From Ewout Steyerberg

There have been proposals by Royston in the direction of stacking I believe:

• here
• Recently in Stat Med more on doing analyses over all sets here

Daniele Giardiello d.giardiello@nki.nl 2017-03-28

I am Daniele a PhD student currently working with professor Ewout Steyerberg. Here a list of the papers that can be useful for you:

• Wahl (2016) - Assessment of predictive performance in incomplete data by combining internal validation and multiple imputation
• Wood (2015) – The estimation and use of predictions for the assessment of model performance using large samples with multiply imputed data

In the simplest way, you could calculate the apparent c-index for each imputed data and average them to get the pooled apparent c-index. In case of large samples the correction for optimism could be avoided.