A brain health comprehensive assessment system for the elderly

Abstract

The application belongs to the technical field of brain network topology analysis, and particularly relates to a brain health comprehensive evaluation system for the elderly, which comprises a data acquisition module, a brain network topology fusion preprocessing module, a multi-scale brain digital twin construction module, and a lesion coupling correlation evaluation module. The brain network topology fusion preprocessing module is used for denoising and normalizing preprocessing of brain macrostructure image data, mesovascular microcirculation data and microscale neural electrical activity data, and determining pathological topology abnormality features based on a brain feature benchmark threshold library of healthy people. The multi-scale brain digital twin construction module is used for constructing a brain digital twin based on a standard brain multi-scale topology template and the brain fusion topology data set. The lesion coupling correlation evaluation module is used for determining brain lesion coupling classification results and risk levels based on topology parameters of the brain digital twin and the brain feature benchmark threshold library of healthy people.

Description

Technical Field

[0001] This invention belongs to the field of brain network topology analysis technology, specifically relating to a comprehensive brain health assessment system for the elderly. Background Technology

[0002] The core pathological feature of neurodegenerative diseases is multi-scale, multi-level, and coupled progressive degeneration. The lesions are not simply damage to a single brain tissue, but rather a natural pathological process involving the interconnected and mutually reinforcing interaction of macroscopic brain tissue structural atrophy, mesoscopic cerebral vascular microcirculatory disturbances, and microscopic neural synaptic network disruptions. Blockage of the vascular microcirculation can induce ischemic apoptosis of neural units, while the decline in neural connectivity exacerbates functional atrophy of brain regions. This coupled interaction of multi-dimensional lesions is the core inducing factor for the rapid deterioration of brain diseases and a key basis for distinguishing between ordinary physiological aging and pathological lesions.

[0003] Existing intelligent brain health assessment technologies have several substantial technical shortcomings and cannot adapt to the natural pathological patterns of brain coupling degeneration, as follows: First, the assessment scale is too narrow to cover the characteristics of lesions across different levels. Most existing brain assessment models are based on single macroscopic structures and only detect organic lesions in brain tissue, ignoring early microscopic degeneration of mesoscopic vascular microcirculation and microscopic neural synaptic networks. These microscopic and mesoscopic abnormalities are the first symptoms of brain diseases, resulting in a very high rate of early missed detection with current technologies.

[0004] Second, it lacks the ability to analyze lesion coupling, and its assessment logic does not conform to pathological patterns. Traditional techniques assume that various brain lesions occur independently, without considering the interconnected deterioration relationship between blood vessels, nerves, and brain tissue. It cannot identify high-risk lesions caused by the coupling of multiple lesions, often resulting in the problem of "normal individual indicators but continuous overall deterioration" in assessment, leading to one-sided and inaccurate assessment results.

[0005] Third, the general models have poor adaptability and cannot accurately distinguish between physiological degeneration and pathological lesions. Existing digital twin brain models have a wide range of fixed parameters and do not adaptively correct for the physiological topological sparsity characteristics of different age groups. They are prone to misclassifying normal physiological aging of the brain as lesions or masking mild pathological coupling degeneration, resulting in a high rate of misjudgment.

[0006] Fourth, the extrapolation mechanism is simple and cannot quantify coupled degradation trends. Traditional time-series extrapolation only predicts changes in the size of a single lesion, and cannot simulate the dynamic process of multi-scale lesions coupling and progressively worsening. It also cannot quantify the degradation rate and associated risks, thus limiting its clinical guidance value.

[0007] In summary, there is currently a lack of an intelligent assessment system that adapts to the natural laws of multi-scale coupled degeneration in the brain, enables cross-level modeling, quantitative assessment of lesion coupling, and accurate inference of network degeneration. New technical solutions are urgently needed to address the aforementioned technical pain points in the industry. Summary of the Invention

[0008] In view of the above-mentioned shortcomings in the prior art, the present invention provides a comprehensive brain health assessment system for the elderly population to solve the problems mentioned in the background art.

[0009] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A comprehensive brain health assessment system for the elderly, including: The data acquisition module is used to collect macroscopic brain structural imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data. The brain network topology fusion preprocessing module is used to perform noise reduction and normalization preprocessing on macroscopic brain structure imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data, and to determine pathological topological abnormalities based on a baseline threshold library of brain features in healthy individuals; based on the pathological topological abnormalities and the deviation of various topological features from their corresponding healthy baseline thresholds, a brain fusion topology dataset is obtained. The multi-scale brain digital twin construction module is based on a standard multi-scale brain topology template and constructs a brain digital twin based on a brain fusion topology dataset. The lesion coupling correlation assessment module determines the brain lesion coupling classification results and risk levels based on the topological parameters of the brain digital twin and the baseline threshold library of brain characteristics of healthy individuals.

[0010] Furthermore, the macroscopic structural imaging data includes brain MRI grayscale images and CT tomographic images; the mesoscopic vascular microcirculation data includes cerebral basal vessel flow velocity, vascular dilation, and capillary perfusion; and the microscopic neural electrical activity data includes resting-state EEG rhythm, neural synaptic response intensity, and brain region synchronous activation data. The noise reduction and normalization preprocessing of macroscopic brain structural imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data includes: Gaussian filtering was used to denoise MRI grayscale images and CT tomographic images. Moving average combined with median filtering was used to denoise vascular flow velocity, perfusion volume, and vascular dilation time series data. Wavelet thresholding combined with power frequency notch filtering was used to denoise resting-state EEG, neural response, and brain region synchronous activation data. The denoised data were then unified to the [0,1] value range using the maximum-minimum normalization algorithm.

[0011] Furthermore, the first threshold based on brain characteristics of healthy individuals, used to determine pathological topological abnormality features, includes: Based on the denoised and normalized MRI / CT image structure data, brain anatomical region segmentation and topological structure calculation are performed to obtain macroscopic image topological features including brain region area change rate, tissue contour deformation coefficient, number of anatomical partition connecting edges, and brain region spatial connectivity. Based on the denoised and normalized vascular microcirculation time series data, the actions of vascular network node identification, blood flow parameter statistics, and topological connectivity calculation are performed to obtain the mesoscopic vascular network topological features including average vascular flow velocity, capillary perfusion, vascular dilation, vascular connectivity coefficient, and vascular node distribution density. Based on the denoised and normalized neural electrical activity data, the brain signal feature analysis, brain region synchronization calculation, and neural connection weight solution are performed to obtain microscopic neural network topology features including neural connection edge weights, brain region synchronous activation degree, neural synaptic response intensity, brain rhythm features, and phase lock values.

