustment amount is:

[0189] ΔS(t) = [1.2·2.138 + 0.9·0.7 + 1.1·0.6] = [2.5656 + 0.63 + 0.66] = 3.8556

[0190] After calculating the adjustment amount ΔS(t) of the stimulation scheme, the new neural stimulation scheme can be obtained:

[0191] S new (t) = S0 + ΔS(t)

[0192] Where S0 = [0.7, 1.0, 0.5] is the initial stimulation scheme, so the new stimulation scheme is:

[0193] S new (t) = [0.7 + 3.8556, 1.0 + 3.8556, 0.5 + 3.8556] = [4.5556, 4.8556, 4.3556]

[0194] This new stimulation scheme will be applied to Xiaozhang's treatment, ensuring that neural regulation can dynamically adapt to his psychological state changes. By introducing real-time psychological state monitoring and combining the neural activity and psychological state coupling features extracted by the variational autoencoder, the neural stimulation scheme can be adjusted in real time to optimize Xiaozhang's treatment effect. In this process, the dynamic adjustment mechanism of the deep learning model effectively responds to Xiaozhang's emotional fluctuations and stress level changes, ensuring the accuracy and adaptability of individualized neural regulation.

Claims

1. The application method of adaptive neural network in individualized neural regulation is characterized by The following steps are involved: S1. Construction of Individualized Neural System Representation Model: S1.

1. Build a model of an individual's neural system representation based on multi-source data including EEG, fMRI, genetic information, and psychological assessments. S1.

1. Using machine learning techniques, including graph neural networks, we can mine key features of individual neural systems, including neural connectivity patterns and synchrony, to construct a basic map of neural activity for each individual. S2. Adaptive control path planning based on neural representation model: S2.

1. Based on the constructed individualized neural representation model, an adaptive control path planning method is introduced. The spatial and temporal trends of individual neural activity data are used to calculate the stimulation path that produces a regulatory effect on the nervous system. S2.

2. Path planning is the stimulation of neural signals. Based on the connectivity and interaction of various functional areas of the brain, the time window, intensity, and frequency of stimulation are calculated to achieve dynamic regulation of individual neural activity. S3. Adaptive activation mechanism and dynamic regulation of the nervous system: S3.

1. Introducing an adaptive activation mechanism based on neural feedback; the adaptive activation mechanism monitors individual neural activity in real time and infers potential activation nodes and connections in the neural network through dynamic changes in neural activity; S3.

2. By adjusting stimulation parameters in real time, we optimize the activation pattern of the nervous system, allowing the neural network to self-adjust and adapt to changes in the external environment at different times; S4. Combining personalized neural regulation with plasticity mechanisms: S4.

1. Use adaptive neural network models to assess the plasticity of individual neural networks in real time and, in combination with the activation and feedback mechanisms of the neural system, design regulatory strategies to promote the development of neural plasticity. S5. Deep coupling between neural regulation and individual psychological state: S5.

