Expert knowledge is a valuable source of information to augment available data or when interpretation/synthesis of data requires expert judgement. Prior elicitation is a key tool for translating this expert knowledge and judgement into a quantitative probability distribution that can then be used in the design, analysis, and interpretation of clinical and observational studies.
On-Demand | 1 hour | Your Desk! |
![]() |
Nicky BestVP and Head of Statistics and Data Science Innovation, GSK |
![]() |
Brad CarlinSenior Advisor, Data Science, PharmaLex |
In this talk, we review the different approaches to elicit information from experts and summarize it in a probabilistic language, Bayesian prior distributions. We then focus on how expert knowledge from multiple experts can be summarized. Software to achieve such summaries is also discussed. The concepts and approaches are illustrated using a variety of real-life examples relating to different aspects of pharmaceutical product development. The talk shows Bayesian prior elicitation to be a feasible and useful aid to internal company decision making across the pharmaceutical product lifecycle chain.