A Construction Method, Device, Equipment and Medium for a Digital Twin Brain Mechanism Model

Abstract

The present application provides a method, apparatus, device and medium for constructing a digital twin brain mechanism model. The construction method includes: aligning multi-source brain state signal data in the time dimension and space dimension through a spatio-temporal alignment method and interpolation calculation, and supplementing the missing multi-source brain state signal data based on an inversion algorithm to obtain target signal data; coupling a multi-dimensional brain mechanism model, a multi-modal brain mechanism model and a multi-level brain mechanism model to obtain an original digital twin brain mechanism model; iteratively updating the parameters of the original digital twin brain mechanism model based on complex brain function information driven by twin data to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model. Through the multi-modal digital twin brain function data processing method and the coupling method of multi-dimensional, multi-modal and multi-level brain mechanism models, the personalization and accuracy of treatment plans are improved.

Description

Technical Field

[0001] The present application relates to the field of medical technologies, and in particular, to a method, device, equipment, and medium for constructing a digital twin brain mechanism model. Background Art

[0002] Currently, most transcranial magnetic clinical treatments use single-modal neuroimaging to evaluate the brain state, which has the deficiencies of missing or outdated historical information and one-sided analysis. Multi-modal neuroimaging data is complementary. However, it is difficult to fuse multi-modal neuroimaging data, and traditional transcranial magnetic therapy only statically evaluates the brain state, thus unable to assist in accurate calculation and positioning of targets, resulting in inaccurate and unreliable diagnosis and treatment of patients' brain diseases in the case of incomplete data.

[0003] The development of the brain state is dynamic, and multi-modal neuroimaging data has relevance and complementarity. Due to differences in time and space in the acquisition and processing methods of different neuroimaging data, the comprehensiveness and dynamics of the analysis of the state information of the digital twin brain function model show different sensitivities to different elements. The traditional brain state evaluation process often only evaluates a single-modal imaging technology, and the data is historical, resulting in relatively one-sided conclusions and unable to comprehensively evaluate the comprehensive state of the brain. In order to improve the efficiency of state information acquisition and evaluation, it is necessary to fully fuse multiple neuroimages, and the research in this area is still in its initial stage. Considering the large differences in brain network characteristics among individuals and the lag in the analysis of brain state changes, the accuracy of traditional brain function analysis for comprehensive analysis of complex brain information is relatively low. Summary of the Invention

[0004] In view of this, the purpose of the present application is to provide a method, device, equipment, and medium for constructing a digital twin brain mechanism model. By using a multi-modal digital twin brain function data processing method and a multi-dimensional, multi-modal, and multi-level brain mechanism model coupling method, the personalization and accuracy of treatment plans are improved; through the perception and processing of multi-modal brain function data, a multi-dimensional digital twin brain model is constructed to achieve personalized brain function analysis and provide support for formulating personalized treatment plans.

[0005] In the first aspect, an embodiment of the present application provides a method for constructing a digital twin brain mechanism model, and the construction method includes:

[0006] Obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data;

[0007] Based on the target signal data, construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model;

[0008] Based on the complex brain function information driven by twin data, perform parameter iterative update on the original digital twin brain mechanism model to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model.

[0009] Further, the method of aligning the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and supplementing the missing parts of the multi-source brain state signal data based on an inversion algorithm to obtain the target signal data includes:

[0010] Align the multi-source brain state signal data in the time dimension through unified timestamps and interpolation calculation, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the space dimension;

[0011] Adopt a generative network to construct an inversion algorithm based on an inverse generator, and perform inversion training using the correlation between multi-modal data to supplement the missing parts of the multi-source brain state signal data and obtain the target signal data.

[0012] Further, the construction of the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model based on the target signal data includes:

[0013] Based on the target signal data, construct a three-dimensional model describing multiple structural information and a functional model describing multiple dynamic responses to obtain the multi-dimensional brain mechanism model;

[0014] On the basis of the multi-dimensional brain mechanism model, construct the multi-modal brain mechanism model through extraction of key feature parameters of the multi-modal system and a modal fusion mechanism;

[0015] On the basis of the multi-dimensional brain mechanism model and the multi-modal brain mechanism model, perform multi-level data fusion based on the theory of neuroimaging source analysis to obtain the multi-level brain mechanism model.

[0016] Further, perform parameter iterative update on the multi-dimensional brain mechanism model in the original digital twin brain mechanism model through the following steps:

[0017] Extract the structural features and functional features of the structural model and functional model of the complex twin brain, and store the extracted features in a structured form;

[0018] Update the structured form in real time to achieve synchronous update of the structured form and the multi-dimensional brain mechanism model.

[0019] Furthermore, the parameters of the multi-modal brain mechanism model in the original digital twin brain mechanism model are iteratively updated through the following steps:

[0020] Extract neuroimaging data in multiple brain states under complex brain information based on a neuroimaging extraction algorithm to obtain multi-modal neuroimaging data;

[0021] Update the multi-modal brain mechanism model according to the multi-modal neuroimaging data;

[0022] Solve the multi-modal mechanism feature model of the multi-modal brain mechanism model and verify the accuracy. Modify the model and adjust the parameters of the multi-modal brain mechanism model according to the model accuracy verification results.

[0023] Furthermore, the parameters of the multi-level brain mechanism model in the original digital twin brain mechanism model are iteratively updated through the following steps:

[0024] Analyze the data characteristics of brain state information to determine brain information data and brain state evaluation results. Construct a correlation analysis data set based on the brain information data and the brain state evaluation results;

[0025] Extract strong association rules based on the association rule algorithm and the correlation analysis data set, and determine strong correlation data from the correlation analysis data set based on the strong association rules;

[0026] Iteratively update the parameters of the multi-level brain mechanism model through the strong correlation data.

