May - June 2024
Core skills: Data processing, feature extraction, time-series/panel data, Python
During this two-month project, I conducted a quantitative research using the K-EmoPhone dataset on screentime, physical activity and their effect on mental health markers like daily stress levels and depressive symptoms in young Korean adults. This largely included high-density trace data from smartphone and smartwatch. Additionally, it included self-reported data on several psychological aspects. The panel had data of 7 days for 77 subjects. Conducting the analysis, the dataset was first preprocessed, features were extracted and main analyses were conducted. A fixed effects model was ran to gain insight into the effects of the predictors on within-subject factors. Following, a multiple regression was ran to explore the effect of the predictors on the between-subjects factors. Lastly, a scientific article was written presenting the results.
In recent years, excessive smartphone use and physical inactivity among young adults have raised concerns about their impact on mental health. This study explores the relationship between screen time, physical activity, and their effects on stress and depression. Using objective measures and self-reported data from the K-emoPhone dataset, we employed fixed effects and multiple linear regression models to analyze these interactions. Contrasting to most research, findings indicate that higher screen time is significantly correlated with lower stress levels across participants. Though, it shows no within-person effects over time. Calorie expenditure consistently correlates with reduced stress, underscoring the importance of physical activity. However, the categorical physical activity feature lacked significance, suggesting the need for refined measures. The analysis of depression data revealed invalid model findings, leading to its removal from the process. The study highlights the complexities of these relationships and suggests future research to incorporate more objective mental health measures, enhanced data collection methods, and broader variable exploration. These insights are crucial for developing strategies to better understand the effects of digital and inactive lifestyles on mental health.