Discovering the Impact on Mental Health for Prolonged Online Dependency in Post-COVID Era Using Deep Transfer Learning

Widespread reliance on digital devices has surged in the post-COVID era, raising critical concerns about their impact on mental health. In this paper, we investigate the correlation between excessive device usage and mental health outcomes using a cutting-edge technique—namely, transfer learning. Real-life quantitative data are collected to incorporate stress, anxiety, depression, and sleep quality in order to correlate factors associated with mental health issues. Complementary qualitative surveys provided insights into user behavior, coping strategies, and perceived psychological effects. Our methodology emphasizes robust data preprocessing and feature selection techniques, selecting an appropriate pretrained base model, building an optimized model, training and evaluating the model, and leading to optimized performance across key evaluation metrics (including accuracy, F1-score, sensitivity, and specificity). An extensive empirical analysis is conducted to highlight the effectiveness of the proposed model. Our findings contribute to the development of practical tools and future research pathways aimed at mitigating the mental health risks associated with prolonged digital device use.