Alzheimer's Disease Diagnosis and Prediction Method and System Guided by Multi-Source Knowledge

Abstract

This invention discloses a multi-source knowledge-guided method for predicting and diagnosing Alzheimer's disease, belonging to the field of medical artificial intelligence-assisted technology. It includes: acquiring participants' MRI images, brain statistics, and clinical data; processing the MRI images using a dual-path feature extraction strategy to obtain global and local brain features and perform collaborative learning to generate fused visual features; fusing the brain statistics and the fused visual features and inputting them into a trained diagnostic module for processing to obtain the current diagnostic result; using the clinical data as guidance, combined with the current diagnostic result, and a trained prediction module, to obtain the disease progression at future moments. This method integrates multimodal data, improving the accuracy and comprehensiveness of Alzheimer's disease diagnosis and prediction, and solving the problem of low accuracy in existing Alzheimer's disease diagnoses, which fails to meet clinical needs.

Description

Technical Field

[0001] This invention relates to the field of medical artificial intelligence-assisted technology, and in particular to a method and system for diagnosing and predicting Alzheimer's disease based on multi-source knowledge guidance. Background Technology

[0002] The statements in this section merely refer to the background art related to this invention and do not necessarily constitute prior art.

[0003] Alzheimer's disease (AD) is an irreversible neurodegenerative disease characterized by progressive memory decline and cognitive impairment over time. This process is typically accompanied by brain tissue atrophy, pathological β-amyloid plaque formation, and abnormal brain metabolism. Since current drug treatments cannot halt the progression of AD, attention has turned to mild cognitive impairment (MCI), a critical stage that precedes the onset of AD. MCI is considered a key transitional stage from normal aging to AD, highlighting its importance for early AD detection through identification and monitoring. However, the diagnosis and intervention of both AD and MCI face significant challenges due to subjective variability among experts.

[0004] Fortunately, multimodal neuroimaging techniques such as magnetic resonance imaging (MRI) and positron emission tomography (PET) combined with fluorescence deoxyglucose PET (FDG-PET) can facilitate diagnosis. Furthermore, various clinical indicators, including age, sex, blood pressure, and objective cognitive function assessments obtained through scales such as the Clinical Dementia Rating Scale (CDRSB) and ADAS-Cog13, can improve the accuracy of AD prediction. For example, apolipoprotein E4 (apoE4), a key clinical biomarker found in more than half of AD patients, indicates disease risk. Therefore, there is an urgent need for objective and effective methods for predicting AD progression based on multi-source knowledge.

[0005] Current research on AI-based diagnostic and prognostic methods for Alzheimer's disease primarily focuses on computer vision data-driven techniques, with most systems relying on single-modality data such as MRI or PET images. These methods often overlook the potential interactions and complementary information between different modalities, thus limiting a comprehensive understanding of the disease. While a few studies have begun to explore the interactions of multimodal data, the methods for feature fusion are relatively simplistic and fail to delve into the complex relationships between modalities, impacting the accuracy and depth of diagnosis. Furthermore, most methods are diagnostically focused and lack the ability to provide comprehensive and effective prognostic information in real-world applications.

[0006] Through in-depth research, the inventors discovered three main problems with current artificial intelligence methods used for the diagnosis and prognosis of Alzheimer's disease:

[0007] First, most systems rely on single-modality patient data and fail to fully explore and utilize the complementary information between different modalities, resulting in insufficient reliability of the system output and thus limiting their effective application in clinical practice.

[0008] Secondly, existing methods are limited in their extraction of visual features and lack effective integration of global and local features, making it impossible for the model to fully and deeply understand the patient's condition.

[0009] Finally, the application of existing systems in prognostic analysis remains insufficient, yet prognostic analysis is crucial for clinical decision-making. These limiting factors collectively contribute to the low accuracy of Alzheimer's disease diagnosis, failing to meet clinical needs. Summary of the Invention

[0010] To address the shortcomings of existing technologies, this invention provides a method, system, electronic device, computer-readable storage medium, and computer program product for predicting Alzheimer's disease diagnosis based on multi-source knowledge. By integrating multimodal data (MRI images, brain statistics, and clinical data), it provides a comprehensive assessment of Alzheimer's disease progression.

