Sankhya: The Indian Journal of Statistics

2004, Volume 66, Pt. 2, 385--398

Optimal Scheduling of Intermediate Examination Times in Simple Illness-Death Model

Anup Dewanji and Atanu Biswas , Indian Statistical Institute, Kolkata, India

SUMMARY. Simple illness-death model arises in  many examples of survival and reliability studies with industrial products. In a typical illness-death model, one is often interested in the event of occurrence of illness ($D$) which is assumed to be unobservable. This event is followed by the event of failure or death ($F$) which is observable in addition to the presence of illness. Failure ($F$) can also be observed before the illness occurs in which case the absence of disease is also recorded. In this work, we consider the problem of finding one or more intermediate examination times in order to make inference on $D$ and/or also to safeguard the event of failure with illness before it is detected at an examination time. The standard likelihood based criteria are difficult to apply since calculation of the expected information  matrix is not straightforward. We discuss two new optimality criteria which are based on some simple probability calculations and easy to apply.

AMS (1991) subject classification.  62K05, 62L05.

Key words and phrases. Illness-death model, optimal designs, performance characteristics, cancer screening studies, risk, adaptive designs.

 Full paper (PDF)