The Right Prediction at the Right Time: Designing AI for Clinical Workflows

Title: "The Right Prediction at the Right Time: Designing AI for Clinical Workflows"
Abstract: More than two-thirds of U.S. hospitals now use predictive models, yet adoption has not consistently translated into better outcomes, raising a simple question: what are these models for, and how do they change care? This talk argues that real-world impact hinges on aligning model design and evaluation with intended use. In the first part ("the what"), I discuss how focusing on clinical goals and decision support can reorient development toward downstream objectives (e.g., reducing sepsis mortality), rather than predictive metrics alone. In the second part ("the how"), I show how model performance is often shaped by human interaction and clinical workflow, and why development and evaluation must account for how clinicians interpret and act on predictions to understand when models help.

Mini Bio: Jenna Wiens is an Associate Professor of Computer Science and Engineering (CSE), the Associate Director of the Artificial Intelligence (AI) Lab , and the co-Director of AI & Digital Health Innovation at the University of Michigan in Ann Arbor. Her primary research interests lie at the intersection of machine learning (ML), artificial intelligence (AI), and healthcare. Dr. Wiens takes a use-inspired approach to research, studying real-world problems through a technical lens. In close collaboration with domain experts and clinicians, her work has enabled the successful integration of ML models developed by my research group into clinical workflows, with a positive impact on patient care. Dr. Wiens received her PhD in 2014 from MIT. In 2015 she was named Forbes 30 under 30 in Science and Healthcare; she received an NSF CAREER Award in 2016; in 2017 she was named to the MIT Tech Review's list of 35 Innovators Under 35; she received a Sloan Fellowship in Computer Science in 2020; and in 2024 she received the Carl Friedrich von Siemens Humboldt Research Award in recognition of her career accomplishments.

Zoom Meeting: https://duke.zoom.us/j/98429803765?pwd=A66McPYaMdwEPbdYEEvX0jbiingvgu.1
Meeting ID: 984 2980 3765
Passcode: 103794