Depression is a major global public health challenge, and social participation (SP) is a key preventive factor. A previous study found that the association between SP and reduced depression risk is particularly pronounced among older adults with low socioeconomic status (SES), suggesting its potential to reduce mental health disparities. However, the extent to which baseline risk is correlated with the magnitude of benefit derived from SP remains unclear. Using data from Japan Gerontological Evaluation Study data for adults aged ≥65, we aimed to (i) develop a prediction model for depression, (ii) identify key predictors, and (iii) evaluate the correlation between predicted risk and preventive benefit to identify populations at high risk and high benefit. Using baseline covariates in 2016 as predictors and depressive symptoms (GDS≥5) in 2022 as the outcome, we compared logistic regression, lasso regression, random forest, XGBoost, and SuperLearner models. For the model with the highest AUC and lowest Brier score, we assessed variable importance. We then examined the correlation between predicted depression risk and individual-level benefit of SP from causal forest model using Spearman’s rank correlation. Among all models, lasso regression outperformed others (AUC=0.741 [95% CI: 0.724-0.759]; Brier score of 0.12). The most influential predictors in the model were 10 baseline GDS items. Predicted depression risk was positivity correlated with the predicted benefit of SP (ρ=0.40). Individuals in the high-risk and high-benefit group were more likely to be older and have lower SES, characterized by unemployment, unmarried or living alone, and lower educational attainment and household income. These findings indicate that individuals at higher risk of depression are also those who derive greater preventive benefit from SP. Promoting SP may therefore be an effective strategy to both prevent depression and reduce socioeconomic disparities in mental health.