Alzheimer’s disease is a global health concern, underscoring the need for early detection to enable timely intervention. By exploring handwriting data from individuals with and without AD across various tasks, we aim to identify predictive features that influence the models. The analysis encompasses data from dictation, copy, graphic, and memory tasks, providing insights into the complexity of handwriting behavior in AD. Notably, copying tasks exhibit higher predictive accuracy. Key features like mean jerk and pressure variance emerge as significant predictors. These findings underscore handwriting as a promising marker for early AD detection, informing personalized screening and intervention strategies. Future research avenues include refining machine learning models, expanding datasets, and exploring additional handwriting features to enhance predictive precision.