To Care and to Code: Transformations in Critical Care in the Time of AI

Abstract

‘Data saves lives’ is an oft-repeated slogan in the world of data-driven healthcare. In the field of critical care, it has also gained traction, refracted in one project’s strapline, “Using data better saves critically ill patients' lives”. The use of routine data to develop algorithms for risk prediction in this population is now well underway, with one tool already on the market and others in development, based on de-identified vital signs data from critically ill patients’ bedside monitors. As large-scale data analytics make their way into routine care, it is salient to ask how the embodied and sensory dimensions of practitioners’ work, and the total system of socio-material relations which characterises the intensive care unit (ICU), intersect with the embodied and sensory dimensions of data science work. How do clinicians in critical care decide how to treat their patients in the era of algorithmic medicine? Is the craft work of critical care changing with increasing digitisation and recourse to clinical decision support systems? What role do ‘clinicians who code’ play in this transformation? In this paper, I start to answer these questions with insights from an ongoing ethnography spanning the clinical spaces of the ICU and the academic spaces of medical informatics. Informed by theoretical work on care in STS, I examine what care means in the data-patient assemblage, which logics drive the practices that healthcare now adopts and how these challenge the repertoires we have for thinking and talking about care.