Abstract: 

In this talk I will discuss some of the considerations when deciding how much data is ‘enough’ when looking to i) develop a new clinical prediction model (CPM) and ii) validate an existing CPM. When designing a study to develop a new CPM, researchers must ensure a large enough sample size to develop a model that predicts as accurately as possible. Conversely, when designing a study to validate an existing CPM, we must ensure a sample size large enough to estimate model performance accurately and precisely in an external sample.

I will introduce a series of work deriving closed-form solutions to identify the minimum required sample size for developing a new CPM (with either a continuous, binary, or time-to-event outcome), and validating an existing CPM (with a binary outcome). These formulae require researchers to pre-specify the expected value of certain parameters akin to standard sample size calculations for trials. I will show how to identify realistic values of these parameters based on published information (e.g., C statistic) for existing models in the same field.

I will also showcase the pmsampsize software package in R and Stata, which has been developed to implement the proposed methodology, and will illustrate this using examples of diagnostic and prognostic prediction models.

Mini-bio: 

Joie Ensor is a Lecturer in Medical Statistics at Keele University where he works primarily within the Centre for Prognosis Research. His research interests focus on methodological advances in prediction modelling, model validation, IPD meta-analysis, and diagnostic test evaluation. Joie is also a keen programmer, writing statistical software to enable others to use novel methodology.