V3I3P60

Deep Learning-Based Early Prediction of Alzheimer’s Disease

Mr. Nitin Barsagde¹, Kunali G. Atkare², Pranjali S. Sarode³, Siddhi R. Malewar⁴, Sonali P. Pradhan⁵, Vaibhav N.  Puri^6*

Abstract

Alzheimer’s disease (AD) is a progressively advancing neurological disorder that leads to irreversible degeneration of brain cells, affecting memory, cognition, and motor abilities. Early detection of AD plays a vital role in enabling timely medical intervention and slowing disease progression. However, limited public awareness and the complexity of neurological changes make early diagnosis challenging, particularly as symptoms worsen with age. For this study, the OASIS dataset obtained from Kaggle was employed, with 80% of the data allocated for model training and the remaining 20% reserved for testing to ensure a reliable evaluation process. To accomplish this objective, an efficient automated diagnostic framework was developed using transfer learning with the EfficientNet-B0 Convolutional Neural Network (CNN).

Furthermore, class-wise sensitivity analysis reveals that the proposed EfficientNet-B0 model performs consistently well across all three classes, including Alzheimer’s Disease (AD), Mild Cognitive Impairment (MCI), and Normal Control (NC), confirming its robustness in stage-wise classification.

Keywords:

Alzheimer’s Disease Detection, EfficientNet-B0 Models, Brain MRI Analysis, CT Imaging