There is increasing concern and interest about the association of cognitive impairments and driving performance among the elderly, and several recent studies have identified significant driving performance deficits in cognitively impaired older people, measured by means of changes in driving simulator metrics. In this paper, it is attempted to reverse the question: can driving at the simulator reveal the presence of cognitive impairments? This question has a two-fold interest: first, driving at the simulator may allow for the detection of subtle changes in driving due to cognitive impairments imperceptible in one’s daily routine; and second, driving simulators may have potential of becoming in the future useful tools for the screening of older individuals and assist clinicians both in the medical examination and the advice on whether to continue driving. Data from a large interdisciplinary driving simulator study were analyzed by means of discriminant analysis techniques, in order to classify individuals as healthy or cognitively impaired on the basis of their simulated driving performance. The analysis sample included 86 individuals, out of which 38 patients with Mild Cognitive Impairment (MCI) and 21 patients with Alzheimer’s disease (AD). The results suggest that variables discriminating between healthy and impaired individuals are average speed and headway, lateral position variability, throttle position, reaction time and accident occurrence at incidents. The functions developed correctly classified more than 65% of the individuals, a share that dropped to around 60% when cross-validation analysis was implemented. Overall, although MCI and AD patients had significant shares of misclassified cases, these misclassifications were mostly between the one pathology and the other; very few pathological cases were classified as healthy, and all of these concerned MCI patients. It is indicated that driving at the simulator may under certain conditions assist in the screening for cognitive impairments.
|cognitive impairment, driver behaviour, driving simulator, older drivers, statistical modelling