Time-to-event results based on interim analyses often exhibit large variability and uncertainty due to immature data. A common clinical question of relevance is to forecast how time-to-event results will look like with continued follow-up. For example, what is the predicted distribution of survival results with an additional 6 months of follow-up.
Using a Bayesian predictive probability approach, we have implemented a prediction model, assuming an underlying piece-wise exponential distribution of the survival times, to simulate future behavior based on the current data.
To demonstrate the approach, we present a comparison of the predicted and actual performance of three different clinical trial datasets. We have also developed an RShiny app based on the methodology, for ease of use and distribution.
|