|
|||||
|
|
Clinical Trials with Exponential Survival Times |
|
| Authors: | Xikui Wang Qier Tan Mikelis G Bickis |
| Institution: | 1. Department of Statistics , University of Manitoba , Winnipeg, Manitoba, Canada xikui_wang@umanitoba.ca;3. Department of Statistics , University of Manitoba , Winnipeg, Manitoba, Canada;4. Department of Mathematics and Statistics , University of Saskatchewan , Saskatoon, Saskatchewan, Canada |
| Abstract: | We consider a bandit process with delayed responses which are exponentially distributed survival times. The objective is to maximize the expected value of the total response from all selections. We formulate the problem and show that the optimal strategy is characterized by a sequence of break-even values. A monotonicity property of this sequence is derived, which implies the non-optimality of the myopic strategy and a special optimal stopping solution. An example is included to illustrate a possible application of the main results. |
| Keywords: | Bandit processes Bayesian method Clinical trial Exponential distribution Survival time |