ABSTRACT
David Williams fitted a log-normal survival-time distribution to a set of data from aquatic toxicology. Additional work in the area is described by Finkelstein who provides full details of a direct analysis of interval-censored failure time data using a proportional hazards model, and in a pair of similar papers by F. J. Aranda-Ordaz and R. J. Tibshirani and A. Ciampi. Frequently, however, mechanistic models will have a clear justification, and estimates are required of the model parameters. A wide range of approaches is available for such data, the spectrum ranging from very crude descriptive measures, to very complex stochastic mechanistic models on the other. Unconventional survival analysis models can be very difficult to analyse and fit to data. The Puri–Senturia model and its extension make use of the terms of survival analysis, through the hazard and the survivor function. In the last case a compartment model is used for time-to-response data for fish swimming in tanks contaminated with a toxicant.
