Fig. 4: Test area under the curve (AUC) for the random forest (RF) and logistic regression (LR) classifiers given all patient history (sociodemographic information, date of prescription, diagnostic and procedure codes) for Site A and Site B, showing modest decrement in discrimination at Site B (mean 0.64 vs 0.67 for LR classifier, and 0.68 vs 0.69 for RF classifier). | Translational Psychiatry

Fig. 4: Test area under the curve (AUC) for the random forest (RF) and logistic regression (LR) classifiers given all patient history (sociodemographic information, date of prescription, diagnostic and procedure codes) for Site A and Site B, showing modest decrement in discrimination at Site B (mean 0.64 vs 0.67 for LR classifier, and 0.68 vs 0.69 for RF classifier).

From: Predicting treatment dropout after antidepressant initiation

Fig. 4

Confidence intervals computed using 500 bootstraps.

Back to article page