The future of model-based drug development will likely include greater use of modeling in a Bayesian framework. It's the "logical outcome" of the integration of three trends, says William Gillespie of the nonprofit Metrum Institute.
The first trend is science's growing understanding of the mechanics of drug action and disposition. Increased use of Bayesian modeling and simulation is the second trend, followed by growing economic and organizational commitment to modeling and simulation.
In a presentation developed for a Statistical and Applied Mathematical Sciences Institute workshop held earlier this month, Gillespie says drug developers need a "more integrated and longitudinal approach to strategic modeling and simulation."
Too many modeling and simulation activities are reactive to short-term needs, he says, leading to "quick-and-dirty" efforts. Such efforts have limited value for other development programs, and even for later phases of same program
Gillespie's presentation includes two drug development examples: trial simulation to design a Phase II dose-finding strategy, and multi-scale systems biology modeling of bone and calcium homeostasis. He finds that use of preclinical and public clinical data aids researchers in building a model for predicting clinical outcomes for the drug candidate. Another finding: leveraging prior information permits more efficient design and analysis of a Phase II trial to select a dose for Phase III.
- here are the presentation slides