O-47 A machine learning approach to modelling intimate partner violence exposure risk among university students in Barbados
Author(s):
MH Campbell, PS Chami, PS Gaski, T Whitby-Best , NS Greaves, MK Emmanuel, JA Ward, SG Anderson
Year of Presentation:
2025
Objective: To present the first comprehensive assessment of
intimate partner violence (IPV) exposure among university
students in Barbados and to elucidate the most important
predictors of IPV exposure through application of machine
learning (ML).
Methods: A cross-sectional survey of 649 students investigated the most reported forms of IPV. ML models, specifically eXtreme Gradient Boosting (XGBoost), were employed to identify predictors of IPV. Data were obtained from the American College Health Association National College Health Assessment (NCHA) conducted at the University of the West Indies Cave Hill campus during the 2021-2022 academic year
Results: Verbal abuse from partners was the most reported type of IPV (15.1 %). Stalking behaviour and physical violence were less common but non-trivial. Findings from ML models indicated students whose parents had associate degrees or technical training are at higher risk of experiencing IPV compared to those whose parents have either lower or higher levels of education. Additionally, married or partnered students and members of gender and sexual orientation minority groups are at higher risk of IPV exposure.
Conclusion: These findings provide novel insights into IPV exposure risk among university students in Barbados and the complex interplay of socio-demographic factors in predicting exposure. The results may be useful to inform the development of targeted interventions and support systems to address IPV on campus and in the broader community