Research

My dissertation and book project, Data Values: Moral Entrepreneurship in Digital Health, examines ongoing efforts to build an ethics of AI. 

Healthcare researchers increasingly turn to techniques of AI and machine learning to guide patient care, but these tools have provoked heated debates about privacy, AI safety, and algorithmic bias. Formal regulation lags behind technological advances, giving researchers wide latitude to establish norms and values of their field. In this project, I study how academic researchers create norms and rules for digital mental health and how these new ideas change healthcare for clinicians and patients alike. 

To understand this moral self-regulation, I completed three years of ethnographic fieldwork with academic teams developing technology for mental healthcare. I argue researchers’ moral rulemaking is a form of entrepreneurship. Moral entrepreneurship shapes the field of digital mental health by setting what it focuses on, what it neglects, and whom it serves. As this project analyzes this process, it helps explain how moral ideas are adjudicated amidst uncertainty and it uncovers new mechanisms for health disparities and social inequality in our digital future. 

This project is funded by several competitive national fellowships, including the ASA/NSF Doctoral Dissertation Research Improvement Grant and the Institute for Citizens and Scholars’ Charlotte W. Newcombe Dissertation Grant. One article from this project is published in Socius.

A second project theorizes how algorithmic systems like electronic health records (EHRs) yield real-world consequences, using the case of decision-making processes for prostate cancer care. 

For this project, I completed 20 months of ethnographic research and interviews in a urologic cancer clinic. My major paper from this project addresses a dilemma in the sociology of algorithms literature: Algorithmic problems like bias and inefficiency are typically attributed to the fact that algorithms are opaque, yet opacity does not explain why there are so many ways algorithmic decision-making can go awry, nor why an algorithm can work well under some circumstances and not others. My paper develops a theoretical model for how algorithmic effects are mediated by social and organizational conditions. I illustrate my argument with the case of medical decision-making with the EHR, a complex algorithmic system. 

This paper has been awarded two best paper prizes, from the Science, Knowledge, and Technology section and the Economic Sociology section of the American Sociological Association. It is currently under review.

My third project studies the role of information technology in efforts to standardize care for health equity. 

This project analyzes how clinicians use electronic health records to document social determinants of health (SDOH). Though professional groups and US federal agencies have issued calls to standardize SDOH data, little is known about how clinicians currently document and use SDOH in daily practice. To investigate this question, I completed ethnographic observation, interviews, and qualitative coding of EHRs to triangulate how clinicians talk and write about patients’ SDOH. I find that even clinicians who prioritize attention to patients’ SDOH do not typically document them. Instead, the paper identifies three “local standards” clinicians enact to integrate SDOH into their care that may be invisible to or even threatened by formal documentation standards. 

A paper from this study is published in Social Science & Medicine. It was awarded the Donald W. Light Award from the Medical Sociology section of the American Sociological Association.