Methods for parameter-driven and wait time models

The project concerns the prediction of survival times for ovarian cancer patients. We have previously explored statistical methods for this purpose; however, it is unclear whether these methods provide enough accuracy and precision to be useful in practice. To study this issue further, we propose quantifying the quality of survival time predictions made by experienced gynecologic oncologists. If our statistical methods outperform these clinical methods, we will have found motivation for refining and implementing our methods for use in a clinical setting.

The student's role will be to design a web-based application that will allow physicians to assess patients' medical records and then enter their predictions of survival times. Subsequently, the student will quantify the error associated with these predictions and compare them with our statistically based predictions.