Advanced Bayesian Methods for Medical Test Accuracy PDF

Preface

Bayesian methods are being used more and more in medicine and biology. For example, at the University of Texas MD Anderson Cancer Center and other institutions, Bayesian sequential stopping rules are implemented somewhat routinely in the design of clinical trials. Also, Bayesian techniques are being used more frequently in diagnostic medicine, such as estimating the accuracy of diagnostic tests and for screening large populations for various diseases. Bayesian methods are quite attractive in many areas of medicine because they are based on prior information, which is usually available in the form of related previous studies. An example of this is in the planning of Phase II clinical trials, where a new therapy will be administered to patients who have advanced disease. Such therapies are developed by pharmaceutical companies and their success depends on the success of previous Phase I or other relevant Phase II trials. Bayes theorem allows a logical way to incorporate the previous information with the information that will accrue in the future. Accuracy of a medical test is an essential component of the diagnostic process and is the key issue of this book. Of course, medicine and biology are not the only areas where the concept of test accuracy plays a paramount role. For example, in the area of sports (e.g., cycling or baseball), the accuracy of a test for "doping" is of extreme importance for maintaining the integrity of the sport

Advanced Bayesian Methods for Medical Test Accuracy is intended as a textbook for graduate students in statistics and as a reference for consulting statisticians. It will be an invaluable resource especially for biostatistics students who will be working in the various areas of diagnostic medicine (e.g., pathology and/or diagnostic imaging). The book is very practical and the student will learn many useful methods for measuring the accuracy of various medical tests. Most of the book is focused on Bayesian inferential procedures, but some is devoted to the design of such studies. A student should have completed a year of introductory probability and mathematical statistics, several introductory methods courses, such as regression and the analysis of variance, and a course that is primarily an introduction to Bayesian inference.

Consulting statisticians working in the areas of medicine and biology will have an invaluable reference with Advanced Bayesian Methods for Medical Test Accuracy, which will supplement the books Statistical Methods for Diagnostic Medicine by Zhou, Obuchowski, and McClish, and The Statistical Evaluation of Medical Tests for Classification and Prediction by Pepe. The two references

are not presented from a Bayesian viewpoint; thus, the present volume is unique and will develop methods of test accuracy that should prove to be very useful to the consultant. Another unique feature of the book is that all computing and analysis is based on the WinBUGS package, which will allow the user a platform that efficiently uses prior information. Many of the ideas in the present volume are presented for the first time and go far beyond the two standard references. For the novice, an appendix introduces the fundamentals of programming and executing BUGS, and as a result, the reader will have the tools and experience to successfully analyze studies for medical test accuracy.

A very attractive feature of the book is that the author's blog: http://medtestacc.blogspot.com provides the BUGS code, which can be executed as one progresses through the book and as one does the exercises at the end of each chapter. Note, each chapter includes the code labeled as BUGS CODE 4.1, BUGS CODE 4.2, etc., and this is also included in the author's blog; thus, the student can cycle between the book and the blog, which reinforces the subject in a beneficial manner.