American Statistical Association
Dynamic treatment regimes (or treatment policies) are used to operationalize multi-stage decision making in the medical field. Common approaches to constructing dynamic treatment regimes from data, such as Q-Learning, employ non-smooth functionals of the data. The non-smoothness leads to non-regular asymptotics under common data generating models. As a result, methods that ignore the non-regularity have poor performance in small samples. In this talk, we present a bootstrap based method for constructing asymptotically valid confidence sets. This method is adaptive in the sense that it provides exact coverage when the true underlying generative model leads to regular asymptotics and is conservative otherwise. We discuss how a variety of modern statistical procedures exhibit this nonregularity and provide interesting statistical challenges in developing measures of confidence.
Susan A. Murphy is H.E. Robbins Professor of Statistics, Research Professor at the Institute for Social Research, and Professor of Psychiatry at the University of Michigan Medical School. Dr. Murphy’s primary research concerns individually tailored treatments, sometimes called adaptive treatment strategies or dynamic treatment regimes. In particular, she works on experimental designs and analysis methods with the goal of constructing adaptive treatment strategies in substance abuse and mental health. She is involved in a variety of experimental studies designed to inform the development of adaptive treatment strategies; these involve studies in ADHD, autism, drug dependence by pregnant women and treatment for alcohol dependence.
She has received numerous NIH grants to fund this work both in psychiatry and substance abuse and led a NIH-roadmap funded network of scientists including psychiatrists, infectious disease specialists, computer scientists, engineers and statisticians on identifying and working to solve the methodological problems arising in the construction of adaptive treatment strategies. She is published in many peer-reviewed journals such as Biometrics, Drug and Alcohol Dependence, Journal of Consulting and Clinical Psychology, Journal of Machine Learning Research, Annals of Behavioral Science, Journal of the American Statistical Association, Journal of the Royal Statistical Society and Annals of Statistics.
|Date:||Monday, November 8, 2010|
|Time:||12:00 - 1:30 P.M.|
1255 Amsterdam Avenue
New York, New York