[0011] In a first aspect, the present invention provides a method for predicting the diagnosis of Alzheimer's disease based on multi-source knowledge;

[0012] A multi-source knowledge-guided method for predicting Alzheimer's disease diagnosis includes:

[0013] Obtain MRI images, brain statistics, and clinical data from participants;

[0014] The MRI images are processed using a dual-path feature extraction strategy to obtain global brain features and local brain features, and then collaboratively learn to generate fused visual features.

[0015] The brain statistics data are fused with the fused visual features and then input into the trained diagnostic module for processing to obtain the diagnostic result at the current moment.

[0016] Guided by the clinical data and combined with the current diagnostic results, the trained prediction module is used to obtain the disease progression at future moments.

[0017] In some implementations, processing the MRI image using a dual-path feature extraction strategy to obtain global and local brain features includes:

[0018] The MRI images were processed using an improved ShuffleNetv2 model to obtain global brain features;

[0019] The MRI images were processed using a 3D CNN network to obtain local brain features;

[0020] The improved ShuffleNetv2 model and the 3D CNN network are configured for parallel operation.

[0021] In some implementations, the collaborative learning of global and local brain features to generate fused visual features specifically involves fusing global and local brain features through an interactive attention mechanism to obtain fused visual features.

[0022] In some implementations, fusing the brain statistics data with the fused visual features specifically involves concatenating the encoded brain statistics data with the fused visual features, adding tokens and location codes, and constructing a feature vector guided by brain statistical knowledge.

[0023] In some implementations, using the clinical data as a guide, combined with the current diagnostic results, and employing a trained prediction module to obtain the disease progression at future moments, specifically involves fusing the current diagnostic results with the encoded clinical data, processing the fusion result through a multi-head attention mechanism, and predicting the disease progression at future moments.

[0024] In some implementations, a comprehensive loss function is constructed by combining the diagnostic phase loss, the prediction phase loss, and the consistency loss to optimize the diagnostic module and the prediction module.

[0025] Secondly, the present invention provides an Alzheimer's disease diagnosis and prediction system based on multi-source knowledge guidance;

[0026] An Alzheimer's disease diagnostic prediction system based on multi-source knowledge includes:

[0027] The acquisition module is configured to acquire participants' MRI images, brain statistics, and clinical data.

[0028] The diagnostic module is configured to: process the MRI image using a dual-path feature extraction strategy to obtain global brain features and local brain features and perform collaborative learning to generate fused visual features; and input the brain statistical data and the fused visual features into the trained diagnostic module for processing to obtain the diagnostic result at the current time.

[0029] The prediction module is configured to: use the clinical data as a guide, combine the current diagnostic results, and utilize the trained prediction module to obtain the disease progression at future moments.

[0030] Thirdly, the present invention provides an electronic device;

[0031] An electronic device includes a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the above-described Alzheimer's disease diagnosis and prediction method based on multi-source knowledge.

[0032] Fourthly, the present invention provides a computer-readable storage medium;

[0033] A computer-readable storage medium having a computer program / instructions stored thereon, which, when executed by a processor, implements the steps of the above-described Alzheimer's disease diagnosis and prediction method guided by multi-source knowledge.

[0034] Fifthly, the present invention provides a computer program product;

[0035] A computer program product includes a computer program / instructions that, when executed by a processor, implement the steps of the above-described Alzheimer's disease diagnosis and prediction method guided by multi-source knowledge.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. The technical solution provided by this invention follows the operational steps of clinical practice, comprehensively utilizes multimodal data, including sMRI, clinical data statistics, and brain statistics knowledge. Through dual-pathway visual feature extraction, it combines global brain region and key local brain region features, and uses an interactive attention mechanism to achieve deep learning and fusion between features. Brain statistics knowledge guides image feature extraction, while clinical knowledge guides prognostic prediction, greatly improving the accuracy of diagnosis and prediction and significantly reducing the misdiagnosis rate.