[0012] Furthermore, the method for determining pathological topological abnormality features based on a first threshold derived from brain characteristics of healthy individuals also includes: The system retrieves a pre-stored database of baseline thresholds for brain features in healthy individuals and compares them one by one with microscopic neural network topology features, mesoscopic vascular network topology features, and macroscopic imaging topology features. If the feature value of a topology feature exceeds the corresponding baseline threshold range, the topology feature is determined to be a pathological topological abnormality feature. If the feature value of a topology feature falls within the corresponding exclusive threshold range, the topology feature is determined to be a physiological normal feature.

[0013] Furthermore, the brain fusion topology dataset is obtained based on pathological topological anomaly features and the deviation of various topological features from their corresponding health benchmark thresholds, including: Initial weights are assigned to microscopic neural network topological features, mesoscopic vascular network topological features, and macroscopic imaging topological features. The deviation of each type of topological feature from its corresponding health benchmark threshold is calculated. The fusion weights are then dynamically updated based on the magnitude of the deviation, with increased weights for pathological features with high deviation and decreased weights for physiological features with low deviation. Finally, cross-scale topological feature weighted fusion is performed according to the updated weights to establish the association mapping between structural, vascular, and neural networks, resulting in a brain fusion topological dataset.

[0014] Furthermore, the steps for constructing the digital twin of the brain include: A pre-constructed standard brain multi-scale topology template is retrieved. The standard brain multi-scale topology template includes three sub-templates: a macroscopic tissue structure layer, a mesoscopic vascular microcirculation layer, and a microscopic neural synapse layer. The three sub-templates are pre-completed with global rigid registration and uniformly adopt the medical standard DICOM three-dimensional spatial coordinates. Align the spatial coordinates of the brain fusion topology dataset with the template coordinates; assign macroscopic image topological features to the anatomical nodes and tissue connection edges of the macroscopic tissue structure layer according to hierarchical rules; assign mesoscopic vascular network topological features to the vascular nodes and vascular segment connection edges of the mesoscopic vascular microcirculation layer; assign neural electrical features to the brain functional nodes and neural connection edges of the microscopic neural synapse layer. Bind anatomical nodes, vascular nodes, and brain function nodes in the same three-dimensional space; load the preset interlayer coupling association function, configure the parameter linkage rules between blood vessels, nerves, and brain tissue, and generate a digital twin of the brain after completing parameter assignment and linkage configuration.

[0015] Further, the step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library includes: The system acquires the topological parameters of the brain digital twin. Based on the brain region area and contour deformation coefficient of the macroscopic tissue structure layer, it determines whether the brain tissue has atrophied. Based on the blood vessel flow velocity, perfusion volume, and dilation of the mesoscopic vascular microcirculation layer, it determines whether the blood vessels are blocked. Based on the neural connection weight and brain region synchronous activation degree of the microscopic neural synapse layer, it determines whether the neural connections are broken. The system compares various parameters with the benchmark threshold library of healthy people. If the parameters are lower than the benchmark lower limit, the corresponding lesion is determined to be in an abnormal state, and the abnormal lesion combination is output. Furthermore, the step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library also includes; For any two types of abnormal lesions, the coupling correlation degree is calculated using the modified Pearson correlation coefficient. The calculation formula is as follows: In the formula, X and Y are the normalized topological parameter sequences of the two types of lesions. For covariance, D(Y) is the variance of the parameter. This is the clinical pathology correction factor; Determine coupling strength based on the range of correlation values: Uncoupled Slight coupling, The coupling is moderate. It is a strong coupling.

[0016] Furthermore, the step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library also includes; If brain tissue atrophy, vascular occlusion, and neural connection disruption are all abnormal simultaneously, a multi-lesion linkage risk calculation is initiated, using a weighted superposition model: In the formula , , These are three sets of pairwise coupling degrees. , , The pathological weighting coefficients are used, and the sum of the three is 1, ultimately yielding a comprehensive linkage risk value in the range [0,1]. .

[0017] Furthermore, based on the comprehensive linkage risk value, the specific components of the coupled risk classification are determined; No abnormal lesions or coupling degree less than 0.3 is classified as Level 1 (no risk); a single abnormal lesion or coupling degree in the range [0.3, 0.6) is classified as Level 2 (mild coupling risk); two types of lesions with coupling degree in the range [0.6, 0.85), or multiple lesions, are considered to have a risk value. Level III is considered moderate coupling risk; pairwise coupling degree ≥ 0.85 or multifocal risk value R ≥ 0.7 is considered level IV severe coupling risk.

[0018] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention constructs a three-dimensional independent detection system encompassing macroscopic imaging, mesoscopic vascular systems, and microscopic neural systems. It employs differentiated noise reduction strategies for different types of heterogeneous data and innovatively sets independent, dedicated benchmark threshold ranges for each of the three types of topological features. This abandons the traditional, coarse-grained approach of using a uniform threshold, accurately distinguishing between physiologically normal fluctuations within their respective threshold ranges and pathologically abnormal features deviating from these ranges. This fundamentally avoids the misjudgment of normal physiological degeneration signals as lesions or the masking of subtle pathological features by physiological noise. Furthermore, this invention precisely quantifies the deviation of features in each dimension using a normalized absolute deviation formula. Based on this deviation, it achieves linear adaptive dynamic weight adjustment, strengthening the weights of high-deviation pathological features and suppressing the weights of low-deviation physiological features. Combined with a cross-scale node association mapping matrix precisely bound to DICOM three-dimensional coordinates, it completes weighted fusion, greatly improving the targeting and fidelity of multi-scale feature fusion. This provides accurate, reliable, and interference-free raw feature data support for subsequent modeling, evaluation, and deduction.

