Skip to main content

Table 4 Results of generalized estimated equations (GEE) for the percentage change in AUC between 10 development and 245 validation studies

From: Does poor methodological quality of prediction modeling studies translate to poor model performance? An illustration in traumatic brain injury

 

Percentage change in AUC (95% CI)

Intercept

9.5% (5.5, 13.4)

Risk of bias (low)

High

− 31.7% (− 48.2, − 15.2)

Unclear

− 13.4% (− 16.4, − 10.3)

Study design (similar)

Cohort to trial

− 18.5% (− 26.2, − 10.8)

Trial to cohort

0.19% (− 3.7, 4.1)

  1. The generalized estimated equations (GEE) model includes a random intercept on model level (N = 18), risk of bias assessment (low, high, unclear based on the original PROBAST), and similarity in study design between the development and validation study (Similar, Cohort to trial, Trial to cohort) to estimate the percentage change in AUC between the development and validation studies. The intercept indicates the percentage change in AUC for low risk of bias models with a similar study design between the development and validation study
Back to article page

Diagnostic and Prognostic Research

ISSN: 2397-7523

Contact us