Whether or not humans are ready, AI predictions are quietly set to become part of end-of-life medical decisions.
getty
AI predictions could soon play a crucial role in end-of-life care decisions, potentially surpassing the accuracy of human predictions. These predictions raise important questions about the intersection of AI, human intelligence, and values.
Recent studies have explored the use of AI algorithms in predicting patient preferences for end-of-life interventions. One study by European researchers found that an AI “patient preference predictor” accurately predicted end-of-life preferences 71% of the time, outperforming human surrogates and couples predicting their partner’s wishes.
AI Prediction Prods People
Another study involved sharing AI-generated predictions of patients’ death risk with clinicians to encourage end-of-life planning. This initiative led to increased use of palliative care and hospice, resulting in a lower-than-expected death rate within 30 days of discharge.
Both sets of researchers emphasized the importance of human involvement in the decision-making process, with the European team cautioning against relying solely on the algorithm. They suggested a collaborative approach where AI and humans work together to make informed decisions.
The St. Louis group focused on training clinicians to initiate discussions about end-of-life care and goals of care. By involving clinicians in the process, they aimed to ensure that AI predictions were used as decision aids rather than replacing human judgment.
The Human Role
While AI shows promise in improving end-of-life care decisions, concerns have been raised about overestimating its capabilities. A commentary on the European research findings highlighted the importance of human surrogates and doctors in guiding patients through complex decisions.
Researchers are exploring the development of a moral AI surrogate that not only predicts outcomes accurately but also honors patients’ values and cultural worldview. This AI surrogate would serve as a decision aid, not a decision maker, with transparent reasoning.
Real-World Worries
As AI continues to be integrated into healthcare, it is essential to consider the limitations of AI in end-of-life decisions. Preferences can change over time, and the role of human surrogates remains crucial in ensuring that patients’ values are respected.
A Moral AI Surrogate
Efforts are underway to develop a moral AI surrogate that combines predictive accuracy with ethical considerations. By involving patients in advance care planning and decision-making, individuals can ensure that their values and preferences are respected.
