The reasons might be two-fold:

• (1) AIDS has become a major worldwide epidemic since it was first reported in the United States in 1981. To prevent and treat this disease, extensive research on AIDS is still required and will be needed for years to come.
• (2) The methodology is general enough that it can be applied to many studies.

The statistical methodology proposed in this paper can be used in many applications where the focus is to determine whether a biological marker can be used to predict survival outcome.

It is the topic of my dissertation. My coauthor, Professor Jeremy G. Taylor of the University of Michigan, is the director of the University of Michigan Cancer Center Biostatistics Unit. Dr. Taylor was my dissertation advisor and is a leader in developing methodologies for analyzing AIDS data.

In many medical studies it is common that a biological marker is collected repeatedly over time to monitor a disease’s progress. A treatment might affect the disease outcome via various pathways (e.g. completely through the marker, or partially through the marker, or independent of the marker.) It is logical and important to study the interrelationships among treatment, the marker, and the disease outcome simultaneously. Our paper proposed a flexible statistical approach to analyze these interrelationships and is general enough to be used in many applications.