Human vs. Machine: Artificial Intelligence-Based Prognostication in End-of-Life and Palliative Care
Saturday, October 14, 2023
9:00 AM – 10:15 AM ET
Location: Waterview CD (Lobby Level)
Artificial Intelligence (AI) can predict patient mortality with accuracy that far outperforms clinicians. Health systems are already using AI with the promise of improved access to end-of-life and palliative care (EOLPC). However, ethical concerns exist regarding choice, unintended harms, and fairness. To inform healthcare organization implementation, we conducted semi-structured interviews with 30 palliative care team members and 15 patients and caregivers from 4 academic medical centers. Interviews were transcribed and analyzed using grounded theory. Interviewees were optimistic that AI-based prognostication would help patients who could benefit from palliative care access it sooner, and that a more accurate sense of prognosis would aid in planning and decision-making. A central theme from our findings, however, was the concern that AI-based prognostication could lead to less humane care. Patients feared that prognostic scores could become stigmatizing “labels,” ultimately limiting their access to other types of care. Care team members worried about over-reliance on AI and an overemphasis on prognosis in decision-making. Often, interviewees expressed distrust that algorithms could capture spirituality or the “human factor.” Participants from racial/ethnic minority groups appeared more aware that algorithms built on historical data could be biased, exacerbating healthcare disparities. Interviewees believed that AI-based prognostication could support high quality EOLPC, but only if it were “just a tool.” For AI-based prognostication to be used ethically, health systems should implement it in ways that preserve a holistic approach to caring in EOLPC. Additional efforts may be needed to educate clinicians about the use or application of AI, especially regarding bias.
Mika Hamer, PhD, MPH – Center for Bioethics and Humanities – University of Colorado; Kathryn Huber, MD, MS – School of Medicine – University of Colorado; Matthew DeCamp, MD, PhD – Center for Bioethics and Humanities – University of Colorado