Abstract:
|
The development of a new drug can be thought of as a stochastic process, where the candidate compound goes through a series of stages. At each stage, the candidate can be discarded, advanced to the next stage, kept in the current stage for further evaluation, or moved back to a previous stage to examine some unexpected complication. The management decisions that are made depend on the information developed for the candidate compound and information-developing for competing compounds (most of which will be in some earlier stage of development.) Historical experience, both at the company modeling the process and through publications from the industry as a whole, allow us to estimate the probability that a compound will be carried to the next stage and the mean and variance of the time a compound will spend in each stage. The necessity for management decisions at each stage introduces complexities in the model. When the process is simulated, the analyst can experiment with different management techniques in the presence of competing compounds at different intensities of competition. One such simulation is described.
|