Predicting Alzheimer’s Disease Using Lifestyle and Cognitive Data

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder with great personal and public health implications. In this study, we employ a publicly available clinical dataset of 2,149 patients to predict AD diagnosis using machine learning. Logistic regression and random forest models were used to a set of demographic, cognitive, and lifestyle features, with the best accuracy attained being 90.5% and a ROC-AUC of 0.942. Important predictive variables were MMSE score, memory complaints, ADL score, and behavioral symptoms. Our findings demonstrated the viability of simple, interpretable models for early Alzheimer’s detection and support future development of risk-scoring tools for preventive care and early intervention.