A Machine Learning Analysis of Socio-Ecological and Psychological Risk Factors for Suicide Among a Nationally Representative Sample of Ghanaian Junior High School Students

Background and Purpose: Suicide among adolescents is a critical global public health issue, particularly in low- and middle-income countries (LMICs) like Ghana, where mental health resources are scarce. Despite the growing recognition of adolescent suicide as a pressing concern, traditional methods for identifying suicide risk often fail to capture the complex interplay of socio-ecological and psychological factors. The advent of artificial intelligence (AI) and machine learning (ML) offers a transformative opportunity to improve suicide risk prediction and intervention strategies. This study aims to utilize AI/ML techniques to analyze socio-ecological and psychological risk factors for suicide among a nationally representative sample of Ghanaian junior high school students, offering culturally relevant, data-driven insights for targeted interventions.

Methods: A cross-sectional survey was conducted with 1,703 junior high school students aged 12–18 years from rural, peri-urban, and urban areas across Ghana. Data on psychological factors (e.g., depression, anxiety) and socio-ecological factors (e.g., bullying, parental support) were collected using validated measures. Descriptive statistics, chi-square tests, and t-tests were used for preliminary analyses, while Random Forest and Logistic Regression models were employed for suicide risk prediction. Model performance was evaluated using accuracy, sensitivity, specificity, and feature importance analysis.

Results: Psychological factors such as depression (r = .42, p < .01), anxiety (r = .38, p < .01), and perceived stress (r = .35, p < .01) were the strongest predictors of suicide risk, while parental support emerged as a significant protective factor (r = –.34, p < .01). The Random Forest model demonstrated good predictive performance (accuracy = 78.3%, AUC = 0.81). Gender differences were observed, with female students reporting higher levels of psychological distress than males.

Conclusions and Implications: This study is the first to apply AI/ML techniques to a nationally representative dataset of Ghanaian adolescents for suicide risk prediction. The findings highlight the critical role of psychological and socio-ecological factors in shaping suicide risk and underscore the potential of AI/ML to provide scalable, precise tools for early identification of at-risk individuals. These insights can inform culturally relevant, evidence-based interventions to reduce adolescent suicide in Ghana and other LMICs.