Regulatory Science Challenges for AI Imaging Devices
Industrial Problems Seminar
Michelle Mastrianni
FDA
Abstract
Most people know that the Food and Drug Administration regulates food and drugs, as per its name—but medical devices also present ever-increasing regulatory challenges for FDA scientists. The field of medical imaging has seen transformative advancements in recent years with the use of artificial intelligence (AI) algorithms to improve the efficacy and efficiency of disease screening. One proposed option to reduce radiologist workload is to automatically "rule-out" AI-negative patients as non-diseased, therefore bypassing the need for radiologist review. A future rule-out device necessitates a study of evaluation metrics different from those used for existing devices, which are currently intended mainly as a diagnostic aid or a tool to triage image review priority. In this talk, I will discuss my work to develop statistical methodology for rule-out device evaluation in an effort to support regulatory science in my division at the FDA. I will also briefly describe some other interesting projects within the division. More generally, I’ll overview my transition in the past year from a math PhD to a research fellowship within the world of regulatory science and medical imaging.