Comparative Analysis of Machine Learning Algorithms for classification of Alzheimer’s disease

Abstract

With significant increase in interest in dementia and Alzheimer’s disease, a form of dementia that deteriorates the brain as well as the memory along with other important mental functions, in which brain neuron connections begin to disintegrate and die. Memory loss and confusion being the main symptoms, physicians and scientists are yet to find a concrete cure for the same. However, different strategies and medications have been found to be helpful, especially if early detection is possible. The report shows the results and analysis of detecting Alzheimer’s using machine learning models and compared them. The OASIS dataset has been applied to different machine learning models like SVM, Logistic Regression, K-Nearest Neighbors, Naive Bayes, Decision Tree, Random Forest, and Ensemble Learning. It has been run through these models before and after fine tuning. The best accuracy was found with Support Vector Machines after fine tuning. Another observation is that the inclusion of feature of CDR’ i.e., Clinical Dementia Rating showed a spike in the evaluation metrics with the best accuracy being 97% achieved by Support Vector Model.