Skip to main content

Table 2 ROC AUC [95% CI] performance comparison of the seven models applied to the internal and external validation datasets. Internal validation was estimated with 5-fold-nested cross-validation while external validation was performed on the YDX dataset

From: Logistic regression has similar performance to optimised machine learning algorithms in a clinical setting: application to the discrimination between type 1 and type 2 diabetes in young adults

Model

Internal validation

External validation

Gradient boosting machine

0.96 [0.94, 0.98]

0.93 [0.90, 0.96]

K-nearest neighbours

0.93 [0.90, 0.97]

0.92 [0.89, 0.95]

Logistic regression

0.96 [0.93, 0.98]

0.95 [0.92, 0.97]

MARS

0.96 [0.90, 0.99]

0.94 [0.92, 0.97]

Neural network

0.96 [0.93, 0.99]

0.94 [0.92, 0.97]

Random forest

0.95 [0.92, 0.98]

0.94 [0.91, 0.96]

Support vector machine

0.96 [0.93, 0.98]

0.94 [0.92, 0.97]