Fred Nugen of University of California - Berkley

Generative AI includes tools like ChatGPT and ThisPersonDoesNotExist.com.  These tools have many helpful applications, like using Denoising Diffusion Probabilistic Models (DDPM) to reduce radiation dose for breast cancer imaging.  In addition, they can generate synthetic data to help reduce bias in healthcare datasets and fill gaps in real-world measurements.  Helping summarize medical visits and investigate a patient's electronic health record is a newer usage of GenAI models being implemented in many hospitals around the world.  While some people tout these innovations as efficiency tools, there are unresolved ethical implications in their use that most users and patients are unaware of, unable to resolve, or blithely ignore.  We will explore the above uses of GenAI tools in the medical space, and close with an open-ended discussion of some of their ethical implications both inside and outside healthcare.

Biography:

Fred Nugen, PhD, is a Senior Collaborator with the Global Burden of Disease (GBD) Study.  Prior to this, he was Assistant Professor of Radiology at Mayo Clinic in the Radiology AI Lab, specializing in AI models for diagnosis, prognosis, and screening of rare cancers.  Since 2020, he has taught Data Science at UC Berkeley, leading the project-based Capstone course.  His students have designed cutting-edge products, including a collaboration with the CDC for an AI model to help investigate breakouts of Legionnaire's disease, and a collaboration with Emergency Department at UC San Francisco hospital to detect unmilled procedures.  Today, he is Chief AI Officer at Seirrowon, a startup specializing in centimeter-level GPS positioning systems for precision agriculture.