Identifying the optimal predictors for adolescent mental and physical health using machine learning methods.

Area of Science:

• Adolescent Health
• Mental Health
• Physical Health
• Cross-cultural Psychology
• Machine Learning Applications

Background:

• Adolescent health is a global concern, with prior research often focusing narrowly on limited predictor domains for mental and physical health issues.
• Existing studies frequently examine single or few predictor domains, potentially overlooking the complex interplay of factors influencing adolescent well-being.

Purpose of the Study:

• To identify optimal predictors for adolescent mental and physical health using advanced analytical techniques.
• To examine the influence of a comprehensive set of 46 predictors across nine domains on adolescent health outcomes.
• To compare predictor importance across different cultural contexts, specifically Hong Kong and the Netherlands.

Main Methods:

• Utilized multiple machine learning algorithms to analyze a large dataset of adolescent health indicators.
• Employed data from the Programme for International Student Assessment (PISA), 2022, encompassing adolescents from Hong Kong and the Netherlands.
• Evaluated the predictive power of 46 variables across nine domains for mental and physical health.

Main Results:

• Emotional and social capital domains emerged as the most significant predictors for adolescent mental and physical health in both studied societies.
• Identified culturally nuanced patterns: environmental domain critical in Hong Kong, physical domain critical in the Netherlands.
• Demonstrated that predictive models trained on data from one society performed less effectively on data from another, highlighting cultural specificity.

Conclusions:

• Interventions focusing on enhancing emotional and social capital show promise for improving adolescent mental and physical health.
• Emphasized the necessity of culturally tailored strategies for interventions, acknowledging both overlapping patterns and unique societal influences.
• Recommended a nuanced approach to adolescent health interventions that respects cultural diversity while leveraging common predictive factors.