Algorithmic assemblages of care: imaginaries, epistemologies and repair work

Publikation: Bidrag til tidsskriftTidsskriftartikelForskningfagfællebedømt

Dokumenter

In the past decade, thefigure of the algorithm has emerged as a matter of concernin discussions about the current state of the healthcare sector and what it maybecome. While analytical focus has mainly centred on‘algorithmic entities’, thepaper argues that we have to move our analytical focus towards‘algorithmicassemblages’, if we are to understand how advanced algorithms will affect healthcare. Departing from thisfigure, the paper explores how an algorithmic system,designed to‘take on’the role of a physiotherapist in physical rehabilitationprogrammes in Denmark, was designed and made to work in practice. On thebasis of ethnographicfieldwork, it is demonstrated that the algorithmic system is afragile accomplishment and outcome of negotiations between the imaginariesembedded in its design and the ongoing adjustments of IT workers, patients andprofessionals. Drawing on recent work on the fragility and incompleteness ofalgorithms, it is suggested that the algorithmic system needs to be creatively‘repaired’to build and maintain enabling connections between bodies in-motionand professionals in arrangements of care. The paper concludes by addressingaccountability for the workings of algorithmic systems in medical practice,suggesting that such questions must also be discussed in relation to encountersbetween algorithmic imaginaries, health professionals and patients, and the variousforms of‘repair work’needed to enable algorithmic systems to work in practice.Such acts of accountability cannot be understood within an ethics of transparency,but are better thought of as an ethics of‘response-ability’, given the need tointervene and engage with the open-ended outcomes of algorithmic systems.
OriginalsprogEngelsk
TidsskriftSociology of Health and Illness
Vol/bind41
Udgave nummerSupplement 1
Sider (fra-til)176-192
ISSN0141-9889
DOI
StatusUdgivet - 2019

Antal downloads er baseret på statistik fra Google Scholar og www.ku.dk


Ingen data tilgængelig

ID: 229268699