Mental disorders, including depression, pose a significant global health challenge. The rise in the prevalence of mental health issues demands innovative diagnostic tools and early intervention. This article presents a study that harnesses the power of machine learning and multi-modal data analysis to develop a robust classifier for distinguishing between healthy individuals and those with depression. The study utilizes graphological signals, such as handwriting and drawing as potential markers for depression. In this study we conducted an analysis on an existing database, upon which we developed machine learning models that outperformed existing literature. The results demonstrate the potential of these signals accurately classify individuals, with implications for early detection and telemedicine applications. Additionally, we collected new data, including handwriting, drawing, and laughter recordings, which will be utilized to create new models with the aim of achieving more effective performance. Our study also includes the integration of an inertial motion sensor into a mobile app, offering prospects for wearable technology and expanded diagnostic capabilities.