Abstract:Alzheimer’s Disease (AD) is a progressive neurodegenerative condition characterized by deterioration of cognitive functions. Although there does not exist a cure for AD, early diagnosis and intervention during stages of Mild Cognitive Impairment (MCI) can be incredibly beneficial in slowing its development. However, predicting the conversion from MCI to AD remains a challenging task. This paper aims to provide an overview of the current research on predicting MCI to AD conversion, with a focus on machine learning methodologies to aid feature extraction and classification. Through a literature review, we aim to offer insights into the latest state-of-the-art techniques to predict MCI to AD conversion as well as prevailing trends in this field including deep learning, transfer learning, contrastive learning, and graph neural networks. We conclude from our study that utilizing AD/HC classes for feature extraction is a promising approach for generating discriminative features for stable MCI and progressive MCI classification. In addition, employing multi-modality models is instrumental in attaining a robust validation framework for the early diagnosis of AD.

Prediction Of Mild Cognitive Impairment to Alzheimer’s Disease Conversion Via Machine Learning