[0038] 2. The technical solution provided by this invention designs a multi-period comprehensive loss function to balance the relationship between diagnostic and prediction tasks, ensuring that the model can maintain consistency in predictions at multiple time points and effectively find the optimal convergence solution of the model.

[0039] 3. The technical solution provided by this invention has strong interpretability. By quantitatively analyzing the attention mechanism of the model, it can accurately locate the patient's key brain regions and clinical indicators, making the diagnostic and prediction results closer to clinical reality and providing new ideas for disease prognosis and prevention. Attached Figure Description

[0040] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0041] Figure 1A flowchart illustrating the Alzheimer's disease diagnosis and prediction method based on multi-source knowledge provided in this embodiment of the invention;

[0042] Figure 2 This is a structural block diagram of the Alzheimer's disease diagnosis and prediction system based on multi-source knowledge provided in an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of the feature matrix fusion module based on clinical knowledge prompting mechanism provided in an embodiment of the present invention. Detailed Implementation

[0044] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0045] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. Furthermore, it should be understood that the terms “comprising” and “having”, and any variations thereof, are intended to cover non-exclusive inclusion, for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0046] Where there is no conflict, the embodiments and features in the embodiments of the present invention can be combined with each other.

[0047] Example 1

[0048] Existing AI-based diagnostic and prediction methods for Alzheimer's disease fail to fully exploit the complementarity and correlation between different modalities of data, and feature extraction has limitations, restricting the accuracy and comprehensiveness of diagnostic predictions. Therefore, this invention provides an Alzheimer's disease diagnostic and prediction method based on multi-source knowledge guidance, which comprehensively utilizes multi-modal data.

[0049] Next, combined Figure 1 , Figure 3 This embodiment discloses a multi-source knowledge-guided method for predicting Alzheimer's disease diagnosis, which includes the following steps:

[0050] S1. Obtain the participants' MRI images, brain statistics, and clinical data, and map them into visual feature matrices, clinical data feature matrices, and brain statistical feature matrices, respectively.

[0051] In this embodiment, brain statistics include gray matter density statistics for each brain region of the participant, and clinical data include the participant's comprehensive scale, education level, and basic information.

[0052] Here, the visual feature matrix is ​​a structured feature matrix, while the clinical data feature matrix and the brain statistical feature matrix are unstructured feature matrices.

[0053] S2. A dual-path feature extraction strategy is used to process MRI images, acquiring global and local brain features and performing collaborative learning to generate fused visual features. Specifically, this includes:

[0054] S201. The visual feature matrix is ​​processed using the improved ShuffleNetv2 model and a 3D CNN network respectively to obtain global brain features. and local brain features C1 represents the dimension of global brain features, and C2 represents the dimension of local brain features.

[0055] In this embodiment, the improved ShuffleNetv2 model includes a channel splitting layer, a 1×1 3D convolutional layer, a 3×3 depth-separable 3D convolutional layer, a 1×1 3D convolutional layer, a stitching layer, and a channel shuffle layer connected in sequence. Meanwhile, the output of the channel splitting layer is directly input to the stitching layer.

[0056] To adapt to 3D MRI images, the network structure was adjusted to incorporate a channel configuration suitable for 3D data. The number of channels in the channel shuffle layer was increased from 1024 to 2048 to accommodate the higher capacity of 3D data. Both depthwise separable convolutions and convolutional layers adopted 3D convolutional layers, with 3D implementations of grouped convolutions and pointwise convolutions, improving the model's flexibility and generalization ability. This made the improved ShuffleNetv2 model architecture suitable for more complex 3D data scenarios.

[0057] Before application, a training set was constructed using collected MRI images, and the improved ShuffleNetv2 model and 3D CNN network were trained respectively.

[0058] Based on this, changes in brain structure can be captured more comprehensively, improving the accuracy of feature expression and the sensitivity of diagnosis; global feature extraction identifies the overall pattern of disease through the macroscopic image structure of the brain, while local feature extraction focuses on fine-grained brain regions to capture local abnormalities.

