Pediatric Cardiac Re-Transplantation

Even in cases without acute rejection, the risk associated with a second heart transplant for pediatric patients appears to be substantially greater than that associated with an initial transplant — even when controlling for the fact that multiple transplant patients tend to be sicker. I’m analyzing transplant data using a version of BART suitable for survival analysis, which is both more flexible than parametric models and avoids making standard assumptions about survival curves like that of proportional hazards. The use of $t$ and $\chi^2$ tests, Kaplan Meier estimators, and Cox models is pervasive in medical survival analysis; I hope to establish BART as a viable alternative and one that is better suited to identifying interesting nonlinear, interacting, or subgroup-specific effects.