Preference Prediction at the End-of-Life: Difficult for Humans and Algorithms
Friday, October 13, 2023
5:00 PM – 6:15 PM ET
Location: Kent A-C (Fourth Floor)
Decision-making for incapacitated patients is messy. Some patients may not have an Advance Directive or a Surrogate Decision-Maker (SDM), or have not had thorough discussions with their SDM. Additionally, SDMs or clinicians may worry about the applicability of previously expressed wishes to new decisions that are circumstance- and time-specific. To resolve this matter, some suggest decision-support for incapacitated patients through the use of algorithms with the main argument that algorithms could be more accurate than surrogates. However, fundamental issues remain unresolved regarding algorithmic decision-support for patient preferences, such as the suitability of data sources for preference prediction. In this working paper, we outline how preference-prediction algorithms are ethically fraught when compared to other clinical uses of algorithms and especially for goals of care decision-making. Subsequently, we identify problems with using proposed algorithms for value-laden decisions and provide suggestions for when, if at all, algorithms should be consulted. These issues, which include problems with accuracy, model verification, and algorithmic persuasion, left unaddressed, may exacerbate a prediction-explanation fallacy where unintentionally biased algorithm(s) misrepresent decision-making for incapacitated patients with or without surrogate decision-makers. The supposed superiority of algorithms over SDMs or clinicians deciding in a patient’s best interests is far from established. This leads to serious decisions with irreversible consequences being made for patients that could be in conflict with their values and preferences. Although well-intentioned, these algorithms potentially undermine the autonomy of incapacitated patients, who are at their most vulnerable, causing harm.
Emma Tumility – Department of Bioethics & Health Humanities – University of Texas Medical Branch