PROJECTING CANCER INCIDENCE AND MORTALITY USING BAYESIAN AGE-PERIOD-COHORT MODELS

  • Dr. Saghir Bashir
  • Centro de Estatística e Aplicações da Universidade de Lisboa
  • Local: FCUL – Bloco C/6 Piso 4 Sala: 6.4.30 – (4ª feira) – 14:30
  • Quarta-feira, 23 de outubro de 2019
  • Referência Projeto: UID/MAT/00006/2019
I will revisit my post-doc work on projecting cancer incidence and mortality. Knowing the extent of future cancer burden can be used for healthcare planning (e.g. through disease reduction measures). Cancer incidence and mortality were projected in order to estimate what would have happened if past trends had continued into the future. Projections were made using a Bayesian age-period-cohort model using second degree auto-regressive smoothing on the age, period and cohort effects. The model performed well in making short to medium term projections when compared to other models. This approach could be used to make projections on more recent publicly available data on a larger scale especially given the advances in computing (both software and hardware).