[0027] In a second aspect, an embodiment of the present application further provides a construction device for a digital twin brain mechanism model. The construction device includes:

[0028] A target signal data determination module, configured to obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and supplement the missing parts of the multi-source brain state signal data based on an inversion algorithm to obtain target signal data;

[0029] A model construction module, configured to construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model based on the target signal data, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model;

[0030] A model update module, configured to perform parameter iterative update on the original digital twin brain mechanism model based on the complex brain function information driven by twin data, so as to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model.

[0031] Further, when the target signal data determination module is used to align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data, the target signal data determination module is further configured to:

[0032] Align the multi-source brain state signal data in the time dimension through unified timestamps and interpolation calculation, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the space dimension;

[0033] Adopt a generative network, construct an inversion algorithm based on an inverse generator, and perform inversion training using the correlation between multi-modal data to perform missing data supplementation on the multi-source brain state signal data to obtain the target signal data.

[0034] In a third aspect, an embodiment of the present application further provides an electronic device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are executed by the processor, the steps of the method for constructing the digital twin brain mechanism model as described above are executed.

[0035] In a fourth aspect, an embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is run by a processor, the steps of the method for constructing the digital twin brain mechanism model as described above are executed.

[0036] According to the method, device, equipment, and medium for constructing the digital twin brain mechanism model provided by the present application, the accuracy and efficiency of the brain function disorder treatment system are improved. Through innovative multi-modal digital twin brain function data processing methods and multi-dimensional, multi-modal, and multi-level brain mechanism model coupling methods, the personalization and accuracy of treatment plans are improved; through the perception and processing of multi-modal brain function data, a multi-dimensional digital twin brain model is constructed to realize personalized brain function analysis and provide support for formulating personalized treatment plans.

[0037] To make the above objects, features, and advantages of the present application more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] To more clearly illustrate the technical solutions of the embodiments of the present application, the accompanying drawings required for the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present application and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0039] Figure 1 It is a flowchart of a method for constructing a digital twin brain mechanism model provided by an embodiment of the present application;

[0040] Figure 2 It is a schematic structural diagram of a device for constructing a digital twin brain mechanism model provided by an embodiment of the present application;

[0041] Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application. Specific embodiments

[0042] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only some, rather than all, of the embodiments of the present application. Usually, the components of the embodiments of the present application described and illustrated in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the present application to be protected, but only represents the selected embodiments of the present application. Based on the embodiments of the present application, every other embodiment obtained by those skilled in the art without creative efforts belongs to the scope of protection of the present application.

[0043] First, the applicable application scenarios of the present application will be introduced. The present application can be applied to the field of medical technology.

[0044] Currently, most transcranial magnetic clinical treatments use single-modal neuroimaging to evaluate the brain state, which has the deficiencies of missing or outdated historical information and one-sided analysis. Multi-modal neuroimaging data is complementary. However, it is difficult to fuse multi-modal neuroimaging data, and traditional transcranial magnetic treatment only statically evaluates the brain state, thus unable to assist in accurate calculation and positioning of the target, resulting in inaccurate and unreliable diagnosis and treatment of patients' brain diseases in the case of incomplete data.

[0045] It has been found through research that the development of brain states is dynamic, and there are correlations and complementarities in multi-modal neuroimaging data. Due to differences in time and space in the acquisition and processing of different neuroimaging data, the comprehensiveness and dynamics of the analysis of the state information of the digital twin brain function model show different sensitivities to different elements. However, the traditional brain state assessment process often only evaluates a single-modal imaging technique, and the data is historical, resulting in relatively one-sided conclusions and being unable to comprehensively evaluate the overall state of the brain. To improve the efficiency of state information acquisition and assessment, it is necessary to fully integrate multiple neuroimages, and the research in this area is still in its initial stage. Considering the large differences in brain network characteristics among individuals and the lag in the analysis of brain state changes, the accuracy of traditional brain function analysis in comprehensively analyzing complex brain information is relatively low.

[0046] Based on this, the embodiments of the present application provide a method for constructing a digital twin brain mechanism model, which improves the personalization and accuracy of treatment plans through a multi-modal digital twin brain function data processing method and a multi-dimensional, multi-modal, and multi-level brain mechanism model coupling method; through the perception and processing of multi-modal brain function data, a multi-dimensional digital twin brain model is constructed to achieve personalized brain function analysis and provide support for formulating personalized treatment plans.

[0047] Please refer to Figure 1 , Figure 1 which is a flowchart of a method for constructing a digital twin brain mechanism model provided by the embodiments of the present application. As Figure 1 shown in

[0048] S101, obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data.

[0049] Regarding the above step S101, in specific implementation, first obtain multi-source brain state signal data, then align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and then perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data.

[0050] During the treatment of brain dysfunction, the brain states of different patients show differences. Through the acquisition method of multi-source brain state signal data, personalized data acquisition for each patient is realized, including the real-time perception and efficient processing of original brain state data such as magnetic resonance images, near-infrared images, and electroencephalogram images. The data sources contain rich spatio-temporal information for subsequent analysis and modeling. Secondly, different brain state signal data often have heterogeneity, including problems such as inconsistent spatio-temporal resolution, high dimensionality, and modal imbalance. Through an efficient data integration method, it is ensured that these multi-source data can jointly represent the brain structure and function. A spatio-temporal alignment method that maximizes the alignment degree of sub-elements of each modal neuroimage is used to unify the time dimension of different data, and interpolation calculations are used to ensure the alignment of neuroimage data in the time dimension. The data at the same moment is projected onto a high-resolution space to achieve the alignment of multi-modal data in the space dimension and process complex heterogeneous data. Based on the perceived data, an inversion algorithm based on an inversion generator is constructed to achieve the consistency of resolution and supplement the missing dynamic acquisition data.

