Paper Title
PREDICTING MENTAL HEALTH SCORES FROM DIGITAL BEHAVIOR USING A MACHINE LEARNING APPROACH

Abstract
The growing integration of digital technologies into daily life has opened new avenues for understanding human behavior and mental health. This study explores the potential of using machine learning techniques to predict individual mental health scores based on digital behavioral data. By analyzing patterns such as social media usage, screen time, activity logs, and communication frequency, the research aims to identify correlations between digital footprints and psychological well-being. A diverse dataset was gathered and pre-processed to extract significant characteristics, which were then used to train and assess several machine learning models, such as XGBoost, AdaBoost, LightGBM, CatBoost, Support Vector Machine, and Neural Networks. All these algorithms showed promising accuracy in assessing mental health scores using standardized psychological evaluation methods, but CatBoost outperformed all with a 97.4% F1-score. This technique demonstrates the viability of using non-invasive, passively gathered digital data for continuous mental health monitoring, which can provide significant insights for early intervention and personalized mental healthcare. Keyword - Artificial Intelligence, Healthcare, Mental Health, Digital Behavior, Prediction, Machine Learning, Non-invasive Monitoring.