Amy Moran Thomas, Associate Professor of Anthropology, MIT

Valencia Joyner Koomson, Associate Professor of Electrical and Computer Engineering, Tufts University & Associate Professor of Computer Science, Tisch College


Following Racially Biased Sensors, How Should We Approach the Algorithms and Models Trained on Distorted Data?

Pulse oximeters have recently been flagged by the FDA for encoding racial bias, since they disproportionately produce clinical errors for nonwhite patients due to how this color-sensing device was designed and calibrated to assess blood oxygen levels using mostly white test groups. Yet many algorithms and machine learning tools across healthcare systems continue to rely on direct input from these sensors and be trained on their data. This presentation highlights a research area that has not yet received adequate public attention: the broader implications for equity that this biased sensor data holds for the next generation of models in computing and healthcare. How are clinical algorithms and medical devices trained on such distorted datasets incorporating them today? Have any machine learning models managed to avoid reproducing racial biases after being fed disproportionately incorrect data? If so, what can be learned from how they managed to identify and address these patterned errors?  And if not, what are the implications for outcomes and care ahead? This presentation offers a critical stock-taking of where this issue currently stands.  It covers a social history of the problem, and describes research areas that would benefit from future interdisciplinary collaborations.


Amy Moran-Thomas is Associate Professor of Anthropology at MIT, interested in environmental aspects of human health and ethnographic approaches to science, technology, and medicine. She is author of the book Traveling with Sugar: Chronicles of a Global Epidemic(2019), an anthropological account of unequal diabetes technologies and the lives they shape in global perspective. Her articles drawing public attention to longstanding racial biases encoded in color-sensing medical devices helped to catalyze clinical reexaminations of the pulse oximeter, including a recent FDA hearing that led to new safety advisories and ongoing revisions of the device ISO. She is interested in how social perspectives on design can contribute to producing more equitable technologies.

Valencia Joyner Koomson is an Associate Professor at Tufts University in the Department of Electrical and Computer Engineering with a joint appointment in the Department of Computer Science, and the Tisch College of Civic Life. She completed the B.S. and M.Eng. degrees in electrical engineering and computer science at the Massachusetts Institute of Technology in 1998 and 1999, respectively. She was awarded the George C. Marshall scholarship in 1999 to pursue post-graduate studies at the University of Cambridge.  She received the M.Phil. and Ph.D. degrees in electrical engineering in 2000 and 2003, respectively, from the University of Cambridge.   Prof. Koomson’s research lies at the intersection of biology, medicine, and electrical engineering.  Her interests are in micro-/nano-scale electronic circuits and systems for wearable and point-of-care biomedical devices, health-IoT systems, health informatics, and advanced microfluidic systems to probe intercellular communication. In 2021 she received the Dr. Martin Luther King Jr. Visiting Professor fellowship at MIT in the Department of Electrical Engineering and Computer Science.