[0051] According to the embodiments provided in the present application, when collecting multi-source brain state signal data, first, in view of the characteristics of complex brain information multi-modal data, personalized collection and collaborative measurement are performed on complex brain information and its multi-modal neuroimaging data. The heterogeneous multi-source data protocol is used for parsing, and the multi-modal neuroimaging data is imported and integrated to achieve comprehensive perception and efficient processing of multi-modal data. Multi-modal neuroimaging includes image data such as magnetic resonance images, near-infrared images, and electroencephalogram images. Among them, the collection of MRI image data usually requires the patient to enter the interior of a magnetic resonance instrument, and images are generated and collected through magnetic fields and radio pulses. NIRS image data reflects brain activities by detecting the absorption of near-infrared light in the blood, and can collect the patient's cerebral oxygenation level data in real time in a portable manner. The EEG device usually consists of multiple electrodes, which are placed on the scalp surface to cover the entire brain area. The three can be collected separately in the same diagnosis and treatment room, and professional technicians are required to operate and ensure the accuracy and stability of data collection. Thus, collaborative measurement and analysis of the brain information of the patient in the current situation are carried out. The data collected from different brain imaging technologies are integrated into a database. When analyzing the data, data preprocessing of the multi-modal image data is required, including steps such as removing noise, correcting image distortion, and standardizing the data format to ensure the quality and consistency of the data; performing a registration process to spatially align the data from different modalities to ensure that the data from different modalities correspond to the same position in the same brain anatomical space for subsequent analysis and comparison; MRI provides detailed information about brain structure, NIRS provides real-time monitoring of cerebral oxygenation level, and EEG provides time-domain and frequency-domain information of brain electrical activities. Through the integration of different modality data, mutual verification and supplementation are achieved, which helps doctors make more accurate diagnosis and treatment decisions. The heterogeneous multi-source data protocol parsing refers to the unified processing and analysis of different structures and formats of different types of data in multi-modal medical imaging data. Through data processing processes and methods such as standardized data preprocessing, feature extraction, and data association, the MRI three-dimensional image data, EEG time series data, and NIRS cerebral oxygenation level are analyzed and associated to obtain corresponding brain state information data.

[0052] Aiming at the problem of blurred individual spatial information of wearable neuroimaging sensing devices, an acquisition technology based on visual reconstruction and sensing device feature recognition is adopted. By recording the spatial position of the sensor and tracing back the neuroimaging data, accurate acquisition of individual neural signals with high spatial fidelity is achieved.

[0053] As an optional implementation manner, for the above step S101, aligning the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and supplementing the missing parts of the multi-source brain state signal data based on an inversion algorithm to obtain target signal data, includes:

[0054] Step 1011: Align the multi-source brain state signal data in the time dimension through unified timestamp and interpolation calculation, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the space dimension.

[0055] Regarding the above Step 1011, in specific implementation, align the multi-source brain state signal data in the time dimension through unified timestamp and interpolation calculation, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the space dimension. Here, in specific implementation, adopt multi-layer semantic matching to analyze low-quality and unbalanced modal data, perform dimensionality reduction on linear and non-linear high-dimensional data respectively. For the problem of inconsistent spatio-temporal resolution of heterogeneous data, use a spatio-temporal alignment method based on maximizing the alignment degree of each modal neuroimaging sub-elements. Ensure the alignment of the time dimension of neuroimaging data through unified timestamp and interpolation calculation. According to the backtracked high-fidelity spatial neuroimaging information, project the data at the same moment into a high-resolution space to achieve the alignment of multi-modal data in the space dimension. Perform semantic matching on data of different modalities, and match and analyze the data at different brain information levels (such as pixel level, region level, feature level). High-dimensional data usually contains a large number of features, and dimensionality reduction can help reduce the dimension of the data. Use principal component analysis (PCA) and t-distributed stochastic neighbor embedding (t-SNE) to process linear and non-linear data, reducing the dimension of the data while retaining the key features. Inconsistent spatio-temporal resolution of heterogeneous data may lead to difficulties in data integration and analysis. Achieve spatio-temporal alignment by maximizing the alignment degree of each modal neuroimaging sub-elements, and use unified timestamp and interpolation calculation for data acquisition to ensure the alignment of the time dimension of neuroimaging data.

[0056] Step 1012: Adopt a generative network, construct an inversion algorithm based on an inverse generator, and perform inversion training using the correlation between multi-modal data to supplement the missing parts of the multi-source brain state signal data and obtain the target signal data.

[0057] Here, the generative network can be an improved generative adversarial network, variational autoencoder, recurrent neural network, Transformer and other generative networks, and this application does not make specific limitations on this.

[0058] Regarding the above step 1012, in specific implementation, a generation network is adopted to construct an inversion algorithm based on an inversion generator, and inversion training is carried out by utilizing the correlation between multi-modal data to supplement the missing multi-source brain state signal data and obtain target signal data. Specifically, based on the perception data, an inversion algorithm based on an inversion generator is constructed, and inversion training is carried out by utilizing the correlation between multi-modal data to improve the data resolution, realize the consistency of resolution and supplement the missing dynamically acquired data. Perception data refers to the original and observed data, which may include low-resolution, incomplete or noisy data. A generation network structure is used to improve the data resolution and integrity. A model that trains and generates new data from the observed data can learn the latent distribution of the data and generate high-quality data similar to the original data. The inversion algorithm uses an inversion generator that can generate high-resolution and complete data from the perception data for data reconstruction and supplementation. By simultaneously training the generation networks of multiple modalities, the generated data can be made more accurate and complete, thereby improving the data resolution and consistency, enabling it to learn the mapping relationship from low-resolution and incomplete data to high-resolution and complete data, and thus realizing data reconstruction and supplementation.

