Predicting HIV self-testing intentions among Chinese college students: a dual-model analysis integrating health belief constructs and machine learning prioritization.

Introduction: As college students emerge as a key HIV-vulnerable population in China, HIV self-testing (HIVST) presents a critical strategy for enhancing detection rates and enabling timely intervention. While observational studies have identified multifactorial influences on HIVST willingness, few investigations integrate behavioral theory with machine learning approaches among college students. This study aims to fill this gap by exploring the determinants of HIVST willingness among college students using the Health Belief Model (HBM) and random forest analytics. Methods: This cross-sectional study employed stratified cluster sampling to recruit 1,015 undergraduates from Xiangnan College (July-August 2022), The Health Belief Model (HBM) was synthesized with random forest analytics to elucidate determinants of HIVST willingness. Data were collected through questionnaires, and logistic regression and random forest modeling were used for analysis. Results: Among participants, 69.3% ( Discussion: Our dual-method analysis establishes that campus HIV control necessitates: 1) Gender-specific prevention programs addressing male students' elevated risk exposure; 2) HBM-informed education strengthening self-efficacy and environmental cues; 3) Structural interventions reducing testing barriers through discreet service delivery. This theoretical-empirical integration advances predictive understanding of HIVST behaviors, providing actionable insights for developing precision public health strategies in academic settings.