Table 2.

Discrimination of machine learning algorithms and logistic regression models across the various prediction settings.

Models, AUC (95% CI)*	Prediction setting (used variables: predicted outcome)
	Baseline: post-mTICI	Baseline: mRS	All variables: mRS
Super learner	0.55 (0.54–0.56)	0.79 (0.79–0.80)	0.90 (0.90–0.91)
Random forests	0.55 (0.55–0.56)	0.79 (0.79–0.79)	0.91 (0.90–0.91)
Support vector machine	0.53 (0.53–0.54)	0.78 (0.77–0.78)	0.88 (0.88–0.89)
Neural network	0.53 (0.53–0.54)	0.77 (0.76–0.77)	0.88 (0.88–0.89)
LR: AUTOMATED SELECTION**
Random forests	0.55 (0.55–0.56)	0.78 (0.78–0.78)	0.90 (0.90–0.90)
LASSO	NA¥	0.78 (0.78–0.79)	0.90 (0.89–0.90)
Elastic net	NA¥	0.77 (0.77–0.78)	0.89 (0.88–0.89)
Backward elimination	0.57 (0.57–0.58)	0.78 (0.77–0.78)	0.90 (0.89–0.90)
LR: prior knowledge‡	0.55 (0.55–0.58)	0.78 (0.78–0.79)	0.90 (0.90–0.90)

^*Model discrimination is assessed by calculating mean Area Under the Curve (AUC) of the receiver operating characteristic across all outer cross-validation folds.

^**Logistic regression using automated variable selection methods.

^¥Variable selection not possible, likely due to insufficient signal-to-noise ratio.

^‡Logistic regression using variables based on prior knowledge.