1. Develop dynamic control methods based on the coupling of neural activity and psychological state. Use deep learning models including transforming autoencoders to jointly model neural signals with psychological states, including mood swings and stress levels, and adjust neural control schemes based on real-time changes in an individual's psychological state. The adaptive control path planning method based on the neural representation model includes: By combining multi-source physiological data including EEG, fMRI, and genomics with psychological assessment data, a neural activity characteristic model for each individual is constructed. This neural activity characteristic model reflects the structure of the nervous system and describes the dynamic changes in neural function, including the connectivity and synchronization functional characteristics of the neural network. The individualized neural representation model is expressed as: in: Indicates time At this moment, the output of the individualized neural representation model represents the comprehensive performance of individual neural activities; It is Input features refer to time series data extracted from different physiological data sources including EEG signals and fMRI image data, reflecting the characteristics of neural activity; For the The weight coefficient of each feature indicates the importance of the feature to the output of the neural representation model; is the bias term, which adjusts the basic output of the model; is the total number of input features, indicating the dimension of the model input features; Stimulation paths are planned based on the functional connectivity of different regions of the nervous system, including the cerebral cortex and basal ganglia. The degree of functional connectivity of each region to other regions determines its contribution to overall neural activity, and stimulation paths are allocated based on connectivity. The formula for spatial optimization is: in: represents the optimized stimulation path, which reflects the stimulation distribution in space; It is the spatial extent of the brain region, representing the set of areas of the nervous system that need to be stimulated; is the spatial location of neural activity and time The intensity of the neural activity in a specific area. It is The weight coefficient of each region represents the contribution weight of the region to the overall neural activity; is the number of neural regions that need to be regulated, which indicates the number of neural regions considered; is the spatial integration variable, which represents the integration process over the spatial area of ​​the brain; In terms of time optimization, a stimulation time window is designed based on the time series of individual neural activity. The stimulation timing is selected within the dynamic time window by weighted summing and maximizing the temporal characteristics of neural activity. The time optimization formula is: in: For in time The optimal stimulation path at the moment represents the stimulation plan optimized for the current time period; It is The time series characteristics of neural activities indicate the time series characteristics of neural activities. Fluctuations at different times; It is The weight coefficient of each neural activity feature indicates the influence of the feature on the time optimization path; is the number of neural activities involved in regulation, indicating the dimension of the time series data considered; A real-time monitoring and dynamic adjustment mechanism is used; an adaptive adjustment mechanism based on real-time data is designed to dynamically adjust the stimulation path according to changes in neural activity; the real-time adjustment formula is: in: Indicates time The amount of stimulation path adjustment at the moment indicates the stimulation direction and amplitude that need to be adjusted; is the adjustment coefficient, which controls the sensitivity of the adjustment amount; It is the neural activity value predicted by the individualized neural representation model, representing the expected state of neural system activity; It is the current neural activity value obtained by real-time monitoring equipment, indicating the actual state of nervous system activity; is the gradient operator, which means the difference The gradient of the change is calculated; The method for deep coupling of neural regulation and individual psychological state includes: First, neural activity signals and psychological state data are preprocessed by the encoder part of the deep learning model. The encoder part fuses multimodal data into a low-dimensional representation through nonlinear mapping, forming the coupling characteristics of neural activity and psychological state. The coupling characteristics are then restored by the decoder part to generate a control output signal corresponding to the neural activity and adjust the neural stimulation scheme according to the current psychological state. The coupling mapping expression formula of the transformed autoencoder is: in: To represent the low-dimensional coupled representation of neural activity and psychological state; by transforming the encoder part of the autoencoder model, the neural activity signal and psychological state data are mapped to a shared low-dimensional feature space, forming a coupled representation of neural and psychological state; The input includes neural activity signals such as EEG and fMRI; The psychological state data is input; the state data is obtained through physiological or psychological assessment; are the weights and biases of the encoder network, which are the parameters adjusted during the network learning process; It is a nonlinear mapping function used to transform neural activity and psychological state signals into low-dimensional coupled representations.

2. The method for applying the adaptive neural network in individualized neural regulation according to claim 1, characterized in that The method for deep coupling of neural regulation and individual psychological state includes: By introducing a transforming autoencoder, neural activity and psychological state are jointly modeled; the structure of the transforming autoencoder includes an encoder, a decoder, and a joint layer; the encoder maps the input neural activity signal and psychological state data to a low-dimensional representation through a nonlinear activation function; the features of neural activity and psychological state in the latent space are learned and fused.

3. The method for applying the adaptive neural network in individualized neural regulation according to claim 2, characterized in that The method for deep coupling of neural regulation and individual psychological state includes: Real-time mental state monitoring is introduced, and the neural control plan is dynamically adjusted by combining the coupling characteristics of neural activity and mental state extracted by the transform autoencoder. Real-time mental state monitoring includes the real-time collection of emotional fluctuations and stress level data. This data is input into the deep learning model, and combined with the changing trends of neural activity signals, a neural stimulation adjustment plan is generated: in, For time The amount of adjustment to the neural stimulation plan at any given moment; this indicates real-time adjustments to the neural regulation plan based on current neural activity and psychological state; The initial nerve stimulation plan is set at the beginning of treatment, including the frequency and intensity of stimulation; For time Momentary psychological state data, including mood swings and stress levels; To regulate the parameters of the model, deep learning model training is used for optimization to control the adjustment of the stimulation scheme; It is a nonlinear control function used to adjust the neural stimulation scheme according to the coupling characteristics of neural activity and psychological state.

Images