In-person seminar: Dr. Evan Kwiatkowski on Adaptively incorporating external data in Bayesian clinical trial design
When & Where
September 5, 2023
12:00 PM - 1:00 PM
RAS 102B ( View in Google Map)
Contact
- Scott Dyson
- [email protected]
Event Description
When: 09/05/2023 12 noon
Where: RAS-102B, See event website for WebEx. Password: xnFXMhnm638
Presenter: Dr. Dr. Evan Kwiatkowski
Abstract:
Bayesian methods provide a natural approach for incorporating external data into the analysis of a clinical trial through an informative prior distribution. Prior-data conflict occurs when the prior primarily favors regions of the parameter space that are far from that supported by the likelihood for the trial data. The predictive distribution based on the trial data can be used to assess the compatibility or relevance of the external data (i.e., Box’s p-value). We determine the compatibility of a prior and the trial data using posterior-predictive assessments and apply this to clinical trial designs for which pertinent preexisting data are available. In an application related to sequential monitoring, we introduce an adaptive monitoring prior for efficacy evaluation that dynamically weighs skeptical and enthusiastic prior components based on the degree to which the trial data are consistent with a specified perspective. In an application related to hybrid designs, we introduce a novel extension of the power prior where the discounting weights are computed separately for each external control based on compatibility with the randomized control data.
Event Site Link
https://uthealth.webex.com/uthealth/j.php?MTID=mb5159c73ed710b61bd2a64009de42d0a
Additional Information
When: 09/05/2023 12 noon
Where: RAS-102B, See event website for WebEx. Password: xnFXMhnm638
Presenter: Dr. Dr. Evan Kwiatkowski
Abstract:
Bayesian methods provide a natural approach for incorporating external data into the analysis of a clinical trial through an informative prior distribution. Prior-data conflict occurs when the prior primarily favors regions of the parameter space that are far from that supported by the likelihood for the trial data. The predictive distribution based on the trial data can be used to assess the compatibility or relevance of the external data (i.e., Box’s p-value). We determine the compatibility of a prior and the trial data using posterior-predictive assessments and apply this to clinical trial designs for which pertinent preexisting data are available. In an application related to sequential monitoring, we introduce an adaptive monitoring prior for efficacy evaluation that dynamically weighs skeptical and enthusiastic prior components based on the degree to which the trial data are consistent with a specified perspective. In an application related to hybrid designs, we introduce a novel extension of the power prior where the discounting weights are computed separately for each external control based on compatibility with the randomized control data.
\n\nhttps://uthealth.webex.com/uthealth/j.php?MTID=mb5159c73ed710b61bd2a64009de42d0a", "startDate":"2023-9-5", "endDate":"2023-9-5", "startTime":"12:00", "endTime":"13:00", "location":"RAS 102B", "label":"Add to Calendar", "options":[ "Apple", "Google", "iCal", "Microsoft365", "MicrosoftTeams", "Yahoo" ], "timeZone":"America/Chicago", "trigger":"click", "inline":true, "listStyle":"modal", "iCalFileName":"Reminder-Event" }