Interpretable Machine Learning for Identifying High-Risk Depression in Men Who Have Sex with Men: A Cross-Sectional Study Based on the Sexual Minority Stress Model (Preprint)

<sec> <title>BACKGROUND</title> Due to unique psychosocial stressors, men who have sex with men (MSM) are at a substantially higher risk of developing depressive symptoms compared with the general population. However, commonly used depression screening tools are primarily designed for the general population and fail to adequately account for MSM-specific stressors, limiting accurate risk identification in this group. In the Chinese context, the mental health of MSM is often underestimated, and there is a lack of precise assessment tools that integrate multidimensional psychosocial variables. This study aims to develop an interpretable machine learning framework to identify key risk factors for depression and enable risk stratification, thereby detecting MSM individuals at high risk for depressive symptoms and supporting timely early intervention. </sec> <sec> <title>OBJECTIVE</title> This study aimed to develop machine learning models and tools to identify MSM at high risk of depression and to assess their depression risk. </sec> <sec> <title>METHODS</title> Between November and December 2022, this study recruited 1,369 MSM using a non-probability sampling approach. Based on the Minority Stress Model, candidate risk factors were collected, encompassing sociodemographic characteristics, MSM-related stigma, psychological resilience, social support, and MSM sexual identity, with depressive symptoms as the primary outcome. 3 feature selection methods combined with ten machine learning algorithms were employed to construct predictive models, and model performance was evaluated to identify the optimal model. SHAP analysis was then applied to interpret the contributions of individual risk factors. To enhance clinical applicability, the risk probabilities of the optimal model were defined as risk scores (RS), which were subsequently used for risk stratification. </sec> <sec> <title>RESULTS</title> Following feature selection, 8 potential risk factors significantly associated with depressive symptoms were identified, which were then used to construct a depression risk identification model. Among all compared algorithms, the LightGBM model demonstrated the best performance, achieving an AUC of 0.945 and exhibiting superior overall performance across multiple evaluation metrics. Key risk factors included psychological resilience, identity superiority, experiences of stigma, and internalized homophobia. Risk stratification analysis revealed that individuals in the high-risk group had a significantly higher prevalence of depressive symptoms compared with those in the low-risk group (P &lt; 0.001). Visualization tools were employed to intuitively illustrate the model’s classification threshold and its potential practical utility. </sec> <sec> <title>CONCLUSIONS</title> Machine learning models can effectively identify MSM individuals at high risk for depressive symptoms by integrating multidimensional features. Risk stratification based on model-derived probability scores facilitates early detection and targeted interventions, thereby supporting personalized mental health screening for this vulnerable population. </sec>