Trip-Based Measures of Naturalistic Driving: Considerations and Connections With Cognitive Status in Older Adult Drivers

Abstract

Older adult drivers can experience age-related health declines, particularly cognitive health, that can negatively impact driving performance and lead to driver license revocation. The measurement of naturalistic in-car driving behavior can be used to evaluate trip complexity, including diversity, length, and frequency of trips and associated destinations. This paper examined measurement methods for trip complexity and driving destinations using GPS data for older adult drivers with differing health statuses, focusing primarily on cognitive health status. Older driver subgroups included four groups with relatively stable health: better overall health, better cognitive health, worse cognitive health, and worse overall health and one group with declining cognitive health. Older drivers with better health status (overall and cognitive) had higher measured trip complexity compared to those with worse health status. Two variables, mean trip distance and percent of trips driven during the workweek evening rush-hour, declined significantly $(p \le 0.049)$ in-line with cognitive declines but did not meaningfully discriminate cognitively declining older drivers from cognitively stable drivers (sensitivity: 18.8%–43.8%; specificity: 58.0%–93.3%). This contrast between significant group differences and nonpredictive declining group changes may suggest that self-referential naturalistic driving measures are needed to identify meaningful changes in driving behavior. In addition to the development of self-referential driving measurement systems, careful consideration of big data analysis as they apply to naturalistic driving is warranted. These include but are not limited to issues of validation, anonymity, measure-based definitions, and occurrence of outlier-type driving events like city-to-city travel.