Letters Comparing risk prediction models

Authors’ reply to Noble and colleagues and Liew and colleagues

BMJ 2012; 345 doi: https://doi.org/10.1136/bmj.e4360 (Published 03 July 2012) Cite this as: BMJ 2012;345:e4360
  1. George C M Siontis, research associate1,
  2. Ioanna Tzoulaki, lecturer1,
  3. Konstantinos C Siontis, research associate1,
  4. John P A Ioannidis, professor2
  1. 1Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece
  2. 2Stanford Prevention Research Center, Department of Medicine, Stanford University School of Medicine, Stanford, CA 94305, USA
  1. jioannid{at}stanford.edu

We agree with Noble and colleagues that impact studies are important for understanding what can be achieved with different predictive models in real practice.1 2 In our systematic review we focused only on existing comparisons of the risk models’ predictive ability (as quantified by discrimination, calibration, or reclassification metrics). Evaluation …

View Full Text

Sign in

Log in through your institution