랜덤포레스트 기반 코로나19 전후 청소년의 우울, 자살 생각 예측 요인 비교 분석

초록

This study systematically compares and analyzes predictors of depression and suicidal thoughts among adolescents before and after the COVID-19 pandemic. To this end, we utilized data from the Korean Child and Adolescent Panel Survey 2018 for 2,019 and 2,022 middle school second-year students, and applied the random forest technique from among machine learning algorithms. From the results of the study, somatic symptoms were the most important predictors of depression among adolescents both before and after the pandemic, and depression was the most important predictor of suicidal ideation. After COVID-19, social withdrawal became a more important predictor of adolescent depression than suicidal ideation, and parenting attitudes were slightly more important as a predictor of depression and suicidal ideation among adolescents after the pandemic. Aggression ranked higher in importance among predictors of adolescent depression and suicidal ideation both before and after COVID-19. Finally, friendships ranked higher in importance among predictors of depression and suicidal ideation both before and after the pandemic. The implications of these findings for counseling and education are also discussed.

키워드