[0059] S102, based on the target signal data, construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model.

[0060] For the above-mentioned step S102, in specific implementation, using the target signal data obtained in the above-mentioned step S101, a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model are constructed, and the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model are model-coupled to obtain an original digital twin brain mechanism model. Specifically, in the above-mentioned step S102, for the coupling of the multi-dimensional, multi-modal, and multi-level brain mechanism models, aiming at the different characteristics of multi-modal information such as "magnetic resonance - near-infrared - electroencephalogram", by extracting key feature parameters, the information of different modalities is integrated into a unified model. The model integration method unifies them into a mathematical equation expression by retaining the characteristics of each modality information while eliminating the differences between modalities, enabling the multi-modal information to be fused and unified in a complex brain information environment. In order to more comprehensively describe the structural information and functional connections of the key parts of the digital twin brain function model, a multi-dimensional brain structure-function model is established. The model covers multiple dimensions of the brain structure and function, including appearance shape, internal structure, function distribution, functional atlas, functional connection, etc. By analyzing the multi-dimensional model, the structure and function of brain information can be understood more accurately, assisting doctors in effective treatment and rehabilitation. On the basis of the establishment of the multi-dimensional model and the integration of the multi-modal model, a multi-level data fusion method based on the theory of neuroimaging traceability analysis is designed. The multi-level model embedding method includes data-level, feature-level, and decision-level fusion. Through this hierarchical fusion method, the multi-level characteristics of brain state information are fully utilized to obtain the data interpretation and brain state evaluation of brain information, providing richer information for the construction of the brain mechanism model.

[0061] In this way, in order to accurately analyze the brain function state data with spatio-temporal information according to different fusion requirements, a multi-dimensional model of the "structure-function" of the twin brain is constructed to specifically describe the structural information and functional connections of the key parts of the digital twin brain function model; for modalities such as "magnetic resonance - near-infrared - electroencephalogram", multi-modal model integration is carried out; around hierarchical information such as "data-level - feature-level - decision-level", a hierarchical data fusion method based on the theory of neuroimaging traceability analysis is adopted for fusion analysis; finally, based on the above "multi-dimensional - multi-modal - multi-level" model, digital twin mechanism model coupling is carried out based on complex brain information. Specifically, when carrying out digital twin brain mechanism model coupling, the digital twin brain mechanism model coupling with multi-dimensional, multi-modal, and multi-level attributes is realized based on modeling software such as Matlab / Mworks / ANSYS.

[0062] As an alternative implementation manner, for the above-mentioned step S102, the construction of the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model based on the target signal data includes:

[0063] Step 1021: Based on the target signal data, construct a three-dimensional model describing multiple structural information and a functional model describing multiple dynamic responses to obtain the multi-dimensional brain mechanism model.

[0064] For the above Step 1021, in specific implementation, based on the target signal data, construct a three-dimensional model describing multiple structural information and a functional model describing multiple dynamic responses to obtain the multi-dimensional brain mechanism model. Specifically, first, perform multi-dimensional model characterization, including the complex twin brain and its associated element structure model: including constructing a three-dimensional model describing structural information such as the appearance shape, internal structure, and mapping relationship of elements; through feature extraction for each type of modality data, for example, for magnetic resonance imaging (MRI) data, extract structural features of the brain such as gray matter volume and white matter fiber bundles; for near-infrared (NIRS) data, extract blood oxygen concentration and metabolic features of the brain; for electroencephalogram (EEG) data, extract spectral features of the EEG signal, etc. Parameterize the extracted features and convert them into specific numerical or vector forms to construct the three-dimensional model. The complex twin brain and its associated element functional model: including constructing a functional model describing dynamic responses such as functional distribution, functional atlas, and functional connection of elements. The voxel-based structure model uses the collected MRI data to describe the anatomical structure and volume of different brain regions. By performing spectral analysis and metabolic analysis on EEG and NIRS data, obtain the distribution of different frequency bands and metabolites in the brain, calculate the covariance between different brain regions, and represent the functional connection relationship between different brain regions as an atlas to describe the network structure and dynamic changes of brain function. Perform time series analysis on the brain imaging data collected multiple times to obtain the change trend of the functional connection between brain regions.

[0065] Step 1022: Based on the multi-dimensional brain mechanism model, construct the multi-modal brain mechanism model through key feature parameter extraction of the multi-modal system and a modality fusion mechanism.

[0066] Regarding the above step 1022, in specific implementation, based on the construction of a multi-dimensional brain mechanism model, through the extraction of key feature parameters of the multi-modal system and the modal fusion mechanism, a multi-modal brain mechanism model is constructed. Specifically, multi-modal model integration is carried out. Based on the construction of a "structure-function" multi-dimensional model, through the extraction of key feature parameters of multi-modal systems such as "magnetic resonance imaging - near-infrared spectroscopy - electroencephalogram", the mathematical equation expression of different modal models is realized, and based on the modal fusion mechanism, the fusion and unification of multi-modal models in a complex brain information environment are achieved. The key feature parameters extracted are usually numerical values or vectors that can represent the important information of each modality. For magnetic resonance imaging (MRI) images, image segmentation and registration techniques are used to process the MRI images, and the structural feature parameters of different brain regions are extracted, including gray matter volume, white matter fiber bundle connection density, thickness of brain regions, etc. For near-infrared spectroscopy (NIRS) data, through preprocessing and signal processing of NIRS data, including noise reduction, filtering, time-domain analysis, and frequency-domain analysis, etc., the brain metabolic feature parameters are extracted, including cerebral blood oxygen concentration, blood flow, etc. For electroencephalogram (EEG) data, time-domain and frequency-domain analysis are carried out, and the feature parameters of different frequency bands and time periods are extracted, including energy density of different frequency bands, coherence of frequency bands, etc.

