Early Detection of the Alzheimer’s Disease: A Novel Cognitive Feature Selection Approach Using Machine Learning

Alzheimer’s Disease (AD) is a dynamic condition that affects cognitive capabilities and functioning. It is a challenging disease to detect, particularly in its early stages. Early diagnosis of AD is the key for its treatment and slowing of its progress. This paper argues and clearly shows the benefits of using cognitive tests for efficient and early AD detection. In this study, a novel approach for the early detection of AD is proposed. We refer to it as Neighborhood Component Analysis and Correlation-based Filtration (NCA-F) and is based on selecting and identifying significant cognitive features. Cognitive features are used to train four Machine Learning (ML) classifiers, including Support Vector Machine (SVM), Naïve Bayes (NB), ANN, and AdaBoost Ensemble (AdB). Our analysis shows that the proposed approach can achieve an 88% classification accuracy. In addition, the performance of various ML classifiers with varying combinations of features has been studied. The proposed feature selection approach that implements AdB is seen to provide the best performance by various metrics.