[0059] S202. Global brain features and local brain features are fused through an interactive attention mechanism to obtain fused visual features.

[0060] Interactive attention mechanisms allow features from different sources to learn from and calibrate with each other, ultimately resulting in richer and more relevant integrated features. In this process, the interactive attention mechanism weighs the importance of features across different feature dimensions to ensure that the final fused features better represent functional abnormalities and structural changes in the brain, thereby enhancing the relevance of each feature.

[0061] For example, the computational flow of the interactive attention mechanism is as follows:

[0062] Add(cross-modal attention(G→L), cross-modal attention(L→G));

[0063]

[0064] G = A G V G L = A L V L ;

[0065] F = G + L;

[0066] Where G represents the weighted global brain features, L represents the weighted local brain features, d represents the scaling factor, and Q represents the local brain features. G The query matrix Q represents global brain features. L The query matrix K represents local brain features. G The key matrix, K, represents global brain features. L The key matrix represents global brain features, F represents fused visual features, and V represents the key matrix. G V represents global brain characteristics. L Indicates local brain features, A G This refers to the calculated global attention, i.e., Attention. G→L A L This represents the calculated local attention, i.e., Attention. L→G .

[0067] In this way, global and local features are weighted and learned through an interactive attention mechanism, and finally feature fusion is achieved by adding them together, thereby better capturing global and local information.

[0068] S3. After fusing brain statistical data with fused visual features, input the data into the trained diagnostic module for processing to obtain the diagnostic result at the current moment.

[0069] Specifically, firstly, the brain statistical feature matrix and fused visual features are concatenated and tokens and location codes are added to construct a feature vector guided by brain statistical knowledge; then, the feature vector guided by brain statistical knowledge is input into the trained diagnostic module to obtain the participant's current Alzheimer's disease diagnosis result.

[0070] Here, the brain statistical feature matrix and fused visual features are concatenated to obtain a fused composite vector. A predefined classification token (cls_token) is expanded to the input batch size and concatenated with the composite vector along the channel dimension. Simultaneously, to add positional encoding, the dimension of the positional encoding matches the length of the fused composite vector sequence, and the positional encoding and the composite vector are directly added. By adding the classification token and positional encoding, the global representation capability of the sequence features is enhanced, and positional and category information is provided for subsequent processing.

[0071] In this embodiment, the diagnostic module is a Transformer encoder.

[0072] Transformer models, leveraging their ability to capture long-range dependencies, can effectively identify complex patterns in brain functional networks, thereby generating diagnostic results at the current moment. Through multi-head self-attention mechanisms, Transformers can analyze the relationships between different brain regions, helping to identify abnormal connections associated with neurodegenerative diseases.

[0073] S4. Guided by clinical data and combined with the current diagnostic results, use a trained prediction module to obtain the disease progression at future moments.

[0074] Specifically, the current diagnostic result E is fused with the encoded clinical data feature matrix G, and the fusion result is processed through a multi-head attention mechanism to predict the disease progression at future moments.

[0075] The clinical data feature matrix is ​​represented as follows:

[0076] G = f clinical (H);

[0077] The fusion result is represented as follows:

[0078] F = concat(E, G);

[0079] The disease progression at a future time, i.e., the prediction result at time t+k, is expressed as follows:

[0080] prediction t+k =f pred (F);

[0081] In the formula, H represents clinical data, and k represents the prediction time step.

[0082] Based on this, it not only helps predict the possible course of a disease, but also assists doctors in developing long-term treatment plans, and in timely intervention and tracking of changes in the condition.

[0083] S5. Conduct interpretive analysis of the diagnostic and predictive results.

[0084] Furthermore, the effectiveness of the brain statistics used was verified using the Lasso regression method. Subsequently, the importance of clinical features was ranked using the self-attention mechanism in Transformer, and their importance in different domains was ranked and visualized. The results were verified in a real dataset, as shown in Table 1.

[0085] Table 1. Validation table of the knowledge-guided multimodal auxiliary diagnostic prediction system for Alzheimer's disease on real datasets.

