Doranne Thomassen

Introduction Prediction models are increasingly used by clinicians to communicate likely outcomes for individual patients during the clinical decision-making process. Although the prediction from such a model, when used for an individual, represents the model’s best estimate of the expected outcome, this estimate is associated with uncertainty.

?Mathematical formulae have been encoded as MathML and are displayed in this HTML version using MathJax in order to improve their display. Uncheck the box to turn MathJax off. This feature requires Javascript. Click on a formula to zoom. Abstract Sample size is an essential indicator of the uncertainty in clinical research results. When studies present time-to-event outcomes with Kaplan-Meier curves, these are often accompanied by the remaining number of patients at risk in a table below the curve.

1 INTRODUCTION When using a statistical model to predict a patient's outcome, there is uncertainty in predictions, even when the model is correctly specified. This uncertainty arises from the fact that a prediction model is estimated using a random sample of limited size from the target population. Typically, sampling uncertainty about predictions varies between individual patients, as not all patient profiles are represented uniformly in model development.