Answers are grounded in this paper and cite their sources.
The Personalised Patient Preference Predictor (P4) has been proposed as an ethically better alternative to the population-level Patient Preference Predictor (PPP). When patients cannot take part in decisions about their own care at all (e.g., those in coma or with severe brain injury and no applicable advance directive), or cannot communicate their will effectively, the P4 customises a generative model with their digital texts. The aim is to create what has been called a ‘digital psychological twin’, intended to predict which treatment a formerly competent self would likely choose. Although its proponents intend the P4 to support surrogate decision-making rather than to replace it, in practice its prediction risks being treated as a reconstructed advance directive that respects a ‘former self’. We criticise this treatment of its output by first mentioning the issues of advance directives that P4 inherits, and second, illustrating important differences between P4’s predictions and actual advance directives that undermine their bindingness. Moreover, drawing on relational autonomy and the ethics of care perspectives, we argue that the main problem with P4 is not just accuracy but its individualistic view of the self and decision-making, which the tool cannot transcend. Finally, we frame this discussion within the supported decision-making paradigm of the Convention on the Rights of Persons with Disabilities (CRPD). We conclude that P4-like tools should be viewed as, at most, a revisable interpretive aid within a dialogical supportive process — and not a replacement for the person’s current will.
No takes yet. Share an insight, caveat, or question.
Hosseiny et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: