Log‐Logistic Regression Models for Survival Data

The log‐logistic distribution has a non‐monotonic hazard function which makes it suitable for modelling some sets of cancer survival data. A log‐logistic regression model is described in which the hazard functions for separate samples converge with time. This also provides a linear model for the log odds on survival by any chosen time. The model is fitted on GLIM and an example is given of its use with lung cancer survival data.