AI Surrogates
Exploring AI-driven mortality prediction and AI surrogates for advance directives to enhance ethical, patient-centered decision-making at the end of life.
This reasearch program focused on mortality prediction using machine learning (Ahmad et al., 2018) and the use of AI surrogates for advance directives in the context of clinical decision support systems. We also explore ethical and responsible AI frameworks to navigate end-of-life decision-making. Mortality prediction models, leveraging patient vitals, medical histories, and treatment responses, can offer clinicians probabilistic insights into a patient’s survival trajectory, aiding in timely and personalized care planning. However, the complexity of such predictions extends beyond statistical accuracy to the deeply human aspects of autonomy, dignity, and informed choice. AI surrogates—digital proxies trained on patient values, past decisions, and expressed wishes—could serve as decision-making aids for advance directives, ensuring that care aligns with a patient’s preferences even when they can no longer communicate them. These AI systems could help mitigate uncertainty, guide families and physicians through difficult choices, and ultimately uphold patient agency and dignity in the face of life-limiting illness.
