NAVER LABS Europe seminars are open to the public. This seminar is virtual and requires registration.
Date: 15th April 2021, 3:00 pm (GMT +01.00)
Anthropology as UX research: promoting human-centered evaluation of a deep learning system for the detection of diabetic retinopathy
Speaker: Elizabeth Baylor is an applied anthropologist and Lead User Experience Researcher at Google’s Engineering Compliance. Previously at Google, she served as the Lead for Google’s new digital payment app for India before moving to Google Health to work with AI. She received her PhD in Applied Anthropology from the University of South Florida and has previous experience at Microsoft, the University of Alabama, and Indiana University of Pennsylvania.Before UX, Elizabeth was a nutritional anthropologist studying foodways in Southeast Asia. She has been trained and funded by the National Science Foundation, and is a Fulbright Scholar, published author, and a regular contributor to the American Anthropological Association, Society for Applied Anthropology, EPIC, and CHI. Elizabeth specializes in adapting and applying classic social science methods to product design and development.
Abstract: Elizabeth Baylor will discuss how UX Research (UXR) enhances product development, drawing on her own career path and a recent Google Health initiative to improve diabetic retinopathy screenings in Thailand with the use of artificial intelligence (AI).
While machine learning (ML) and AI have significant potential in health contexts, it is rare to see deep learning algorithms demonstrating real world improvements in clinician workflows and patient outcomes. Why? In short, we move what we measure. To date, there has been a strong focus on prospective evaluations of model accuracy but less attention to the human processes and needs that our tools ultimately aim to serve. This talk outlines a more human-centered approach.
Based on interviews and observations, you will see how current eye-screening workflows, user expectations for an AI-assisted screening process, and post-deployment experiences impact both model performance and ultimately the provider and patient experience. Join us to explore the value of conducting human-centered research alongside prospective evaluations of model accuracy.