[0067] Step 1023, based on the multi-dimensional brain mechanism model and the multi-modal brain mechanism model, multi-level data fusion is carried out based on the theory of neuroimaging source analysis to obtain the multi-level brain mechanism model.

[0068] Regarding the above step 1023, in specific implementation, based on the construction of a multi-dimensional brain mechanism model and a multi-modal brain mechanism model, multi-level data fusion is carried out based on the theory of neuroimaging source analysis to obtain a multi-level brain mechanism model. Specifically, multi-level model embedding is carried out. Based on the construction of a "structure-function" multi-dimensional model and multi-modal model integration, multi-level data fusion is carried out based on the theory of neuroimaging source analysis. First, data-level fusion uses a multi-modal neuroimaging data preprocessing method to obtain a neuroimaging visualization expression and aligned data of each modality; feature-level fusion uses algorithms such as correlation analysis and multiple hypothesis testing to extract data features of brain states; decision-level fusion uses methods such as fuzzy inference algorithms and D-S inference algorithms to make full use of the complementary characteristics of brain state information to obtain data interpretations of brain information and brain state evaluations.

[0069] For multimodal imaging data, there are often various types of noise, such as environmental noise, motion artifacts, etc. Filters (such as Gaussian filters, median filters) are used to remove this noise to reduce interference with subsequent analysis. Data from different time points or different sources may have spatial inconsistencies. Feature-based registration is used to map the data into a unified spatial coordinate system (such as a whole-brain mask). There may also be intensity differences in different scan sequences, different devices, or different time points. Z-score normalization, i.e., intensity normalization, is performed to make the intensity values between different images comparable. Dynamic imaging data may be affected by subject movement. Image registration-based methods are used to remove motion artifacts and correct the data. For brain imaging data, the functional connectivity between different brain regions can be revealed by calculating the correlation coefficients (such as Pearson correlation coefficient, Spearman correlation coefficient) between them, and the Bonferroni correction multiple hypothesis testing method is used to correct the multiple comparison problem to identify significant differences between different brain regions or different time points. Pattern recognition and classification tasks are performed using the extracted correlation and significant features in combination with machine learning algorithms (such as support vector machine (SVM), random forest (RandomForest), deep learning neural networks, etc.). These algorithms can use the extracted features to classify and predict different brain states, thereby achieving a deeper understanding and analysis of brain state information.

[0070] S103. Based on the complex brain function information driven by twin data, perform parameter iterative update on the original digital twin brain mechanism model to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain the target digital twin brain mechanism model.

[0071] Regarding the above step S103, in specific implementation, perform parameter iterative update on the original digital twin brain mechanism model according to the complex brain function information driven by twin data to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain the target digital twin brain mechanism model.

[0072] Specifically, in the above step S103, the evolution of the brain mechanism model driven by twin data is carried out: based on the multi-dimensional brain structure-function model, by analyzing the structural and functional characteristics of brain information, the iterative update of multi-dimensional model parameters is realized. Through the real-time analysis and extraction of different brain states, it is ensured that the model is continuously optimized and updated as the multi-dimensional data is updated, providing a basis for the dynamic evolution of the brain mechanism model. For different brain states of multi-modal information, multi-modal neuroimaging devices are used for data collection and analysis. By extracting key feature parameters and updating the model of multi-modal information, the update of the mechanism characteristics of different brain states supported by multi-modal neuroimaging data is realized, ensuring that the model can accurately reflect the dynamic changes of brain states. Strong association rules are extracted through the association rule algorithm to realize the update of the multi-level model. Through the correlation analysis of multi-level data, more accurate data support is provided for the brain mechanism model to predict the internal feature change trends of the brain model and key parts.

[0073] To predict the internal feature change trends of the brain model and key parts, based on the coupling of the complex brain information digital twin brain function mechanism model, the iterative update of the "structure-function" multi-dimensional model parameters driven by twin data; the mapping update of multi-modal information such as "nuclear magnetic resonance - near-infrared - electroencephalogram"; and the update of the multi-level data correlation degree of "data level - feature level - decision level" are carried out according to the evolution mechanism of the complex brain function information mechanism model driven by twin data.

[0074] As an optional embodiment, the parameter iterative update of the multi-dimensional brain mechanism model in the original digital twin brain mechanism model is carried out through the following steps:

[0075] A: Extract the structural characteristics and functional characteristics of the structural model and functional model of the complex twin brain, and store the extracted characteristics in a structured form.

[0076] B: Update the structured form in real time to achieve the synchronous update of the structured form and the multi-dimensional brain mechanism model.

[0077] For the above steps A - B, for the multi - dimensional model parametric update, first, the structural features and functional features of the structural model (appearance shape, internal structure, association relationship, etc.) and functional model (functional distribution, functional atlas, functional connection, etc.) of the complex twin brain are analyzed and extracted in real - time, and the extracted features are stored in a structured form. The structured form is updated in real - time to achieve the synchronous update of the structured form and the multi - dimensional model. For image data such as magnetic resonance imaging (MRI), image processing and segmentation algorithms are used to extract the contours and appearance shape features of different brain regions, and volume, density and other internal structure features of different brain tissues (such as gray matter, white matter) are extracted through segmentation and registration techniques. Through methods such as correlation analysis and network analysis, the association relationship between different brain regions is analyzed, a brain functional connection network is constructed, and then the features of the association relationship are extracted. For dynamic functional brain image data, such as electroencephalogram (EEG), near - infrared spectroscopy (NIRS), etc., functional distribution features including functional connection strength, frequency features, etc. of different frequency bands or time periods can be extracted through frequency - domain analysis, time - domain analysis and other methods.

