Egenis seminar: "A feel for the data: from sensory to algorithmic prediction in critical care"

As machine learning algorithms move into the everyday routines of acute medicine, critical care is emerging as a key site where sensory, embodied and computational ways of knowing meet. Bedside work in the ICU has long relied on clinicians’ cultivated feel for patients’ trajectories: attunements to breathing patterns, skin tone, trace sounds, and subtle shifts in affect or physiology. Today, these forms of expertise encounter growing expectations that large streams of monitor data, cleaned, curated, modelled and repurposed, can be transformed into predictive tools capable of guiding intervention. Commercial systems based on de-identified vital-sign datasets are already in circulation, and others are in development. Yet the introduction of such prediction models raises questions about how these so-called ‘digital solutions’ sit alongside the socio-material relations that make intensive care possible, and how clinicians might integrate (or resist) them in decision-making.

Drawing on an ongoing ethnography that moves between the clinical world of the ICU and the academic world of medical data science, this talk explores how prediction is crafted and contested in practice. Following healthcare practitioners, data scientists, and ‘clinicians who code’, I examine the shifting practices of care as they become newly mediated by data infrastructures, analytic pipelines, and the promise of healthcare AI. Drawing on the STS scholarship on care, I analyse how different logics - from efficiency to intuition – are negotiated in the data-patient assemblage. In doing so, I argue that the future of algorithmic medicine cannot be understood without attending to the embodied, affective and relational labour that continues to underpin clinical judgement, and to the forms of care that data work itself demands.

Venue Byrne House, Streatham Campus, Exeter

Online: via Zoom

Free to attend. Register here