What explains adolescents' physical activity and sports participation during the COVID-19 pandemic? – an interpretable machine learning approach
Lingyi Fu Ryan D. Burns Shandian Zhe Yang Bai a Department of Health & Kinesiology,University of Utah,Salt Lake City,USAb Kahlert School of Computing,University of Utah,Salt Lake City,USA
DOI: https://doi.org/10.1080/02640414.2024.2404783
IF: 3.9428
2024-09-21
Journal of Sports Sciences
Abstract:Adolescents' physical activity (PA) and sports participation declined due to the COVID-19 pandemic. This study aimed to determine the critical socio-ecological factors for PA and sports participation using a machine learning approach. We did a cross-sectional secondary data analysis utilising the 2021 National Survey of Children's Health (NSCH) dataset ( N =16,166; 49.0% female). We applied an interpretable machine learning approach (e.g. decision tree-based models) that examined the critical factors associated with PA and sports participation. The factors related to the intrapersonal, interpersonal, organisational, and community levels of the socio-ecological model. Out of the 25 factors examined, our findings unveiled the 11 critical factors associated with PA and the 10 critical factors associated with sports participation. Factors at the intrapersonal levels (e.g. age, screen time, and race) held greater importance to PA than those at the other three levels. While interpersonal factors (e.g. parent participation in children's events/activities, family's highest educational level, and family income level) were most important for sports participation. This study identified that the common critical factors of physical activity and sports participation during the COVID-19 pandemic mainly relied on intrapersonal and interpersonal levels. Unique factors were discussed. In this study, we identified 11 critical factors for PA, with the top five being age, neighbourhood amenities, screen time, missed school days, and family income level . Additionally, we identified 10 critical factors for sports participation, with the top five factors being parent participation in a child's events/activities, family's highest educational level, family income level, screen time, and school engagement . These findings emphasise the shared significance of intrapersonal and interpersonal factors as common determinants of both PA and sports participation. Notably, PA appears to be primarily influenced by intrapersonal factors (e.g. age, screen time, and race), reflecting its more internally driven nature. In contrast, sports participation appears to be more externally driven, primarily shaped by interpersonal factors (e.g. parent participation in the child's events/activities, family's highest educational level, and family income level). This distinction underscores the need for educators and policymakers to carefully consider these common and unique factors when devising promotion strategies during the COVID-19 pandemic. By recognising these distinctions, interventions can be better tailored to encourage both PA and sports participation among adolescents.
sport sciences