Table 2.

Model performance of non-regularized and elastic net regularized logistic regressions

	Logistic regression		Elastic net
	Mean AUC (95% CI)	Model fit	Mean AUC (95% CI)	Model fit	λ	α
Year 1
Eating disorder diagnosis	0.48 (0.41–0.56)	Extremely poor	0.62 (0.53–0.71)	Poor	0.070	0.2
Binge eating	0.69 (0.65–0.73)	Poor	0.77 (0.73–0.81)	Fair	0.082	0.8
Compensatory behaviors	0.83 (0.80–0.86)	Good	0.88 (0.87–0.90)	Good	0.309	0.1
Underweight BMI	0.80 (0.74–0.85)	Good	0.93 (0.91–0.95)	Excellent	0.025	1
Year 2
Eating disorder diagnosis	0.47 (0.40–0.55)	Extremely poor	0.61 (0.56–0.67)	Poor	0.110	0.3
Binge eating	0.64 (0.61–0.68)	Poor	0.71 (0.68–0.74)	Fair	0.150	0.3
Compensatory behaviors	0.72 (0.66–0.77)	Fair	0.85 (0.81–0.89)	Good	0.100	0.5
Underweight BMI	0.70 (0.65–0.77)	Fair	0.89 (0.86–0.91)	Good	0.178	0.3

AUC, Area Under the Receiving Operating Characteristics Curve; BMI, body mass index; CI, confidence interval.

The λ (penalization) and α (mixing) parameters were identified through 10-fold cross-validation repeated three times for the elastic net models.