[0078] As an optional embodiment, the multi - modal brain mechanism model in the original digital twin brain mechanism model is updated by parameter iteration through the following steps:

[0079] a: Extract neuroimage data in multiple brain states under complex brain information based on the neuroimage extraction algorithm to obtain multi - modal neuroimage data.

[0080] b: Update the multi - modal brain mechanism model according to the multi - modal neuroimage data.

[0081] c: Solve the multi - modal mechanism feature model and verify the accuracy of the multi - modal brain mechanism model, and modify the model and adjust the parameters of the multi - modal brain mechanism model according to the model accuracy verification results.

[0082] For the above steps a - c, in specific implementation, for the update of multi-modal mechanism features, first, according to different mechanism brain states of complex brain information, it is divided into: brain resting state, brain task state, brain abnormal state, etc.; then the neuroimaging model uses a variety of neuroimaging devices to check and verify before and after treatment and monitor in real time during treatment. Based on the neuroimaging extraction algorithm, the extraction of neuroimaging data in multiple brain states under complex brain information is realized. According to the multi-modal neuroimaging data, different mechanism feature models are updated respectively; finally, the solution and accuracy verification of the multi-modal mechanism feature model are carried out. According to the model accuracy verification results, the model is modified and the parameters are adjusted, so as to realize the update of the mechanism features of different brain states supported by multi-modal neuroimaging data in the digital twin brain mechanism model; based on the defined different states such as resting state, task state and abnormal state, the neuroimaging extraction algorithm is designed and used to extract the features strongly correlated with each state from the multi-modal neuroimaging data. The extracted features can include structural features (such as brain region volume, gray matter density, etc.) and functional features (such as frequency spectrum, blood oxygen concentration, etc.). The extracted neuroimaging features are corresponded to their respective brain states. Updating the multi-modal mechanism feature model includes operations such as adjusting model parameters and adding new features, and integrating it into the overall digital twin brain mechanism model.

[0083] As an alternative embodiment, the parameters of the multi-level brain mechanism model in the original digital twin brain mechanism model are iteratively updated through the following steps:

[0084] (1): Analyze the characteristic features of the data features of the brain state information, determine the brain information data and the brain state evaluation results, and construct a correlation analysis data set based on the brain information data and the brain state evaluation results;

[0085] (2): Extract strong association rules based on the association rule algorithm and the correlation analysis data set, and determine strong correlation data from the correlation analysis data set based on the strong association rules;

[0086] (3): Iteratively update the parameters of the multi-level brain mechanism model through the strong correlation data.

[0087] For the above steps (1)-(3), in specific implementation, for the update of the multi-level data correlation degree, first, analyze the data characteristic features of the brain state information, interpret the brain information data and evaluate the brain state, and construct a correlation analysis data set; secondly, extract strong association rules based on the association rule algorithm; finally, update the multi-level model through the strong correlation degree data. Collect and organize the data characteristic features of the brain state information such as multi-modal neuroimaging data, physiological signal data, and clinical evaluation data, and conduct statistical analysis on the collected data characteristic features, including descriptive statistics, frequency distribution, correlation analysis, etc., to obtain the basic characteristics and mutual relationships of the data. Construct a correlation analysis data set through steps such as selecting appropriate data characteristic features, data preprocessing (such as missing value processing, outlier processing, etc.), and data encoding, which includes the correlation information between different features. Strong association rules refer to rules with high support and high confidence. Use association rule mining algorithms (such as Apriori algorithm, FP-Growth algorithm, etc.) to extract strong association rules, and apply the extracted strong association rules to the update of the multi-level model to update the parameters, structure, etc. of the model.

[0088] This application introduces the concept of digital twin, and proposes a method for constructing and evolving a digital twin data and mechanism model based on complex brain information. On the basis of the coupling of multi-dimensional, multi-modal, and multi-level brain mechanism models, through the brain function data inversion algorithm and correlation analysis, the state evolution law of the digital twin brain function model under large-scale brain state simulation is deduced, effectively solving the problems of one-sidedness, staticness, and fuzziness in the evaluation of brain state by single-dimensional neuroimaging. It improves the comprehensiveness of brain state evaluation, has obvious innovation, and greatly improves the treatment effect.

[0089] According to the method for constructing the digital twin brain mechanism model provided by this application, improve the accuracy and efficiency of the brain dysfunction treatment system. Through the innovative multi-modal digital twin brain function data processing method and the multi-dimensional, multi-modal, and multi-level brain mechanism model coupling method, improve the personalization and accuracy of the treatment plan; through the perception and processing of multi-modal brain function data, construct a multi-dimensional digital twin brain model to achieve personalized brain function analysis and provide support for formulating personalized treatment plans. Make up for the technical difficulties in multi-modal data processing, solve the problems of heterogeneity, high dimension, and inconsistency of multi-modal brain function data, and use spatio-temporal alignment methods and data fusion technologies to efficiently process multi-modal data and ensure the integrity and accuracy of brain function information. Through the evolution of the brain mechanism model driven by twin data, realize the prediction of the internal feature change trend of the brain model and key parts, and provide in-depth understanding of the brain mechanism and its evolution process. Through the innovative digital twin brain mechanism model coupling method, provide more accurate, personalized, and dynamic treatment plans and support for the treatment of brain dysfunction, and improve the treatment effect and the patient recovery rate.

