Bias and Ethics in AI and Healthcare

Time: 
10:45 AM to 11:25 AM
Room: 
Rutter Center Robertson Auditorium 3
Track: 
Data & Analytics
Description: 

This session will examine the potential sources of bias when building artificial intelligence (AI) systems in health care and lay out a framework for proactively identifying and preventing or mitigating bias . Topics will include identifying biases due to missing data and groups not identified by machine learning algorithms, sample size and underestimation, and misclassification and measurement error. We will discuss both explicit and implicit bias and examples of how bias might contribute to socioeconomic disparities in health care along with recommended approaches to managing these ethical and policy issues, with current understanding and thinking into the future of AI and machine learning. This presentation will leverage several examples of UCSF projects to illustrate these critical concepts.

Slides: https://ucsf.box.com/s/17n30io0qwu6opfgqng44enpds5dop1x (MyAccess login required)

Presenter(s): 
Jinoos Yazdany
Gabriela Schmajuk
Milena Gianfrancesco
Sara Murray
Session Type: 
Skill Level: 
Beginner
Previous Knowledge: 

This session is designed to be accessible to both a general audience interested in the topic of bias in AI and health care and to the more seasoned researcher who is using AI methods in health care.

Speaker Experience: 

Jinoos Yazdany MD MPH, Professor of Medicine and Chief, ZSFG Division of Rheumatology, Director Quality and Informatics Lab
Sara Murray MD MAS, Director of Medical Informatics UCSF
Milena Gianfrancesco, PhD, Adjunct Assistant Professor of Medicine, UCSF
Gabriela Schmajuk, MD MSc, Associate Professor of Medicine, Interim Chief of Rheumatology SFVA