[0019] 2. Compared to traditional single-structure or single-function brain modeling methods, this invention constructs an integrated digital twin architecture encompassing macroscopic tissue structure, mesoscopic vascular microcirculation, and microscopic neural synapses. Utilizing a unified DICOM 3D coordinate system, it achieves one-to-one binding and alignment of the three layers of topological nodes, ensuring absolute synchronization of the anatomical space. Simultaneously, it pioneers three sets of exclusive quantitative interlayer coupling correlation functions. Through linear calculations with fixed coefficients, it accurately simulates the physiological linkages and pathological deterioration patterns of brain tissue-blood vessels, blood vessels-nerves, and brain tissue-nerves. It can quantify and update the changes in the correlation parameters of the other two layers in real time based on changes in the topological parameters of a single layer, dynamically simulating the synergistic and progressive pathological process of brain tissue atrophy, vascular occlusion, and neural degeneration. Attached Figure Description

[0020] Figure 1 This is a schematic diagram illustrating the components of a comprehensive brain health assessment system for the elderly population according to the present invention. Figure 2 This is a schematic diagram of the process for obtaining a brain fusion topology dataset in one embodiment of the present invention; Figure 3 This is a schematic diagram illustrating the process of constructing and iteratively updating a multi-scale digital brain twin in one embodiment of the present invention; Figure 4 This is a schematic diagram of the process for quantitative assessment of lesion coupling correlation in one embodiment of the present invention. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be further described below with reference to embodiments.

[0022] Example 1: like Figures 1-4 As shown, a comprehensive brain health assessment system for the elderly includes: a data acquisition module, a brain network topology fusion preprocessing module, a multi-scale brain digital twin construction module, and a lesion coupling correlation assessment module. The data acquisition module is used to simultaneously acquire heterogeneous data adapted to the characteristics of brain lesions across scales, including brain structural imaging data, cerebral vascular microcirculation blood flow data, neural electrical activity data, and population-adapted cognitive behavioral data. Structural imaging data, including grayscale MRI images and CT tomographic images of the brain, are used to characterize the organic structural state of brain tissue. For example, brain MRI grayscale images can be obtained using a medical superconducting magnetic resonance imaging (MRI) scanner that scans the entire brain, primarily acquiring T1-weighted and T2-weighted grayscale images (the MRI grayscale images referred to here), and simultaneously outputting multi-view sequences of transverse, sagittal, and coronal images; focusing on key areas such as the cerebral cortex, ventricles, white matter, gray matter, cerebellum, and brainstem. The equipment natively outputs grayscale image sequences in DICOM standard format, preserving original structural features such as pixel grayscale values, spatial coordinates, and interslice spacing, for subsequent digital twin modeling of macroscopic brain tissue structures. The equipment transmits complete MRI grayscale images in batches / in real-time to the system's data acquisition module via the hospital's PACS image archiving system and DICOM communication interface. Brain CT tomographic images were acquired using a medical spiral CT scanner in plain head scan mode, with standard slice thickness set (1mm~5mm). Transverse tomographic images were continuously acquired, and sagittal and coronal reconstructed tomographic images were added as needed. This clearly presented organic structural features such as brain tissue density, intracranial cavities, bone, hemorrhage points, calcifications, and infarcts. The DICOM format CT tomographic sequence images were output, retaining key parameters such as tomographic location, tissue density values, and spatial scale, serving as the core data source for macroscopic identification of organic brain lesions. The data was connected to this system via the hospital's PACS system and a DICOM interface for data parsing and storage.

[0023] Microcirculatory blood flow data include cerebral vascular velocity, capillary perfusion, and vascular dilation data; For example, cerebral basal vessel flow velocity data can be obtained using transcranial Doppler ultrasound (TCD) to target and detect the main basal vessels such as the anterior cerebral artery, middle cerebral artery, and basilar artery, and to collect precise parameters such as systolic velocity, diastolic velocity, and mean blood flow velocity in real time, serving as static baseline data. Capillary perfusion data can be obtained using laser speckle flow imaging systems and functional near-infrared spectroscopy (fNIRS) to non-invasively detect the blood perfusion flux, perfusion volume per unit area, and blood flow perfusion rate of cerebral cortical capillaries, acquiring high-precision static parameters of microcirculation perfusion. Vascular dilation data can be obtained by combining cerebral vascular ultrasound detection with vascular pressure sensing equipment to collect end-diastolic and systolic vessel diameters, calculate the vasodilation coefficient and vascular compliance, and accurately obtain quantifiable parameters of vasodilation.

[0024] Microscopic neural electrical activity data includes resting-state brain electrical rhythms, synaptic response intensity, and brain region synchronous activation data; cognitive behavioral data includes assessment data on memory, executive function, language logic, and limb coordination. Resting-state EEG rhythms refer to the spontaneous, continuous electrical activity rhythms generated by clusters of neurons in the cerebral cortex in a subject who is awake, eyes closed, completely relaxed, and without any active thinking or motor tasks. This can be acquired using a medical digital electroencephalogram (EEG). The acquisition process involves the subject sitting quietly with eyes closed, maintaining physical and mental relaxation, and entering a standard resting state. The device acquires raw time-domain EEG waveforms of the entire brain according to the international 10-20 lead rule, with each acquisition session lasting at least 5 minutes. The device has a built-in frequency domain analysis module that uses Fast Fourier Transform (FFT) to divide the raw signal into frequencies and extract the absolute power, relative power, center frequency, and rhythm fluctuation coefficient for each frequency band: δ (0.5~3Hz), θ (4~7Hz), α (8~13Hz), β (14~30Hz), and γ (>30Hz).

[0025] Synaptic response intensity: Characterizing the transmission efficiency, response amplitude, and reaction speed of electrochemical signal transmission between neurons via synapses, it is a core indicator of the integrity of neural pathway function; synaptic damage and neurodegeneration directly lead to decreased response intensity and prolonged latency. Since the microscopic structure of synapses cannot be directly probed in the living human brain non-invasively, standardized external stimuli are applied to the subject, and the evoked electroencephalogram (EEG) response signals generated by the brain are detected. The response peak amplitude, signal latency, and peak area are used as equivalent quantitative indicators of synaptic response intensity.

[0026] Synchronous activation in brain regions: This quantifies the degree of coordination in the electrical activity of neuronal clusters in different functional areas of the brain (frontal lobe, temporal lobe, parietal lobe, occipital lobe, central motor area, etc.). It is characterized by the coherence coefficient, phase lock value (PLV), and cross-correlation coefficient of the EEG signal: higher values ​​indicate stronger neural network linkages between brain regions; brain lesions and neural network breaks will cause a significant decrease in synchronous activation. This indicator is a core parameter for constructing brain network topology.

