Validation of FAST Model Sleep Estimates with Actigraph Measured Sleep in Locomotive Engineers

Final report; Dec. 2005-Oct. 2011.Raslear, ThomasPDFTech ReportDOT/FRA/ORD-12/05DFRA.080088Human characteristicsHuman factorsPhysical conditionSleepFatigue (Physiological condition)DrowsinessLocomotive engineersUnited StatesUnited States. Federal Railroad AdministrationGertler, JudithHursh, StevenFanzone, JosephRaslear, ThomasQinetiQ North America, Inc.Institutes for Behavior ResourcesUnited States. Federal Railroad AdministrationUnited States. Federal Railroad Administration. Office of Railroad Policy and DevelopmentNTL-RAIL TRANSPORTATION-Rail SafetyNTL-RAIL TRANSPORTATION-Rail Planning and PolicyNTL-SAFETY AND SECURITY-Rail SafetyNTL-SAFETY AND SECURITY-Human FactorsUS Transportation CollectionThis report presents the results of a study to validate the AutoSleep sleep prediction algorithm, which is a component of the Fatigue Avoidance Scheduling Tool (FAST). Researchers collected work and sleep data from 41 locomotive engineers by using actigraphy and daily log books and compared these data with AutoSleep predictions developed according to the log-book\ue2\u20ac\u201crecorded work periods. Comparison of the actigraphy data with model predictions on a minute-by-minute basis found an overall agreement between the two 87 percent of the time. Application of Signal Detection Theory to the data indicates that AutoSleep is biased toward underestimating daily sleep. These findings validate the sleep prediction algorithm of FAST and validate its utility for assessing fatigue risk created by typical railroad schedules