Digital Phenotyping and the Cumulative Burden of Self-Management of Bipolar Disorder

Abstract

Living with a chronic condition involves ongoing work: tracking symptoms, managing medication, coordinating care, and maintaining routines to prevent recurrence of acute episodes. In psychiatric care, this workload is equally complex yet rarely examined in HCI and design research. Digital phenotyping, a nascent AI/ML-based approach to monitoring behaviour, passively collects sensor and device data and is promoted as an unobtrusive and scalable tool. Yet, this promise of passivity obscures the skilled labour that takes place-from data sense-making to fitting new tools into existing care practices. In this position paper, situated in the context of outpatient bipolar disorder treatment in the Netherlands, we seek to problematise the current approach and apply Burden of Treatment Theory to unpack this hidden workload. We argue that digital phenotyping design lacks sensitivity to the care ecologies of self-management it intervenes into and call for HCI researchers to further engage with this problem space.