Hidden in Plain Sight: An Update to the Reading Race Project

November 10, 2022

Judy Wawira Gichoya, MD, MS | Department of Radiology and Imaging Services, Emory University School of Medicine

There is an old saying, “a picture is worth more than a thousand words.” We increasingly observe this phenomenon where AI models are shown to outperform radiologists to detect features like gender, race, social determinants of health, and health care costs. This ability has potential to be harnessed to mitigate disparities in prediction systems, such as breast cancer risk prediction. However, our understanding is limited due to the black box nature of AI models.

In this lecture, Dr. Gichoya will:

• Review several use cases where this ability is being harnessed
• Review some possible explanations for this ability and then 
• Recommend future possible areas of research that participants can explore