[0027] The brain network topology fusion preprocessing module, connected to the data acquisition module, is used to perform hierarchical denoising, scale normalization, and topology feature extraction on cross-scale heterogeneous data. It also employs a brain network topology adaptive denoising fusion algorithm to remove interference noise from physiological degeneration of the brain, retain pathological topological abnormal features, and construct a unified brain structure-blood vessel-nerve fusion topology dataset. The brain network topology fusion preprocessing module uses Gaussian filtering to denoise MRI grayscale images and CT tomographic images, moving average combined with median filtering to denoise vascular flow velocity, perfusion volume, and vascular dilation time series data, and wavelet thresholding combined with power frequency notch filtering to denoise resting-state EEG, neural response, and brain region synchronous activation data. After denoising, all features are unified to the [0,1] value range using the maximum-minimum normalization algorithm, and macroscopic image topology features, mesoscopic vascular network topology features, and microscopic neural network topology features are extracted respectively. The brain network topology adaptive denoising and fusion algorithm is further run. The pre-stored baseline threshold library of brain features of healthy people is retrieved. Features falling within the threshold range are identified as physiological degeneration interference noise and removed. Features exceeding the threshold range are identified as pathological topological abnormalities and retained. The algorithm dynamically adjusts the fusion weights of the three types of data according to the deviation of the features from the baseline threshold, and completes the cross-scale feature association mapping by weighting. Finally, a unified brain structure-blood vessel-nerve fusion topology dataset is constructed. For example, Gaussian filtering with a size of 3×3 to 7×7 is used for MRI grayscale images and CT tomographic images to remove image noise and motion artifacts; for time-series data such as cerebral vascular flow velocity, capillary perfusion, and vascular dilation, a moving average filter with a window length of 5 to 15 sampling points combined with median filtering is used to remove transient abnormal fluctuations; for neuroelectric data such as resting-state EEG rhythm, synaptic response intensity, and synchronous activation of brain regions, db4 wavelet basis soft threshold wavelet denoising is used, and 50Hz power frequency notch filtering is superimposed to filter out environmental interference and electromyographic artifacts.

[0028] The maximum-minimum normalization algorithm is used to uniformly map all denoised feature data to the [0,1] interval, eliminating the differences in the units and value ranges of different types of data. The normalization calculation is performed based on the maximum and minimum values ​​of the healthy population features built into the system.

[0029] Macroscopic topological features such as brain region morphology, area, and spatial connectivity are extracted from imaging data; mesoscopic topological features such as vascular nodes, blood flow distribution, and vessel diameter connectivity are extracted from vascular data; and microscopic topological features such as brain electrical rhythm, response amplitude, and brain region synchronization matrix are extracted from neuroelectrical data.

[0030] Physiological noise removal: The system retrieves a pre-stored baseline threshold database of brain features from healthy individuals and compares each extracted topological feature with the baseline threshold value. If the feature value is within the baseline threshold range, it is identified as physiological degeneration interference noise and removed. If the feature value exceeds the baseline threshold range, it is identified as pathological topological abnormality features and retained.

[0031] First, initial weights are assigned to the three categories of topological features: macroscopic, mesoscopic, and microscopic. Second, the deviation of each type of feature from the health baseline threshold is calculated. Then, the fusion weights are dynamically updated according to the magnitude of the deviation, with increased weights for pathological features with high deviation and decreased weights for physiological features with low deviation. Finally, cross-scale topological feature weighted fusion is completed according to the updated weights to establish the association mapping of three types of networks: structural, vascular, and neural. Ultimately, a standardized, high-fidelity brain fusion topological dataset is generated and input into the multi-scale brain digital twin construction module.

[0032] Based on the real-time detection values ​​of topological features in each dimension and the standard benchmark intervals corresponding to the benchmark threshold library of brain features in healthy individuals, the benchmark threshold library of brain features in healthy individuals contains independent macroscopic-specific threshold intervals, mesoscopic-specific threshold intervals, and microscopic-specific threshold intervals. The three types of topological features correspond one-to-one with their respective exclusive threshold intervals and do not share threshold ranges. Topological features whose detection values ​​fall within their corresponding exclusive threshold intervals, are caused only by normal physiological aging and minor physiological fluctuations, and have no pathological progression trend are defined as physiologically normal features. Topological features whose detection values ​​deviate from their corresponding exclusive threshold intervals, are caused by brain lesions and malformations, and have a pathological deterioration trend are defined as pathologically abnormal features.

[0033] The overall deviation of a single-class feature is calculated using the normalized absolute deviation formula, which is: In the formula, X is the current feature detection value, Xmean is the mean of the healthy baseline interval, Xmax and Xmin are the upper and lower limits of the healthy baseline interval, respectively, and D is the normalized feature deviation, with a value range of [0,1]. Dynamic weight updates are performed based on the calculated deviations of the three types of features. The initial state is that the weights of the macroscopic, mesoscopic, and microscopic features are equally assigned as 1 / 3, and the total weight is always 1. Physiological normal features refer to the real-time detection values ​​of macroscopic image topology features, mesoscopic vascular network topology features, and microscopic neural network topology features, which fall within the standard normal interval corresponding to the brain feature baseline threshold library of healthy people. These features only come from normal brain aging and daily physiological fluctuations, without lesions or pathological progression trends, and are considered invalid interference features, with the corresponding feature deviation D approaching 0.

[0034] Real-time detection values ​​of pathological abnormality features, including macroscopic imaging topological features, mesoscopic vascular network topological features, and microscopic neural network topological features, are topological features that deviate from the standard normal range (above the upper limit or below the lower limit) corresponding to the baseline threshold library of brain features in healthy individuals. These features are generated by real lesions such as brain tissue atrophy, vascular occlusion, and broken neural connections, and have a trend of pathological progression and coupling deterioration. They are valid pathological features, and the larger the deviation degree D value of the corresponding feature, the more severe the pathological abnormality. For pathological abnormal features with higher deviation, the corresponding dimension fusion weight is increased linearly in the positive direction; for physiological normal features with lower deviation, the corresponding dimension fusion weight is decreased linearly, thereby strengthening the weight of pathological features and suppressing the weight of physiological noise. Based on the updated real-time weight matrix, cross-scale node association matching mapping of structural topology, vascular topology, and neural topology is performed to establish a one-to-one correspondence between macroscopic structural nodes, mesoscopic vascular nodes, and microscopic neural functional nodes at the same anatomical location, forming a three-dimensional network coupling association mapping matrix. Finally, weighted fusion is performed to obtain a standardized, high-fidelity brain fusion topology dataset.

