HISTORICALLY, BIOMATHEMATICAL PERFORMANCE PREDICTION MODELS HAVE FOCUSED ON PREDICT-

This modeling strategy necessarily de-emphasizes individual differences in resilience and vulnerability to sleep loss and con-tradicts recent findings that indicate significant and systematic differences among individuals.4-7 For example, on a psychomo-tor vigilance test (PVT) scale,8 restricted-sleep laboratory stud-ies show that interindividual differences account for nearly 70% of the total variance in performance among a group of individu-als, and that the differences are trait-like, stable, and innate to an individual.5 Hence, even if a group-average model were ca-pable of accurately predicting mean-group performance, such a model would be of limited benefit without knowing how this translates into predictions at an individual level.9 Recently, two methods have been proposed for developing in-dividual-specific performance prediction models for individuals subjected to total sleep loss with uncertain initial states (i.e., ini