ABSTRACT
In an ideal healthcare setting, caregivers can account for relevant factors, such as treatment history or genetic variations, when making treatment decisions for chronic diseases. These treatment plans are referred to as dynamic treatment regimes (DTRs). This chapter presents a literature review of statistical methods used to estimate DTRs in the context of using data collected in a Sequential Multiple Assignment Randomization Trials (SMART). A DTR is a treatment strategy that tells the caregiver which adaptive, sequential treatment plan will most likely lead to the highest probability of success. A SMART design is an experimental trial with the purpose of determining which DTRs are optimal for which patients. A SMART is able to evaluate the various DTRs because it compares them by randomizing patients to a set preselected treatment at each phase. A SMART can have two goals: compare a small number of pre-specified DTRs embedded in the SMART design or construct new DTRs which may not be naturally embedded in the design.
