Multi-Modal Data Exploration and Analysis for Predicting Severity Impairment Index Using Actigraphy and Behavioral Metrics

Understanding and predicting internet use behaviors and their impact on mental health is pivotal in addressing challenges like problematic internet usage. This project focuses on exploratory data analysis (EDA) of the Healthy Brain Network (HBN) dataset, comprising physical activity data, internet usage behavior, and clinical measures. The aim is to uncover patterns and insights that contribute to predicting the Severity Impairment Index (SII), a standard measure of problematic internet use. The study emphasizes the integration of diverse data sources, including tabular demographic and health data and time-series actigraphy data, for a comprehensive exploration. Through feature engineering, advanced correlation analyses, and data visualization, the project delves into critical aspects such as physical activity patterns, internet addiction tendencies, and their interplay with mental health metrics. Novel techniques like dimensionality reduction, clustering, and time-lagged analysis are utilized to highlight key predictors of SII. This project also explores the use of machine learning-ready features, emphasizing data-driven decision-making to identify risk factors and trends. By leveraging advanced EDA techniques, the study underscores the importance of structured data exploration in improving mental health diagnostics and intervention strategies. This approach empowers researchers to address problematic internet use effectively and contributes to a better understanding of behavioral patterns across diverse demographics.