K.C.S.Pillai Memorial Lecture
Individualizing Dynamic Treatments: Statistical Challenges and Some Solutions
Susan Murphy
Professor
University of Michigan
Start Date and Time: Thu, 22 Apr 2010, 4:30 PM
End Date and Time: Thu, 22 Apr 2010, 5:30 PM
Venue: MATH 175
Refreshments: Refreshments will be served at 4:00 PM in HAAS 111.
Abstract:
An emerging and exciting area of clinical science involves the use of data to directly inform the tactics and strategies used to provide treatment in clinical practice. Clinical practice involves the use of dynamic information on patients to individualize treatment, that is, clinical practice requires sequential decision making. To a large extent, the development of methods to inform sequential decision making has occurred outside of statistics (computer science, operations research, engineering). However the developed methods are primarily algorithmic; these methods either do not provide inferential tools, or at best, provide ad hoc inferential tools.