Priors from Envisioned Posterior Judgments: A Novel Elicitation Approach With Application to Bayesian Clinical Trials

Abstract

Background: The uptake of formalized prior elicitation from experts in Bayesian clinical trials has been limited due to challenges such as complex statistical modeling, lack of practical tools, and the cognitive burden placed on experts arising from needing to quantify their uncertainty probabilistically. Existing methods also fail to address prior-posterior coherence, i.e., how do we ensure that the posterior distribution, obtained mathematically from combining the estimated prior with the trial data, reflects the expert's actual posterior beliefs? Method: In this study, we propose a new elicitation approach that effectuates prior-posterior coherence and reduces cognitive burden. This is achieved by eliciting expert responses, comprising point estimates only, about envisioned posterior judgments under various data outcomes and inferring the prior distribution by minimizing discrepancies between these responses and expected responses derived from the posterior distribution. Via an iterative process, experts receive feedback on the degree of coherency of their responses, and are invited to revise their responses to achieve greater coherency. The feasibility and potential value of this new approach are illustrated through an application to an ongoing trial. Results: We involved 10 experts from Walk 'n watch trial research team. Experts were presented with 16 hypothetical outcome scenarios to experts and elicit the priors followed by the developed elicitation framework. Following two rounds of elicitation, experts' judgments showed substantial improvement in coherency, demonstrating the practical applicability of the proposed elicitation approach. Conclusion: The proposed method provides a practical solution to the challenges of formalized prior elicitation in Bayesian clinical trials by addressing prior-posterior coherence and reducing cognitive demands on experts.