Impact of highway work zones on drivers' cognitive workload, visual activity, and risky driving behaviour

Road infrastructure is essential for economic and social development. However, although the presence of work zones is crucial, it also increases the risk of accidents. This study aims to evaluate drivers' subjective parameters when driving through work zones compared to normal conditions and to identify factors associated with risky driving behaviour. The analysis of abrupt manoeuvres using kinematic, visual and neurological data, still limited in the literature, offers a promising approach to identifying accident precursors, enhancing road safety. A driving test was conducted on a real highway, including sections with and without work zones of varying complexity. Drivers completed the route while wearing eye-tracking, electroencephalographic and kinematic devices, which recorded their visual, neural and driving behaviour under consistent experimental conditions. Data were analysed to compare driving conditions and predict abrupt manoeuvres using LSBoost ensemble regression models, identifying key predictive parameters in Work Zone and No Work Zone sections. Sections with work zones exhibited greater variability, lower average speeds and higher number of events. Visual analysis revealed an increased cognitive load, evidenced by higher numbers and total durations of fixations, greater pupil diameter and higher numbers and total amplitudes of saccades, while saccade peak velocity decreased. Global increases in EEG theta power, particularly in parietal areas, also reflect greater cognitive demands. LSBoost model predicted abrupt manoeuvres in Work Zone sections, with an R2 of 0.527. Curve, velocity, complexity, and EEG index were identified as the most relevant predictors. In conclusion, kinematic and neurological factors emerged as primary indicators of risky driving behaviour, with visual parameters contributing moderately but still significantly.