The number of people requiring palliative care in the UK is projected to rise significantly, creating an urgent need for earlier and more systematic recognition of those approaching the end of life. Current clinical markers and tools for predicting prognosis are limited in their accuracy, and prone to human and systemic biases. Artificial intelligence (AI) offers potential to improve prediction of deterioration and dying, and early studies suggest that it may support timely interventions and advance care planning. However, integration of AI must prioritise data integrity, accountability and minimising the amplification of existing inequities. Crucially, recognising dying remains a fundamentally human task with ethical, relational and existential dimensions that AI cannot replicate. Successful implementation will depend on thoughtful human–AI collaboration that strengthens clinical insight without compromising the compassionate, person-centred approach that is central to palliative care.
Lester et al. (2026) studied this question.