Summer 2007 - Challenges in Dynamic Treatment Regimes and Multistage Decision-Making
Workshop Information
Schedule
Tutorials: June 18-20, 2007
Monday, June 18, 2007
Room H, 3rd Floor
8:00-9:00 a.m. | Registration and Continental Breakfast |
9:00 a.m. | Introduction and Welcome Marie Davidian, North Carolina State University |
9:30-10:45 | Introduction to Causal Inference Miguel Hernan, Harvard School of Public Health |
10:45-11:00 | Break |
11:00-12:15 | Introduction to Causal Inference (continued) Miguel Hernan, Harvard School of Public Health |
12:15-2:00 p.m. | Lunch (Room FG, 3rd Floor) |
2:00-3:15 | Introduction to Dynamic Treatment Regimes Butch Tsiatis, North Carolina State University |
3:15-3:30 | Break |
3:30-4:45 | Introduction to Dynamic Treatment Regimes (continued) Butch Tsiatis, North Carolina State University |
Tuesday, June 19, 2007
Room H, 3rd Floor
8:30-9:00 a.m. | Registration and Continental Breakfast |
9:00-10:30 | RL with Additional Discussion of Connections to Classification Ron Parr, Duke University |
10:30-10:45 | Break |
10:45-12:15 | RL with Additional Discussion of Connections to Classification (continued) Ron Parr, Duke University |
12:15-2:00 p.m. | Lunch (Room FG, 3rd Floor) |
2:00-3:15 | Computational Challenges with High Dimensional Data Joelle Pineau, McGill University |
3:15-3:30 | Break |
3:30-4:45 | Computational Challenges with High Dimensional Data (continued) Joelle Pineau, McGill University |
Wednesday, June 20, 2007
Room H, 3rd Floor
8:30-9:00 a.m. | Registration and Continental Breakfast |
9:30-10:30 | Introduction to Mechanistic Models and Control Theory Daniel Rivera, Arizona State University |
10:30-10:45 | Break |
10:45-12:00 | Introduction to Mechanistic Models and Control Theory (continued) Daniel Rivera, Arizona State University |
12:00-1:30 p.m. | Lunch (Room FG, 3rd Floor) |
1:30-2:00 p.m. | Poster Presentation Session (2 minutes each, up to one transparancy) |
2:00-3:15 | Introduction to Nonstandard Statistical Inference Experimental Trials Susan Murphy, University of Michigan |
3:15-3:30 | Break |
3:30-4:45 | Introduction to Nonstandard Statistical Inference (continued) Susan Murphy, University of Michigan |
5:30-7:30 | Poster Session and Reception, Room BC (Please have your poster set up by 5:15) SAMSI will provide poster presentation boards and tape. The board dimensions are 4 ft. wide by 3 ft. high. They are tri-fold with each side being 1 ft. wide and the center 2 ft. wide. Please make sure your poster fits the board. The boards can accommodate up to 16 pages of paper measuring 8.5 inches by 11 inches. |
Opening Workshop (Thursday, June 21 - Friday, June 22 at Radisson RTP):
The workshop featured one or two overview talks presenting the "big picture" of the methodological challenges followed by more advanced and targeted talks on research relevant to development of dynamic treatment regimes that build on the foundation provided by the tutorials. There were talks the first full day and second morning, after which participants were divided into discussion groups centered around four key areas that formed the basis for "brainstorming" by working groups during the next week of the program (below). The discussion groups developed lists of important challenges and questions centered around their theme.
Thursday, June 21, 2007
Room H, 3rd Floor
8:00-9:00 a.m. | Registration and Continental Breakfast |
9:00-9:05 | Welcome |
9:05-9:35 | Introduction by Susan Murphy, University of Michigan |
9:35-10:05 | Sample Complexity of Policy Search with Known Dynamics Ambuj Tewari, Univ. of California-Berkeley |
10:10-10:35 | Discussion What is the promise of work in sample complexity and upper bounds on generalization error? |
10:35-10:50 | Break |
10:50-11:20 | Clinical data based optimal STI strategies for HIV: a reinforcement learning approach Damien Ernst, Supélec |
11:25-11:55 | From Population to Individual Drug Dosing in Chronic Illness - Intelligent Control for Management of Renal Anemia Adam Gaweda, University of Louisville |
12:00-12:30 | Discussion What are the issues with model based work? |
12:30-1:45 | Lunch (Room FG, 3rd Floor) |
1:45-2:15 | Estimation of the effect of dynamic treatment regimes under flexible dynamic visit regimes Andrea Rotnitzky , Di Tella University and Harvard University |
2:20-2:50 | Asymptotic Bias Correction for Estimates of Optimal Dynamic Treatment Regimes Erica Moodie, McGill University |
2:55-3:20 | Discussion What are the issues that concern statisticians? |
3:20-3:35 | Break |
3:35-4:05 | Adaptive stimulation design for the treatment of epilepsy Joelle Pineau , McGill University |
4:10-4:40 | Bias and Variance in Value Function Estimates Peng Sun, Duke University |
4:45-5:15 | Wrap-up discussion. Plan for Friday. |
Friday, June 22, 2007
Room H, 3rd Floor
8:30-9:00 a.m. | Registration and Continental Breakfast |
9:00-10:30 | Tutorial/Talk --Inference for Dynamic Regimes Jamie Robins, Harvard University |
10:30-10:45 | Break |
10:45-12:00 | Tutorial/Talk --Inference for Dynamic Regimes (continued) Jamie Robins, Harvard University |
12:00-1:30 | Lunch (Room FG, 3rd Floor) |
1:30-3:00 | Discussion of Working Group Projects and Activities |
Working Groups (Monday, June 25 - Wednesday, June 27 at SAMSI):
Four Working Groups convened to discuss and prioritize challenges in their respective areas. Participants identified the most pressing problems and outline modes of attack and specific research directions to be pursued. The working group foci and potential lead participants are:
- Difficulties In Statistical Inference (Peter Bartlett, Susan Murphy, Sasha Rakhlin, Jamie Robins)
- Bayesian Approaches (Brad Carlin, Peter Thall)
- The Role of Mechanistic Models (Tom Banks, Victoria Chen, Marie Davidian, Daniel Rivera)
- Practical Challenges and Applications (Erica Moodie, Joelle Pineau, Butch Tsiatis)
Working Groups will met daily according to a schedule that will allow participants to be involved with more than one group if desired.
Transition Workshop (Thursday, June 28 - Friday, June 29 at SAMSI):
Each Working Group presented their results, findings, and recommendations for the future to all Working Group members along with additional participants who may return for this final activity. Discussion followed each group's presentation. These presentations and discussions formed the basis for a white paper outlining methodological challenges in the area of dynamic treatment regimes to be written by the Program Leaders for submission to a leading statistical or mathematical science journal, with input from participants.