[0035] The multi-scale brain digital twin construction module is connected to the brain network topology fusion preprocessing module. It is used to build a three-layer linkage digital twin based on the standard brain multi-scale topology template and the fusion topology dataset, namely the macroscopic tissue structure layer, the mesoscopic vascular microcirculation layer, and the microscopic neural synapse layer, so as to realize the 1:1 real-time mapping of the real brain's cross-scale physical features, network connection relationships, and dynamic operating status. For example, the multi-scale brain digital twin construction module is connected to the brain network topology fusion preprocessing module. The standard multi-scale brain topology template is a three-layer integrated benchmark template with pre-complete spatial registration, uniformly adopting the DICOM three-dimensional coordinate system, including a geometric topology framework and a standard parameter library for healthy individuals. After loading the standard multi-scale topology template, the module completes coordinate alignment of the fused topology dataset through rigid registration, and maps topological features and quantitative parameters to the macroscopic tissue structure layer, mesoscopic vascular microcirculation layer, and microscopic neural synapse layer according to their levels. Among them, the macroscopic tissue structure layer carries brain anatomical structure and imaging features, the mesoscopic vascular microcirculation layer carries vascular blood flow and diameter features, and the microscopic neural synapse layer carries neurophysiological and brain network features. The module establishes the static spatial binding relationship and dynamic parameter coupling and linkage relationship of the three-layer model, realizes cross-level parameter synchronous update based on inter-layer association rules, and finally instantiates and generates a three-layer linked brain digital twin. The steps involved in constructing and iteratively updating a multi-scale digital twin of the brain include: The first step is to load the standard topology template: The system retrieves the pre-constructed standard brain multi-scale topology template, which is generated based on MRI, CT, cerebral vascular and EEG data of a large sample of healthy people. It is divided into three sub-templates: macroscopic tissue structure layer, mesoscopic vascular microcirculation layer and microscopic neural synapse layer. The three sub-templates have completed global rigid registration in advance, uniformly adopt the medical standard DICOM three-dimensional spatial coordinates, and have a built-in parameter benchmark library of healthy people.

[0036] The second step, coordinate registration and feature mapping: Receive the fused topology dataset output by the brain network topology fusion preprocessing module, and use a rigid registration algorithm to align the spatial coordinates of the dataset with the template coordinates; perform feature mapping according to hierarchical rules: assign the structural topology features corresponding to MRI and CT to the anatomical nodes and tissue connection edges of the macroscopic tissue structure layer; assign microcirculation features such as vascular flow velocity, capillary perfusion, and vascular dilation to the vascular nodes and vascular segment connection edges of the mesoscopic vascular microcirculation layer; assign neuroelectric features such as resting-state EEG rhythm, synaptic response intensity, and synchronous activation degree of brain regions to the brain functional nodes and neural connection edges of the microscopic neural synapse layer.

[0037] The third step is to configure a three-layer linkage mechanism: First, execute static spatial linkage binding, which forces the macroscopic, mesoscopic, and microscopic topological nodes under the same three-dimensional spatial location to ensure the synchronization of anatomical positions; second, enable dynamic parameter coupling linkage, load the preset interlayer coupling association function, configure the parameter linkage rules between blood vessels, nerves, and brain tissues, and simulate the real pathological evolution.

[0038] The multi-scale brain digital twin construction module, based on a three-layer topological sub-template of the unified DICOM three-dimensional coordinate system and a fused topological dataset that completes cross-scale correlation mapping, performs precise spatial node binding, inter-layer parameter coupling correlation function configuration, and dynamic and static dual-state linkage matching actions. Specifically, the inter-layer coupling correlation functions and the specific parameter linkage rules and quantification calculation methods for blood vessels, nerves, and brain tissue are as follows: Based on the unified DICOM three-dimensional coordinates (x, y, z), macroscopic structural nodes, mesoscopic vascular nodes, and microscopic neural functional nodes corresponding to the same anatomical location are forcibly bound and paired, achieving one-to-one overlap of the spatial coordinates of the three-layer topological nodes and completing static spatial synchronization configuration. Three types of exclusive, quantifiable inter-layer coupling correlation functions are pre-encapsulated: brain tissue-blood vessel coupling correlation function, blood vessel-nerve coupling correlation function, and brain tissue-nerve coupling correlation function. Each function has a built-in fixed linkage influence coefficient used to accurately quantify the synchronous change amplitude of topological parameters at different levels. The specific quantification rules are as follows: First, the brain tissue-blood vessel coupling correlation function uses a negative linear coupling formula: In the formula, Vout represents the updated vascular microcirculation parameters, V0 represents the baseline value of vascular parameters, α represents the brain tissue linkage attenuation coefficient, which is fixed at 0.6, and Sdef represents the comprehensive deformation coefficient of brain tissue contour and brain region area atrophy rate. When the degree of brain tissue deformation and atrophy increases, the attenuated vascular dilation and capillary perfusion parameters are automatically calculated, realizing the linkage evolution of brain tissue lesions negatively dragging down vascular microcirculation. Secondly, the vascular-neural coupling correlation function adopts a positive linear coupling formula: In the formula, Nout represents the updated neural function parameters, β is the vascular blood supply linkage coefficient, fixed at 0.75, and Vnow is the normalized combined value of the current vascular flow velocity and capillary perfusion. When vascular blood flow parameters decrease, the intensity of neural synaptic response and the synchronous activation degree of brain regions are synchronously and linearly decreased, thus reproducing the pathological pattern of neural function decline caused by insufficient blood supply. Thirdly, the brain tissue-neuron coupling correlation function adopts a negative coupling formula: In the formula, Nnew represents the updated neural connectivity parameters, N0 represents the baseline value of the neural parameters, γ represents the structural damage linkage coefficient, which is fixed at 0.65, and Sloss represents the quantitative value of the brain region structural connectivity loss. When the connectivity of brain tissue structures decreases or the region shrinks, the weights of local neural connection edges and neural synchronization parameters are weakened accordingly. The coefficients of the three types of coupling correlation functions are all fixed constants preset by the system and do not change with the scene. When any single-level topology parameter undergoes abnormal changes, it is immediately substituted into the corresponding coupling function to solve the updated values ​​of the other two layers of correlation parameters in real time, and the topology state of the three-layer model is dynamically refreshed. At the static level, the anatomical spatial position of the three-layer model is completely synchronized and aligned. At the dynamic level, the collaborative pathological evolution process of brain tissue, blood vessels, and nerves is simulated in real time, and finally a coupled brain digital twin with cross-scale spatial alignment and parameter linkage evolution is obtained. The fourth step is model instantiation: After completing parameter assignment and linkage configuration, a three-layer linked brain digital twin that can run in real time is generated. The fifth step is periodic iterative updates: For dynamic time-series data such as blood flow, EEG, and cognition, daily real-time iterations are performed; for static imaging structural data such as MRI and CT, monthly batch iterations are performed; when new data is introduced, parameters are automatically remapped, and the three-layer model linkage updates are triggered simultaneously to ensure that the digital twin and the user's real brain state remain dynamically consistent.

