Fatigue causes decrements in vigilant attention and reaction time and is a major safety hazard in the trucking industry. There is a need to quantify the relationship between driver fatigue and safety in terms of operationally relevant measures. Hard-braking events are a suitable measure for this purpose as they are relatively easily observed and are correlated with collisions and near-crashes. We developed an analytic approach that predicts driver fatigue based on a biomathematical model and then estimates hard-braking events as a function of predicted fatigue, controlling for time of day to account for systematic variations in exposure (traffic density). The analysis used de-identified data from a previously published, naturalistic field study of 106 U.S. commercial motor vehicle (CMV) drivers. Data analyzed included drivers’ official duty logs, sleep patterns measured around the clock using wrist actigraphy, and continuous recording of vehicle data to capture hard-braking events. The curve relating predicted fatigue to hard-braking events showed that the frequency of hard-braking events increased as predicted fatigue levels worsened. For each increment on the fatigue scale, the frequency of hard-braking events increased by 7.8%. The results provide proof of concept for a novel approach that predicts fatigue based on drivers’ sleep patterns and estimates driving performance in terms of an operational metric related to safety. The approach can be translated to practice by CMV operators to achieve a fatigue risk profile specific to their own settings, in order to support data-driven decisions about fatigue countermeasures that cost-effectively deliver quantifiable operational benefits. (Author/publisher)
Abstract