Metaheuristic-driven dual-layer model for classifying Alzheimer's disease stages.

Area of Science:

• Neuroimaging and Artificial Intelligence
• Computational Neuroscience
• Medical Diagnostics

Background:

• Accurate Alzheimer's disease (AD) staging is vital for patient care and treatment.
• Distinguishing between AD progression stages using neuroimaging is challenging.

Purpose of the Study:

• To develop an advanced machine learning framework for multi-stage Alzheimer's disease classification.
• To enhance the accuracy and interpretability of AD staging using MRI data.

Main Methods:

• A two-tier machine learning architecture combining Convolutional Neural Networks (CNNs) for feature extraction and ensemble models (XGBoost, LightGBM) for classification.
• Application of metaheuristic optimization strategies to refine model performance.
• Evaluation on a public Alzheimer's disease dataset.

Main Results:

• The proposed framework achieved a maximum classification accuracy of 89.55% for multi-stage AD classification.
• Demonstrated robust predictive performance and strong generalization capabilities across different experimental configurations.
• Explainable AI (XAI) techniques provided insights into neuroimaging biomarkers related to AD progression.

Conclusions:

• The developed framework effectively classifies Alzheimer's disease progression stages using MRI data.
• Incorporation of XAI enhances model interpretability and clinical relevance.
• This approach offers a promising direction for data-driven Alzheimer's diagnosis and staging research.