An Explainable Deep Learning Model for Classification and Analysis of Alzheimer's Disease for Clinical Trust

Alzheimer’s Disease (AD), a progressive neurodegenerative disorder, presents significant challenges in early diagnosis and treatment. While deep learning models have demonstrated high accuracy in analyzing medical imaging data, their lack of interpretability limits clinical adoption. This study proposes an explainable deep learning framework that integrates a Convolutional Neural Network (CNN) with the Local Interpretable Model-agnostic Explanations (LIME) technique to enhance transparency and clinical trust in AD classification. The model highlights key biomarkers, such as hippocampal atrophy and amyloid plaque density, that contribute to its predictions. The approach addresses the need for consistent evaluation metrics and the integration of domain expertise in AI-driven diagnosis. Experiments conducted on the OASIS benchmark dataset achieved a classification accuracy of 97%, a precision of 97.2%, a recall of 96.8%, and an AUC of 0.97, with LIME providing localized, interpretable insights that align with clinical understanding. By combining predictive performance with explainability, the framework addresses ethical concerns by fostering transparency, accountability, trustworthiness, and practical AI-assisted diagnostic tools for precision healthcare.