Ethical Dimensions of Fall Detection Artificial Intelligence: An Ethnographic Study in Geriatric Care

Geriatric patients face a chronic risk of falls due to age-related physical and cognitive decline. AI-driven clinical decision support systems have been proposed to mitigate these risks, with image-based AI gaining attention for its real-time capabilities and associated ethical concerns, such as privacy and prioritization. Geriatric care staff using such technology must navigate these ethical challenges in their practice. However, our understanding of the ethical dimensions of fall detection AI in geriatric care remains limited, limiting ethical implementation. <br/><br/>We present an ethnographic study at a Danish geriatric hospital ward, which uses image-based AI to prevent falls among geriatric patients. We observed the ward’s general practices of geriatric care, as well as the AI system’s integration into and change of these care practices. We analyzed our data using intra-normativity from Jeanette Pols’ empirical ethics, i.e. explaining how the AI system is used to enact good geriatric care for chronic fall risks. <br/><br/>Our study demonstrates how geriatric care staff ethically engage with the use of image-based AI detection to mitigate the chronic risk of falls among their patients. The AI system is consistently adjusted and interpreted based on clinical judgement of patients and the varied rhythms of falls. However, the AI system also functions both as a stress factor and a tool to mitigate other stress factors. Our results emphasize how ethical use of fall detection AI hinges on continued adjustments serving both patients and staff, paving the way for ethical implementation of AI systems mitigating geriatric patients’ chronic fall risks.