DISCLOSURE STATEMENT
No potential conflict of interest was reported by the author(s).
Notes
1 This is not to say that the algorithm cannot be trained based on previous decisions—it can—but rather that those training data would not produce the right predictions, that is, predictions judged by their congruence with what the patient would have decided had she been offered, and made, a choice.
2 Consider, for example, work in economics on the notion of projection bias (Loewenstein, O’Donoghue, and Rabin Citation2003); work in psychology on affective forecasting (Gilbert and Wilson Citation2009); work in disability studies on perceptions quality of life (Albrecht and Devlieger Citation1999).