A Novel Light-Weight Convolutional Neural Network Model to Predict Alzheimer’s Disease Applying Weighted Loss Function

Abstract

Alzheimer’s disease (AD) is a progressive neurological disorder that presents a significant public health concern. Early detection of Alzheimer’s has the potential to greatly improve patient care and treatment. Artificial intelligence (AI) has the potential to revolutionize healthcare by improving patient outcomes and empowering healthcare providers. In recent years, significant breakthroughs in medical diagnosis have occurred, thanks to the use of AI, particularly through the application of deep learning (DL) techniques. These advancements have the potential to greatly improve patient care and outcomes. Several proposals have been developed utilizing DL techniques to identify AD. This study proposes a DL model to classify individuals with AD using magnetic resonance imaging images. The study aims to evaluate DL’s effectiveness in predicting AD. The proposed model used a custom-weighted loss function, resulting in a 99.24% training accuracy, 96.95% test accuracy, a Cohen’s kappa score of 0.931, and a weighted average precision of 97%. The model is evaluated against several pre-trained models. Regarding accuracy findings and Cohen’s kappa score, the suggested model performs better than the others.