Table 5 Performances of the models measured using stratified K-fold cross-validation.

From: Complement C3 identified as a unique risk factor for disease severity among young COVID-19 patients in Wuhan, China

 

K in cross-validation

AUC

Brier score

Logistic regression—AIC

Logistic regression—BIC

Random forest

Logistic regression—AIC

Logistic regression—BIC

Random forest

Young patients

3

0.5512

0.5542

0.5962

0.2740

0.2584

0.2438

4

0.5405

0.5454

0.5847

0.2764

0.2606

0.2462

5

0.5327

0.5344

0.5838

0.2772

0.2640

0.2461

All patients

3

0.6069

0.6186

0.6310

0.2502

0.2425

0.2369

4

0.6027

0.6140

0.6271

0.2525

0.2444

0.2378

5

0.6059

0.6156

0.6276

0.2500

0.2433

0.2378

  1. For each value of K, the stratified K-fold cross-validation was repeated twenty times and the results averaged. Higher values of the AUC and smaller values of the Brier score are preferable.
Back to article page