Alzheimer’s Disease Prediction Using Optimized DNN and CNN with Mini-batch Gradient Descent

Abstract

Alzheimer’s disease (AD) is a progressive neurological disorder that leads to memory loss and cognitive decline, creating a major challenge for global healthcare systems. Early and reliable prediction of AD is essential to support timely clinical intervention and effective disease management. This study presents a comparative evaluation of optimized deep learning (DL) models and conventional machine learning (ML) techniques for Alzheimer’s disease prediction using the OASIS dataset. Several traditional ML algorithms, including Support Vector Machine, Random Forest, and Gradient Boosting, are implemented and improved through systematic hyperparameter tuning. In parallel, a deep neural network and Optimized convolutional neural networks is designed with optimized training strategies such as adaptive learning rates, dropout regularization, and early stopping to enhance model generalization and reduce overfitting. The performance of all models is assessed using standard evaluation metrics, including accuracy, precision, recall, F1-score, and AUC-ROC. Experimental results show that optimized convolutional neural networks achieve superior predictive performance compared to traditional ML methods, particularly in identifying positive AD cases. However, conventional ML models demonstrate competitive accuracy with lower computational cost, making them suitable for practical deployment in resource-limited environments. The findings highlight the strengths and limitations of both approaches and provide useful insights for selecting effective predictive models for Alzheimer’s disease diagnosis.