Dynamic survival models for predicting the future in health technology assessment and incorporating external evidence

Abstract: Predicting future survival outcomes is often a key task within health technology assessment, and a number of different survival models can be used. Future predictions can be sensitive to model choice, and may be improved by incorporating external evidence. In this talk, I shall introduce dynamic survival models, which combine survival and time series methods. The performance of these models is compared against other models in a simulation study of survival from a disease when some patients are 'cured', giving mixture cure models. The use of dynamic survival models is also demonstrated by replicating an existing NICE appraisal. Potential areas for future research are presented.