[0090] Compared with clinically used products, the multi-modal brain function data perception and processing solution mentioned in this application has the advantages of personalized acquisition, heterogeneous data integration, and consistent resolution. It can more comprehensively and efficiently perceive and process the brain states of different patients, providing more accurate data support for subsequent treatment of brain function disorders. The multi-dimensional, multi-modal, multi-level brain mechanism model coupling solution realizes the comprehensive modeling and analysis of brain structure and function by integrating multi-modal information, establishing multi-dimensional brain structure-function models, and multi-level data fusion, providing doctors with richer information and contributing to the precise treatment of brain diseases. Adopting this solution can ensure real-time performance while maximizing the recognition accuracy, enabling patients to obtain a better personalized and real-time treatment experience and providing stronger support for the treatment of brain function disorders. In the specific processing of multi-dimensional data information feature extraction, this solution proposes to use the correlation analysis and inversion algorithm of multi-modal neuroimaging data to achieve consistent resolution and supplement dynamic data, providing an effective solution for accurately extracting multi-dimensional data information.

[0091] Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of a digital twin brain mechanism model construction device provided by an embodiment of this application. As Figure 2 shown in

[0092] The target signal data determination module 201 is configured to obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data;

[0093] The model construction module 202 is configured to construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model based on the target signal data, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model;

[0094] The model update module 203 is configured to perform parameter iterative update on the original digital twin brain mechanism model based on complex brain function information driven by twin data to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model.

[0095] Further, when the target signal data determination module 201 is used to align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data, the target signal data determination module 201 is further configured to:

[0096] Perform time - dimension alignment on the multi - source brain state signal data through unified timestamp and interpolation calculation, and project the multi - source brain state signal data at the same moment into a high - resolution space to perform space - dimension alignment on the multi - source brain state signal data;

[0097] Adopt a generative network, construct an inversion algorithm based on an inverse generator, and perform inversion training using the correlation between multi - modal data to supplement the missing parts of the multi - source brain state signal data and obtain the target signal data.

[0098] Furthermore, when the model construction module 202 is used to construct a multi - dimensional brain mechanism model, a multi - modal brain mechanism model, and a multi - level brain mechanism model based on the target signal data, the model construction module 202 is further used for:

[0099] Based on the target signal data, construct a three - dimensional model describing multiple structural information and a functional model describing multiple dynamic responses to obtain the multi - dimensional brain mechanism model;

[0100] On the basis of the multi - dimensional brain mechanism model, construct the multi - modal brain mechanism model through key feature parameter extraction of the multi - modal system and a modal fusion mechanism;

[0101] On the basis of the multi - dimensional brain mechanism model and the multi - modal brain mechanism model, perform multi - level data fusion based on the theory of neuroimaging source analysis to obtain the multi - level brain mechanism model.

[0102] Furthermore, the model update module 203 is further used to perform parameter iterative update on the multi - dimensional brain mechanism model in the original digital twin brain mechanism model through the following steps:

[0103] Extract the structural features and functional features of the structural model and functional model of the complex twin brain, and store the extracted features in a structured form;

[0104] Update the structured form in real - time to achieve synchronous update of the structured form and the multi - dimensional brain mechanism model.

[0105] Furthermore, the model update module 203 is further used to perform parameter iterative update on the multi - modal brain mechanism model in the original digital twin brain mechanism model through the following steps:

[0106] Extract neuroimaging data in multiple brain states under complex brain information based on a neuroimaging extraction algorithm to obtain multi - modal neuroimaging data;

[0107] Update the multi - modal brain mechanism model according to the multi - modal neuroimaging data;

[0108] Solve the multi-modal mechanism feature model and verify the accuracy of the multi-modal brain mechanism model, and modify the model and adjust the parameters of the multi-modal brain mechanism model according to the model accuracy verification results.

[0109] Furthermore, the model update module 203 is further configured to perform parameter iterative update on the multi-level brain mechanism model in the original digital twin brain mechanism model through the following steps:

[0110] Analyze the data characteristics of the brain state information to determine the brain information data and the brain state evaluation results, and construct a correlation analysis data set based on the brain information data and the brain state evaluation results;

[0111] Extract strong association rules based on the association rule algorithm and the correlation analysis data set, and determine strong correlation data from the correlation analysis data set based on the strong association rules;

[0112] Perform parameter iterative update on the multi-level brain mechanism model through the strong correlation data.

[0113] Please refer to Figure 3 , Figure 3 which is a schematic structural diagram of an electronic device provided by an embodiment of the present application. As Figure 3 shown in

[0114] The memory 320 stores machine-readable instructions executable by the processor 310. When the electronic device 300 runs, the processor 310 communicates with the memory 320 through the bus 330. When the machine-readable instructions are executed by the processor 310, the steps of the method for constructing the digital twin brain mechanism model in the method embodiment as described above can be executed. The specific implementation manner can refer to the method embodiment and will not be elaborated here. Figure 1 shown.

[0115] An embodiment of the present application further provides a computer-readable storage medium. A computer program is stored on the computer-readable storage medium. When the computer program is run by a processor, the steps of the method for constructing the digital twin brain mechanism model in the method embodiment as described above can be executed. The specific implementation manner can refer to the method embodiment and will not be elaborated here. Figure 1 shown.

[0116] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, and units described above can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated here.

[0117] In several embodiments provided in this application, it should be understood that the disclosed systems, devices, and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some communication interfaces. The indirect coupling or communication connection of the devices or units can be in electrical, mechanical, or other forms.

[0118] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0119] In addition, in each embodiment of this application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit.

[0120] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a non-volatile computer-readable storage medium executable by a processor. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in each embodiment of this application. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks, or optical discs that can store program codes.

