The talk will describe the contributions to this class of problems from event history generalizations of survival analysis and its important complement: survival synthesis, where the many individual transition intensities are synthesized to transition probabilities. It is important to take advantage of the parallel development in biostatistics and sociometrics.
Several approaches have been proposed to handle this issue, among them the dynamic probabilistic causality framework of Arjas and Eerola, and several by James M. Robins (G-computation (being very similar to surival synthesis and the A-E approach), and structural nested failure time models).
The theory will be illustrated with analyses of Bone Marrow Transplant data.