GWU, Office of CEHP – Watergate Office Building, 1st floor conf. room
2600 Virginia Ave. NW, Washington, D.C. 20052
Muhammad Aurangzeb Ahmad
University of Washington Tacoma / KenSci
As AI and Machine Learning are being increasingly integrated into healthcare, challenges in creating responsible AI systems that are interpretable, fair, transparent, unbiased, robust and reliable are coming under tighter scrutiny. This talk will focus on one important aspect of such systems – Explainability – and will explore what constitutes explainable AI in healthcare, and what are the nuances, challenges, and requirements for its design. Drawing on insights from both academia and industry, the talk will delve into why explainability in healthcare is different from other domains and how deployed systems may have counterintuitive implications: e.g., explainability leading to distrust, explainable systems leading to more opaque models, etc. We will discuss the application of explainability techniques and the practical challenges in creating effective explainable AI models in healthcare. Finally, open problems and research directions for AI in the healthcare community will be described.