[0086]

[0087] In this embodiment, a deep learning model integrating clinical knowledge and brain statistical knowledge was designed based on multi-source information for the diagnosis and prediction of Alzheimer's disease. By fusing multiple types of information, brain statistical knowledge is used to guide the extraction of dual-path visual features during the diagnosis stage, and clinical knowledge is applied to further guide the diagnostic output features during the prediction stage, thereby improving the accuracy of predictions for future time points.

[0088] To balance the relationship between diagnostic and predictive tasks, ensure the model maintains consistency across multiple time points, and effectively find the model's optimal convergent solution, as an implementation method, the following is included before the predictive and diagnostic modules:

[0089] A comprehensive loss function is constructed by combining the loss from the diagnosis phase, the loss from the prediction phase, and the consistency loss, and the diagnosis module and prediction module are optimized accordingly.

[0090] During the diagnosis phase, the model generates the current diagnostic result D through dual-path feature extraction, interactive attention mechanism, and Transformer processing. To measure the difference between this diagnostic result and the true label, a classification loss (typically cross-entropy loss) is used to optimize the model's classification ability. This loss can be expressed as:

[0091]

[0092] Where N represents the number of samples, y i This represents the true label of sample i. This represents the output diagnostic probability distribution.

[0093] In the prediction phase, the model fuses diagnostic features E with coded clinical features G and inputs these into the prediction module to generate predicted outputs for multiple future time points. Prediction loss measures the difference between the model's predicted future time points and the actual disease progression. Prediction loss is typically measured using mean squared error (MSE) or time-series correlation regression loss. This part can be defined as:

[0094]

[0095] Where T represents the number of time points to be predicted, y t+k and and represent the true label and model prediction value at the (t+k)th time node in the future, respectively.

[0096] By minimizing the prediction loss, the model can improve the accuracy of predicting the future development of a patient's condition.

[0097] After diagnosis and prediction, the consistency loss can be obtained based on the diagnosis and prediction results, as shown in the formula:

[0098]

[0099] Among them, f i,j This indicates the output of the diagnostic module. This represents the output of the prediction module at the current time step. Mean squared error (MSE) is used as the loss function to evaluate the consistency between the two modules. This choice allows for a higher penalty for significant errors when the model's predictions deviate.

[0100] Based on this, a comprehensive loss function was constructed:

[0101]

[0102] Where λ is the regularization coefficient, a non-negative real number (λ∈R+), used to adjust the model's focus on consistency loss during training and optimization.

[0103] To verify the optimal value of the hyperparameter λ, this embodiment was tested on a real dataset. The experiment shows that the result is best when λ = 0.4. The verification results are shown in Table 2.

[0104] Table 2. Verification table of the optimal value of α in the objective function. All indicators in the table are average values.

[0105] Accuracy MAUC SEN SPE λ = 0.2 0.785 0.911 0.820 0.901 λ = 0.4 0.801 0.920 0.819 0.911 λ = 0.6 0.790 0.904 0.806 0.897 λ = 0.8 0.779 0.910 0.798 0.889 λ=1 0.770 0.904 0.786 0.870

[0106] Example 2

[0107] Combination Figure 2Based on the Alzheimer's disease diagnosis and prediction method guided by multi-source knowledge described in Embodiment 1, this embodiment discloses an Alzheimer's disease diagnosis and prediction system guided by multi-source knowledge, comprising:

[0108] The acquisition module is configured to acquire participants' MRI images, brain statistics, and clinical data.

[0109] The diagnostic module is configured to: process the MRI image using a dual-path feature extraction strategy to obtain global brain features and local brain features and perform collaborative learning to generate fused visual features; and input the brain statistical data and the fused visual features into the trained diagnostic module for processing to obtain the diagnostic result at the current time.

[0110] The prediction module is configured to: use the clinical data as a guide, combine the current diagnostic results, and utilize the trained prediction module to obtain the disease progression at future moments.

[0111] It should be noted that the acquisition module, diagnosis module, and prediction module described above correspond to the steps in Embodiment 1. The examples and application scenarios implemented by these modules and their corresponding steps are the same, but they are not limited to the content disclosed in Embodiment 1. It should also be noted that these modules, as part of the system, can be executed in a computer system, such as a set of computer-executable instructions.

