Abstract
This article addresses a topic not considered in previous models of patient treatment: the possible downstream availability of improved treatment options coming out of the medical research and development (RD) pipeline. We provide clinical examples in which a patient may prefer to wait and take the chance that an improved therapy comes to market rather than choose an irreversible treatment option that has serious quality of life ramifications and would render future treatment discoveries meaningless for that patient. We then develop a Markov decision process model of the optimal time to initiate treatment, which incorporates uncertainty around the development of new therapies and their effects. After deriving structural properties for the model, we provide a numerical example that demonstrates how models that do not have any foresight of the RD pipeline may result in optimal policies that differ from models that have such foresight, implying erroneous decisions in the former models. Our example quantifies the effects of such errors.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 632-642 |
| Number of pages | 11 |
| Journal | IIE Transactions (Institute of Industrial Engineers) |
| Volume | 42 |
| Issue number | 9 |
| DOIs | |
| State | Published - Sep 2010 |
| Externally published | Yes |
Funding
We would like to thank Turgay Ayer, Fatih Erenay, Mahesh Nagarajan, Martin Puterman, and Greg Werker for their helpful comments. We also thank three anonymous referees and the Associate Editor for all of their thoughtful feedback. Steven Shechter was supported by the Natural Sciences and Engineering Research Council Discovery Grant (341415-07). Oguzhan Alagoz was supported by the National Science Foundation Grant (CMMI-0700094).
Keywords
- Markov decision processes
- Medical decision making
- Research and development
ASJC Scopus subject areas
- Industrial and Manufacturing Engineering