Traditionally, clinical events are reviewed by a group of adjudicators to determine whether they meet the definition outlined in the study protocol. Adjudication of events is important in securing FDA approval for new drugs and devices. The DCRI’s CEC group has about 50 faculty reviewers who dedicate time weekly to event adjudication. However, the work has expanded exponentially—CEC now performs about 30,000 adjudications per year, in comparison with 5,000 per year about five years ago.
Because of this growth, it is essential to enhance CEC’s processes, said Matt Wilson, RN, (pictured right) operations director of CEC. The group is working closely with machine learning experts at the DCRI and Duke to develop machine learning algorithms that are able to adjudicate the more straightforward cases.
The group plans to test the capabilities of these algorithms in a prospective study that will compare a pure machine learning approach to a hybrid approach that pairs the algorithms with human adjudication. Renato Lopes, MD, PhD, (pictured left) one of CEC’s faculty directors, said he believes the hybrid approach will be most accurate, as human adjudication will likely be necessary for complex cases.
As CEC looks forward to implementing these algorithms in real-time adjudication, it is also diversifying the therapeutic areas it serves. Three years ago, its portfolio was 70 percent cardiovascular. However, with this year’s additions of endocrinology, nephrology, neurology, infectious diseases, and pediatrics, it is now 49 percent cardiovascular—which suggests that other clinical areas are realizing the benefit of using CEC before seeking FDA approval.
This article originally appeared in the DCRI's 2018-2019 Annual Report. View more articles from this publication.
Renato Lopes, MD, PhD, MHS, one of the faculty co-leads for DCRI’s Clinical Events Classification group, discusses the future of clinical event adjudication with Matt Wilson, director of operations for the group.