Classification and Analysis of Dementia using Machine Learning Algorithms

Abstract

Dementia is associated to one of the early phases of fatal diseases such as Huntington's disease or, in more severe situations, death. It is a chronic and degenerative disease that affects millions of people throughout the world. Memory loss, difficulty in concentration, mood changes, and being confused are some of the symptoms of dementia. Early detection can prevent or postpone the course of dementia, which is the most common degenerative illness among the elderly. This paper's principal objective is to deploy a variety of machine learning algorithms like Logistic Regression, ExtraTreesClassifier, Random Forest, ExtremeBoost Classifier (XGBoost), Light Gradient Boost (LGBM), Decision Tree, Gradient Boosting, Gaussian Nave Bayes, and Classifier Support Vector Machine (SVM) on selected features from the dataset for multi-class classification of dementia patients as 'demented,' 'non-demented,' and 'converted'. The model evaluation was based on the F1-score, precision and accuracy and it was found that Gradient Boost performed well with an accuracy of 0.96.