Analysis of machine learning and deep learning models for early Alzheimer's disease diagnosis

Abstract

Machine learning (ML) models and deep learning (DL) algorithms have become widely utilized tools in the healthcare sector. They are helpful in identifying patterns, analyzing imaging data, and utilizing various biomarkers to identify between a healthy and unhealthy individual. This technology contributes to confirming early diagnoses, particularly for Alzheimer’s disease, and mild cognitive impairments (MCIs). Alzheimer’s disease is a neurological disorder that causes memory loss and inhibits the individual’s ability to function independently. This paper compares the ML models used to classify the disease, differentiating between how often they are used and their accuracy. The ML models include support vector machines (SVMs), random forest (RF), and logistic regression, while the DL algorithms utilized include convolutional neural networks (CNNs) and recurrent neural networks (RNNs). Through the comparison, we find that the most effective ML models are SVMs due to their ability to handle large datasets and provide precise image classification, whereas the most effective DL models are CNNs, because of their multiple variations in structures and highest accuracy rates in analyzing patterns in medical images through in-depth analysis for early AD diagnosis.