Early Prediction of Dementia Using Machine Learning

Abstract

This study explores the application of machine learning algorithms for the early prediction of dementia, aiming to improve diagnostic accuracy and reliability. Utilizing a comprehensive dataset from Kaggle, which includes both continuous and categorical variables, four machine learning models—Random Forest, Decision Tree, Logistic Regression, and Support Vector Machine (SVM)—were implemented and evaluated. The study identifies cognitive test scores, the APOE ε4 allele, and depression status as key predictors of dementia. Tree-based models demonstrated superior performance, achieving perfect scores across metrics such as accuracy, recall, precision, and F1. Despite these promising results, the study acknowledges limitations such as the reliance on a single dataset, limited predictors, and challenges in real-world validation. Future research should incorporate larger, more diverse datasets, longitudinal data, and additional predictors to improve model robustness and applicability. These findings highlight the potential of machine learning as a transformative tool in clinical settings for timely dementia diagnosis and intervention.