Fig. 2. Precision-recall curves for each prediction modeling approach (using optimal tuning parameters, estimated in the validation dataset) in in both settings (mental health specialty and general medical) for 30- and 90-day suicide attempt (fatal and nonfatal) models and 90-day suicide death models.

Each figure panel represents precision-recall curves for different samples, outcomes, and follow-up periods. (a): 30-day suicide attempt prediction models, mental health specialty visits; (b): 30-day suicide attempt prediction models, general medical visits; (c): 90-day suicide attempt prediction models, mental health specialty visits; (d): 90-day suicide attempt prediction models, general medical visits; (e): 90-day suicide prediction models, mental health specialty visits, (f): 30-day suicide prediction models, general medical visits. Original parsimonious (yellow long-dashed line); Logistic regression with lasso variable selection (green long-short-dashed line); Random forest (blue short-dashed line); Artificial neural network (orange medium-dashed line); Ensemble = ensemble model using logistic regresssion, random forest, and artificial neural network models with detailed temporal predictors (solid black line).