Background: The COVID-19 pandemic has posed risks to public mental health worldwide. University students, who are already recognised as a vulnerable population, are at elevated risk of mental health issues given COVID-19-related disruptions to higher education. To assist universities in effectively allocating resources to the launch of targeted, population-level interventions, the current study aimed to uncover predictors of university students' psychological wellbeing during the pandemic via a data-driven approach.
Methods: Data were collected from 3973 Australian university students ((median age = 22, aged from 18 to 79); 70.6% female)) at five time points during 2020. Feature selection was conducted via least absolute shrinkage and selection operator (LASSO) to identify predictors from a comprehensive set of variables. Selected variables were then entered into an ordinary least squares (OLS) model to compare coefficients and assess statistical significance.
Results: Six negative predictors of university students' psychological wellbeing emerged: White/European ethnicity, restriction stress, perceived worry on mental health, dietary changes, perceived sufficiency of distancing communication, and social isolation. Physical health status, emotional support, and resilience were positively associated with students' psychological wellbeing. Social isolation has the largest effect on students' psychological wellbeing. Notably, age, gender, international status, and educational level did not emerge as predictors of wellbeing.
Conclusion: To cost-effectively support student wellbeing through 2021 and beyond, universities should consider investing in internet- and tele- based interventions explicitly targeting perceived social isolation among students. Course-based online forums as well as internet- and tele-based logotherapy may be promising candidates for improving students' psychological wellbeing.
Keywords: COVID-19; intervention; machine learning; psychological wellbeing; university students.