Publication Description
We present a flexible family of link-based regression models with time-independent (baseline) covariates. Particular choices of the link function yield a proportional hazard or a proportional odds model, among others. The background hazard function in the model is estimated using B-splines. This yields a full likelihood under both continuous time and random interval censoring. Residuals for influence and goodness-of-fit are described. Two examples are shown, one in continuous time and the other under interval censoring.