[0039] The lesion coupling correlation assessment module is connected to the multi-scale brain digital twin construction module. It is used to retrieve cross-scale topological parameters of the twin and, through the brain lesion coupling correlation mining algorithm, quantify and identify the pairwise coupling correlation degree and multi-lesion linkage risk of vascular occlusion, brain tissue atrophy, and neural connection breakage. It completes the classification assessment and coupling risk classification of Alzheimer's disease, vascular dementia, multiple lacunar infarctions, and brain degenerative atrophy. S4. Lesion Coupling Correlation Quantitative Assessment: Step 1, Cross-scale Lesion Parameter Extraction and Initial Judgment of Single Lesion Abnormalities: Retrieve the topological parameters of the three-layer digital twin. Extract parameters such as brain region area and contour deformation coefficient from the macroscopic tissue structure layer to determine brain tissue atrophy. Extract parameters such as blood vessel flow velocity, perfusion volume, and dilation from the mesoscopic vascular microcirculation layer to determine blood vessel blockage. Extract parameters such as neural connection weight and brain region synchronous activation degree from the microscopic neural synapse layer to determine neural connection breakage. Compare various parameters with the system's built-in healthy population benchmark threshold library. If the parameters are lower than the benchmark lower limit, the corresponding lesion is determined to be in an abnormal state, and the abnormal lesion combination is output.

[0040] The second step is to calculate the coupling correlation between pairs of lesions: For any two types of abnormal lesions, the modified Pearson correlation coefficient is used to calculate the coupling correlation. The calculation formula is as follows:

[0041] In the formula, X and Y are the normalized topological parameter sequences of the two types of lesions. For covariance, D(Y) is the variance of the parameter. The clinical pathology correction coefficient is used; the coupling strength is determined based on the correlation range. Uncoupled Slight coupling, The coupling is moderate. It is a strong coupling.

[0042] The third step is the calculation of the risk of multiple lesions: If brain tissue atrophy, vascular occlusion, and neural connection disruption are all abnormal at the same time, the calculation of the risk of multiple lesions is initiated, using a weighted superposition model. In the formula , , These are three sets of pairwise coupling degrees. , , The pathological weighting coefficients are used, and the sum of the three is 1, ultimately yielding a comprehensive linkage risk value in the range [0,1]. .

[0043] The fourth step is disease classification assessment: classification is performed based on the combination and coupling characteristics of abnormal lesions: only the neural connection is broken and the nerve is strongly coupled with the brain tissue, which is identified as Alzheimer's disease; only the blood vessel is blocked and the blood vessel is strongly coupled with the nerve, which is identified as vascular dementia; the main feature is blood vessel blockage and local blood vessels are weakly coupled in multiple areas, which is identified as multiple lacunar infarctions; the main feature is brain tissue atrophy, accompanied by mild neurodegeneration and no strong vascular coupling, which is identified as degenerative brain atrophy; if all three types of lesions are abnormal at the same time, it is identified as mixed type of degenerative brain disease.

[0044] Step 5, Coupling Risk Classification: Based on the pairwise coupling correlation and comprehensive linkage risk value, four levels of risk are classified: Level 1 (no risk) is defined as no abnormal lesions or a coupling degree less than 0.3; Level 2 (mild coupling risk) is defined as a single abnormal lesion or a coupling degree in the range [0.3, 0.6); Level 3 (risk value) is defined as two types of coupled lesions with a coupling degree in the range [0.6, 0.85], or multiple lesions. Level III is considered moderate coupling risk; pairwise coupling degree ≥ 0.85 or multifocal risk value R ≥ 0.7 is considered level IV severe coupling risk.

[0045] Step 6, Report Generation: Integrate abnormal lesion results, coupling correlation, linkage risk value, disease classification, and risk level, along with pathological interpretation and intervention recommendations, to generate a standardized brain health assessment report.

[0046] The brain network degeneration prediction module is connected to the lesion coupling and correlation assessment module. It is used to fit the natural degeneration law of brain network topology based on the temporal topological change data of multi-scale twins, predict the progressive development trend of brain connection network breakage, microcirculation failure and neurological function decline, and quantify the brain network degeneration level and the probability of lesion deterioration in the future preset period. S5. Brain Network Topological Degeneration Trend Inference: Step 1, Time-Series Dataset Construction: Retrieve time-series topological change data from multi-scale brain digital twins for at least 6 consecutive months. The dataset includes daily sampled brain connectivity network parameters, vascular microcirculation parameters, and neural function parameters, as well as monthly updated structural feature data, forming a complete long-term time-series sample set. Step 2, Degeneration Pattern Fitting: Use an exponential decay model to fit the natural degradation pattern of brain network topology. The model formula is as follows: Simultaneously, time-series samples are input into an LSTM (Long Short-Term Memory) network for training, learning the decay characteristics and cross-scale linkage relationships of three types of topological parameters. The third step is progressive trend deduction: based on the trained network, trend prediction is performed. On one hand, changes in indicators within a single scale are monitored to simulate the single-dimensional progressive degradation process of brain connectivity network breakage, microcirculatory failure, and neurological function decline; on the other hand, relying on inter-layer parameter coupling rules, the overall development trend of multi-scale parameter linkage degradation is simulated. The fourth step is degradation level determination: various topological indicators within the prediction period are extracted, and a weighted comprehensive degradation index S is calculated. Based on the value range of S, four levels are classified: Level 1 (normal), Level 2 (mild degradation), Level 3 (moderate degradation), and Level 4 (severe degradation). The fifth step is lesion deterioration probability calculation: based on historical data, the average degradation rate V is calculated, combined with the deviation D of the predicted index from the healthy baseline threshold, according to the formula... Calculate the probability of disease progression. Step 6: Multi-cycle prediction and prognostic simulation: Output the degeneration trend, degeneration level, and probability of progression for three preset cycles of 3 months, 6 months, and 12 months respectively; At the same time, substitute the intervention correction parameters, simulate the repair effect of topological indicators under intervention methods such as drug treatment and rehabilitation training, and generate a personalized prognostic simulation plan.