[0121] Finally, it should be noted that the above-described embodiments are only specific embodiments of the present application, which are used to illustrate the technical solutions of the present application, rather than limiting it. The protection scope of the present application is not limited thereto. Although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that any person skilled in the art within the technical scope disclosed by the present application can still modify the technical solutions described in the foregoing embodiments or can easily think of changes, or perform equivalent replacements on some of the technical features; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for constructing a digital twin brain mechanism model, characterized in that The construction method includes: Obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data; Based on the target signal data, construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model; Based on the complex brain function information driven by twin data, perform parameter iterative update on the original digital twin brain mechanism model to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model; The constructing a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model based on the target signal data includes: Based on the target signal data, construct a three-dimensional model describing multiple structural information and a function model describing multiple dynamic responses to obtain the multi-dimensional brain mechanism model; On the basis of the multi-dimensional brain mechanism model, construct the multi-modal brain mechanism model through extraction of key characteristic parameters of the multi-modal system and a modal fusion mechanism; On the basis of the multi-dimensional brain mechanism model and the multi-modal brain mechanism model, perform multi-level data fusion based on the theory of neuroimaging source analysis to obtain the multi-level brain mechanism model; wherein, the multi-level data fusion includes data-level fusion, feature-level fusion, and decision-level fusion.

2. The construction method according to claim 1, characterized in that, The aligning the multi-source brain state signal data in the time dimension and space dimension through a spatio-temporal alignment method and interpolation calculation, and performing missing data supplementation on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data includes: Align the multi-source brain state signal data in the time dimension through unified timestamps and interpolation calculation, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the space dimension; Adopt a generative network to construct an inversion algorithm based on an inverse generator, and perform inversion training using the correlation between multi-modal data to perform missing data supplementation on the multi-source brain state signal data to obtain the target signal data.

3. The construction method according to claim 1, characterized in that, Perform parameter iterative update on the multi-dimensional brain mechanism model in the original digital twin brain mechanism model through the following steps: Extract the structural features and functional features of the structural model and function model of the complex twin brain, and store the extracted features in a structured form; Update the structured form in real time to realize the synchronous update of the structured form and the multi-dimensional brain mechanism model.

4. The construction method according to claim 1, characterized in that, Perform parameter iterative update on the multi-modal brain mechanism model in the original digital twin brain mechanism model through the following steps: Extract neuroimaging data in multiple brain states under complex brain information based on a neuroimaging extraction algorithm to obtain multi-modal neuroimaging data; Update the multi-modal brain mechanism model according to the multi-modal neuroimaging data; Solve the multi-modal mechanism feature model and verify the accuracy of the multi-modal brain mechanism model, and modify the model and adjust the parameters of the multi-modal brain mechanism model according to the model accuracy verification results.

5. The construction method according to claim 1, characterized in that Iteratively update the parameters of the multi-level brain mechanism model in the original digital twin brain mechanism model through the following steps: Analyze the data characteristics of the brain state information to determine the brain information data and the brain state evaluation results, and construct a correlation analysis data set based on the brain information data and the brain state evaluation results; Extract strong association rules based on the association rule algorithm and the correlation analysis data set, and determine strong correlation data from the correlation analysis data set based on the strong association rules; Iteratively update the parameters of the multi-level brain mechanism model through the strong correlation data.

6. A device for constructing a digital twin brain mechanism model, characterized in that, The construction device includes: A target signal data determination module, configured to obtain multi-source brain state signal data, align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing supplement on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data; A model construction module, configured to construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model based on the target signal data, and perform model coupling on the multi-dimensional brain mechanism model, the multi-modal brain mechanism model, and the multi-level brain mechanism model to obtain an original digital twin brain mechanism model; A model update module, configured to iteratively update the parameters of the original digital twin brain mechanism model based on complex brain function information driven by twin data, so as to realize the evolution of the digital twin brain mechanism model driven by twin data and obtain a target digital twin brain mechanism model; When the model construction module is used to construct a multi-dimensional brain mechanism model, a multi-modal brain mechanism model, and a multi-level brain mechanism model based on the target signal data, the model construction module is further configured to: Construct a three-dimensional model describing multiple structural information and a functional model describing multiple dynamic responses based on the target signal data to obtain the multi-dimensional brain mechanism model; On the basis of the multi-dimensional brain mechanism model, construct the multi-modal brain mechanism model through extraction of key feature parameters of the multi-modal system and a modal fusion mechanism; On the basis of the multi-dimensional brain mechanism model and the multi-modal brain mechanism model, perform multi-level data fusion based on the theory of neuroimaging trace analysis to obtain the multi-level brain mechanism model; wherein, the multi-level data fusion includes data-level fusion, feature-level fusion, and decision-level fusion.

7. The construction device according to claim 6, characterized in that, When the target signal data determination module is used to align the multi-source brain state signal data in the time dimension and the space dimension through a spatio-temporal alignment method and interpolation calculation, and perform missing supplement on the multi-source brain state signal data based on an inversion algorithm to obtain target signal data, the target signal data determination module is further configured to: Align the multi-source brain state signal data in the time dimension through unified timestamps and interpolation calculations, and project the multi-source brain state signal data at the same moment into a high-resolution space to align the multi-source brain state signal data in the spatial dimension; Adopt a generative network, construct an inversion algorithm based on an inversion generator, and use the correlation between multi-modal data for inversion training to supplement the missing parts of the multi-source brain state signal data to obtain the target signal data.

8. An electronic device, characterized in that, It includes: A processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the electronic device runs, the processor communicates with the memory through the bus. When the machine-readable instructions are run by the processor, the steps of the method for constructing the digital twin brain mechanism model according to any one of claims 1 to 5 are executed.

9. A computer-readable storage medium, characterized in that, A computer program is stored on the computer-readable storage medium. When the computer program is run by the processor, the steps of the method for constructing the digital twin brain mechanism model according to any one of claims 1 to 5 are executed.

Images