This book was written to summarize and describe the state of the art of statistical
methods developed to address questions of estimation and inference for dynamic
treatment regimes, a branch of personalized medicine. The study of dynamic treatment
regimes is relatively young, and until now, no single source has aimed to provide
an overview of the methodology and results which are dispersed in journals,
proceedings, and technical reports so as to orient researchers to the field. Our primary
focus is on description of the methods, clear communication of the conceptual
underpinnings, and their illustration via analyses drawn from real applications as
well as results from simulated data. The first chapter serves to set the context for the
statistical reader in the landscape of personalized medicine; we assume a familiarity
with elementary calculus, linear algebra, and basic large-sample theory. Important
theoretical properties of the methods described will be stated when appropriate;
however, the reader will, for the most part, be referred to the primary research articles
for the proofs of the results. By doing so, we hope the book will be accessible to
a wide audience of statisticians, epidemiologists, and medical researchers with some
statistical training, as well as computer scientists (machine/reinforcement learning
researchers) interested in medical applications