[0047] The hierarchical early warning visualization and interactive module is used to receive lesion coupling assessment results and network degradation simulation data. It divides the early warning into multiple levels according to the coupling risk level and degradation speed. At the same time, it visualizes and displays multi-scale brain twin models, lesion coupling regions, and network topology degradation trajectories. It supports online interactive query and report export for users, family members, and medical staff. S6, tiered early warning, visualization and human-computer interaction: The first step is indicator quantification and joint judgment: The module receives the lesion coupling assessment results and network degradation simulation data, extracts the lesion coupling risk level, and divides the degradation speed into four levels: low speed, medium speed, high speed, and extreme speed according to the average degradation rate V obtained from the time series simulation; according to the joint judgment rules of coupling risk level and degradation speed, it automatically divides the warning into four levels: normal warning, mild warning, moderate warning, and emergency warning.

[0048] The second step is tiered information push and suggestion matching: When the alert is determined to be Level 1 (normal), a pop-up reminder and daily maintenance suggestions are pushed only to the user on the local terminal; when the alert is determined to be Level 2 (mild warning), a local pop-up and mobile mini-program message push are triggered to push a regular check-up reminder to the user; when the alert is determined to be Level 3 (moderate warning), multiple channels such as pop-ups, SMS, and mobile terminals are activated to simultaneously inform the user and their family, along with home intervention plans and suggested check-up cycles; when the alert is determined to be Level 4 (emergency warning), voice broadcasts, SMS, and APP reminders are activated to simultaneously push to the user, their family, and the linked medical terminals, and emergency medical treatment guidance is automatically pushed.

[0049] The third step is multi-format visualization: The system loads a three-layer digital twin of the brain, highlights the lesion area and the lesion coupling related area, and supports model rotation, scaling, and layered display; plays dynamic simulation animations of brain network degeneration in the next 3 months, 6 months, and 12 months; synchronously displays two-dimensional time series charts such as coupling risk curves and degeneration rate curves, and displays quantitative values ​​such as coupling correlation degree, comprehensive degeneration index, and probability of lesion deterioration on the data panel.

[0050] The fourth step is tiered human-computer interaction: Operational permissions are divided into three levels: user, family member, and medical staff. All users can switch model perspectives, view charts, and query historical warning records. Users can preview and download assessment reports online; family members can only view warning information and regression trends; medical staff have additional permissions to review historical twin models, edit treatment plans, and interact via online messaging. The fifth step is log recording: All warning events and interactive operations are automatically stored in cloud logs, enabling full traceability.

[0051] The privacy-encrypted storage and backtracking module is used to encrypt and store all cross-scale raw data, multi-scale twin model iteration data, lesion coupling assessment reports, and degeneration inference time series data. It supports long-term data backtracking comparison and model iteration optimization, enabling full-cycle traceability and tracking of brain health status.

[0052] The multi-scale brain digital twin construction module includes a general topology template matching unit, a three-layer scale linkage modeling unit, a topology parameter iterative update unit, and a cross-scale association binding unit; A general topology template matching unit is used to pre-construct three-layer standard topology templates of macroscopic structure, mesoscopic blood vessels, and microscopic nerves based on large-sample brain samples from healthy individuals, adapting to the physiological topological degeneration characteristics of the brains of different age groups. The three-layer scale linkage modeling unit is used to map the preprocessed cross-scale topological features to three-layer standard templates, independently construct twin models at each scale, and establish inter-layer parameter linkage mapping relationships. The topology parameter iterative update unit is used to iteratively update the topology node parameters, connection edge weights, blood perfusion parameters, and neural activation parameters of the multi-scale twin model on a weekly and monthly basis, ensuring that the twin and the real brain network state are synchronized in real time. Cross-scale correlation binding units are used to bind the correlation mapping relationship between macroscopic tissue deformation parameters, mesoscopic vascular blood flow parameters, and microscopic neural connectivity parameters, realizing the simulation basis for single-scale lesions triggering multi-scale linkage responses.

[0053] The above are merely embodiments of the present invention. The circuits, electronic components, and modules involved are all prior art, fully achievable by those skilled in the art, and require no further explanation. The scope of protection in this application does not involve improvements to the software and methods. Commonly known structures and characteristics in the solutions are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all prior art in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, under the guidance of this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the application.

Claims

1. A comprehensive brain health assessment system for the elderly, characterized in that: include; The data acquisition module is used to collect macroscopic brain structural imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data. The brain network topology fusion preprocessing module is used to perform noise reduction and normalization preprocessing on macroscopic brain structure imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data, and to determine pathological topological abnormality features based on a baseline threshold library of brain features in healthy individuals. Based on the pathological topological anomalies and the deviation of various topological features from their corresponding health baseline thresholds, a brain fusion topology dataset is obtained. The multi-scale brain digital twin construction module is based on a standard multi-scale brain topology template and constructs a brain digital twin based on a brain fusion topology dataset. The lesion coupling correlation assessment module determines the brain lesion coupling classification results and risk levels based on the topological parameters of the brain digital twin and the baseline threshold library of brain characteristics of healthy individuals.

2. The comprehensive brain health assessment system for the elderly as described in claim 1, characterized in that: The macroscopic structural imaging data includes brain MRI grayscale images and CT tomographic images; the mesoscopic vascular microcirculation data includes cerebral basal vessel flow velocity, vascular dilation, and capillary perfusion; and the microscopic neural electrical activity data includes resting-state EEG rhythm, neural synaptic response intensity, and brain region synchronous activation data. The noise reduction and normalization preprocessing of macroscopic brain structural imaging data, mesoscopic vascular microcirculation data, and microscopic neural electrical activity data includes: Gaussian filtering was used to denoise MRI grayscale images and CT tomographic images. Moving average combined with median filtering was used to denoise vascular flow velocity, perfusion volume and vascular dilation time series data. Wavelet thresholding combined with power frequency notch filtering was used to denoise resting state EEG, neural response and brain region synchronous activation data. The denoised data is normalized to the range of [0,1] using the maximum-minimum normalization algorithm.

