Uncovering Positionality in Data and Models for Healthcare

June 3, 2019, Berkman Klein Center, Cambridge, MA

We examine classification systems and the forces that shape them with a focus on AI healthcare applications. This workshop aims to shed light on perspectives that shape classification systems to uncover what we call positionality -- ie, how identity and/or outlook can influence the ways in which data is labelled and rendered. To demonstrate this phenomenon, we consider several examples in the healthcare industry, where AI applications continue to proliferate. For example, the International Classification of Disease (ICD) has separate codes for legal and illegal abortion--a deliberate choice in data rendering that we are able to unpackage when examined through a historical lens. The workshop will examine how positionality materializes in other settings of healthcare, like medical imaging and precision medicine.

[Agenda] [Introduction to Positionality] [W61.62 Struck by Duck]

Case studies: [Ontologies and Patient Care] [Medical Imaging] [Precision Medicine]