[0112] Example 3

[0113] Embodiment 3 of the present invention provides an electronic device, including a memory and a processor, as well as computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, they complete the steps of the above-mentioned Alzheimer's disease diagnosis and prediction method guided by multi-source knowledge.

[0114] Example 4

[0115] Embodiment 4 of the present invention provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the above-described Alzheimer's disease diagnosis and prediction method guided by multi-source knowledge.

[0116] Example 5

[0117] Embodiment 5 of the present invention provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-described Alzheimer's disease diagnosis and prediction method based on multi-source knowledge.

[0118] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment, whereby a series of operational steps are performed to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] The descriptions of each embodiment in the above embodiments have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0122] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-source knowledge-guided method for predicting and diagnosing Alzheimer's disease, characterized in that, include: Obtain MRI images, brain statistics, and clinical data from participants; Brain statistics include gray matter density statistics for each brain region of the participants, and clinical data include the participants' comprehensive scale, education level and basic information; The MRI image is processed using a dual-path feature extraction strategy to obtain global and local brain features. This includes: processing the MRI image using an improved ShuffleNetv2 model to obtain global brain features; and processing the MRI image using a 3D CNN network to obtain local brain features. The improved ShuffleNetv2 model and the 3D CNN network are configured in parallel and perform collaborative learning to generate fused visual features. Specifically, global and local brain features are fused through an interactive attention mechanism to obtain fused visual features. The improved ShuffleNetv2 model consists of a channel splitting layer, a 1×1 3D convolutional layer, a 3×3 depthwise separable 3D convolutional layer, a 1×1 3D convolutional layer, a stitching layer, and a channel mixing layer connected in sequence. The output of the channel splitting layer is directly input to the stitching layer. A channel configuration adapted to 3D data is introduced, increasing the number of channels in the channel mixing layer from 1024 to 2048. Both the depthwise separable convolutional layer and the convolutional layer use 3D convolutional layers. The brain statistics data are fused with the fused visual features and then input into the trained diagnostic module for processing to obtain the diagnostic result at the current moment. Guided by the clinical data and combined with the current diagnostic results, the disease progression at future moments is obtained using a trained prediction module. Specifically, the current diagnostic results are fused with the encoded clinical data, and the fusion result is processed through a multi-head attention mechanism to predict the disease progression at future moments. A comprehensive loss function is constructed by combining the diagnostic stage loss, the prediction stage loss, and the consistency loss to optimize the diagnostic module and the prediction module.

2. The Alzheimer's disease diagnosis and prediction method based on multi-source knowledge as described in claim 1, characterized in that, The fusion of the brain statistics data with the fused visual features specifically involves: concatenating the encoded brain statistics data with the fused visual features, adding tokens and location codes, and constructing a feature vector guided by brain statistical knowledge.

3. An Alzheimer's disease diagnosis and prediction system guided by multi-source knowledge, characterized in that, The method for predicting Alzheimer's disease diagnosis based on multi-source knowledge as described in any one of claims 1-2 includes: The acquisition module is configured to acquire participants' MRI images, brain statistics, and clinical data. The diagnostic module is configured to: process the MRI image using a dual-path feature extraction strategy to obtain global brain features and local brain features and perform collaborative learning to generate fused visual features; and input the brain statistical data and the fused visual features into the trained diagnostic module for processing to obtain the diagnostic result at the current time. The prediction module is configured to: use the clinical data as a guide, combine the current diagnostic results, and utilize the trained prediction module to obtain the disease progression at future moments.

4. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the Alzheimer's disease diagnosis and prediction method based on multi-source knowledge as described in any one of claims 1-2.

5. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the Alzheimer's disease diagnosis and prediction method based on multi-source knowledge as described in any one of claims 1-2.

6. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they implement the steps of the Alzheimer's disease diagnosis and prediction method based on multi-source knowledge as described in any one of claims 1-2.

Images