Event Details

Tom Park of Berry Consultants will present a seminar to Department of Management Science

Abstract: Many clinical trials are part of a larger process of drug development, and when we do try to optimize the design of the trial we often try to optimize some outcome specific to the trial - its power, duration, number of subjects, or ability to inform our post trial decision making such as selecting “the right dose” or choosing which patient subgroups or disease sub-types to proceed in. But we don’t know that we’ve optimized the right characteristic, or made the right trade-off between the different characteristics. Is it, for example, better to be quicker and cheaper, quicker and more expensive, or slower but more accurate?

Drug development is a very unusual activity, perhaps only oil exploration as a project has such high upfront costs, such high potential pay-offs, but such small probability of success. Normal business decision tools don’t seem to have the right structure to capture the process. Perhaps we can make more optimal design decisions if we simulate not just the trial in question, but the whole development process? We can then see the effects of trade-offs in time, cost and accuracy on the expected Net Present Value of the program? How hard is this to do? How sensible is it to try? What do the results look like when we do?

We’ve written a software package that does this and I’ll share some results."