3. The comprehensive brain health assessment system for the elderly as described in claim 2, characterized in that: The first threshold based on brain characteristics of healthy individuals is used to determine the characteristics of pathological topological anomalies, including: Based on the denoised and normalized MRI / CT image structure data, brain anatomical region segmentation and topological structure calculation are performed to obtain macroscopic image topological features including brain region area change rate, tissue contour deformation coefficient, number of anatomical partition connecting edges, and brain region spatial connectivity. Based on the denoised and normalized vascular microcirculation time series data, the actions of vascular network node identification, blood flow parameter statistics, and topological connectivity calculation are performed to obtain the mesoscopic vascular network topological features including average vascular flow velocity, capillary perfusion, vascular dilation, vascular connectivity coefficient, and vascular node distribution density. Based on the denoised and normalized neural electrical activity data, the brain signal feature analysis, brain region synchronization calculation, and neural connection weight solution are performed to obtain microscopic neural network topology features including neural connection edge weights, brain region synchronous activation degree, neural synaptic response intensity, brain rhythm features, and phase lock values.

4. The comprehensive brain health assessment system for the elderly as described in claim 3, characterized in that: The first threshold based on brain characteristics of healthy individuals is used to determine pathological topological abnormality features; it also includes: The system retrieves a pre-stored database of baseline thresholds for brain features in healthy individuals and compares them one by one with microscopic neural network topology features, mesoscopic vascular network topology features, and macroscopic imaging topology features. If the feature value of a topology feature exceeds the corresponding baseline threshold range, the topology feature is determined to be a pathological topological abnormality feature. If the feature value of a topology feature falls within the corresponding exclusive threshold range, the topology feature is determined to be a physiological normal feature.

5. A comprehensive brain health assessment system for the elderly as described in claim 4, characterized in that: The brain fusion topology dataset is obtained based on pathological topological anomaly features and the deviation of various topological features from their corresponding health benchmark thresholds, including: Initial weights are assigned to the microscopic neural network topology features, the mesoscopic vascular network topology features, and the macroscopic image topology features; Calculate the deviation of various topological features from their corresponding health benchmark thresholds; then dynamically update the fusion weights based on the magnitude of the deviation, increasing the weights for pathological features with high deviations and decreasing the weights for physiological features with low deviations. Finally, cross-scale topological feature weighted fusion is performed according to the updated weights to establish the association mapping of three types of networks: structural, vascular, and neural, resulting in a brain fusion topological dataset.

6. The comprehensive brain health assessment system for the elderly as described in claim 1, characterized in that: The steps for constructing the brain digital twin include: A pre-constructed standard brain multi-scale topology template is retrieved. The standard brain multi-scale topology template includes three sub-templates: a macroscopic tissue structure layer, a mesoscopic vascular microcirculation layer, and a microscopic neural synapse layer. The three sub-templates are pre-completed with global rigid registration and uniformly adopt the medical standard DICOM three-dimensional spatial coordinates. Align the spatial coordinates of the brain fusion topology dataset with the template coordinates; assign macroscopic image topological features to the anatomical nodes and tissue connection edges of the macroscopic tissue structure layer according to hierarchical rules; assign mesoscopic vascular network topological features to the vascular nodes and vascular segment connection edges of the mesoscopic vascular microcirculation layer; assign neural electrical features to the brain functional nodes and neural connection edges of the microscopic neural synapse layer. Bind anatomical nodes, vascular nodes, and brain function nodes in the same three-dimensional space; load the preset interlayer coupling association function, configure the parameter linkage rules between blood vessels, nerves, and brain tissue, and generate a digital twin of the brain after completing parameter assignment and linkage configuration.

7. A comprehensive brain health assessment system for the elderly as described in claim 6, characterized in that: The step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library includes: The system acquires the topological parameters of the brain digital twin. Based on the brain region area and contour deformation coefficient of the macroscopic tissue structure layer, it determines whether the brain tissue has atrophied. Based on the blood vessel flow velocity, perfusion volume, and dilation of the mesoscopic vascular microcirculation layer, it determines whether the blood vessels are blocked. Based on the neural connection weight and brain region synchronous activation degree of the microscopic neural synapse layer, it determines whether the neural connections are broken. The system compares various parameters with the benchmark threshold library of healthy people. If the parameter is lower than the benchmark lower limit, the corresponding lesion is determined to be in an abnormal state, and the abnormal lesion combination is output.

8. A comprehensive brain health assessment system for the elderly as described in claim 7, characterized in that: The step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library also includes: For any two types of abnormal lesions, the coupling correlation degree is calculated using the modified Pearson correlation coefficient. The calculation formula is as follows: In the formula, X and Y are the normalized topological parameter sequences of the two types of lesions. For covariance, D(Y) is the variance of the parameter. This is the clinical pathology correction factor; Determine coupling strength based on the range of correlation values: Uncoupled Slight coupling, The coupling is moderate. It is a strong coupling.

9. A comprehensive brain health assessment system for the elderly as described in claim 8, characterized in that: The step of determining the brain lesion coupling classification result based on the cross-scale topological parameters of the brain digital twin and the health benchmark threshold library also includes: If brain tissue atrophy, vascular occlusion, and neural connection disruption are all abnormal simultaneously, a multi-lesion linkage risk calculation is initiated, using a weighted superposition model: In the formula , , These are three sets of pairwise coupling degrees. , , The pathological weighting coefficients are used, and the sum of the three is 1, ultimately yielding a comprehensive linkage risk value in the range [0,1]. .

10. A comprehensive brain health assessment system for the elderly as described in claim 9, characterized in that: Based on the comprehensive linkage risk value, the specific components of the coupled risk classification are determined; No abnormal lesions or coupling degree less than 0.3 is classified as Level 1 (no risk); a single abnormal lesion or coupling degree in the range [0.3, 0.6) is classified as Level 2 (mild coupling risk); two types of lesions with coupling degree in the range [0.6, 0.85), or multiple lesions, are considered to have a risk value. Level III is considered moderate coupling risk; pairwise coupling degree ≥ 0.85 or multifocal risk value R ≥ 0.7 is considered level IV severe coupling risk.

Images