Predictions from the three-process model of alertness

Abstract

This paper summarizes a computer model for predicting alertness/performance in daily life. The model uses the timing of work hours (and/or sleep hours) as input and uses both a circadian and a homeostatic component (amount of prior wake and amount of prior sleep) that are summed to yield predicted subjective alertness (on a scale between 1 and 21). The model also includes prediction of psychomotor performance on various tasks, an identification of levels at which the risk of performance/alertness impairment start, as well as a prediction of sleep latency and time of awakening from sleep. The model is currently being used to evaluate work/rest schedules for navy, airline and railway applications. It is also used for teaching sleep/wake regulation and for generating research hypotheses.