P3‐405: Comparison of a Novel, Naturalistic Driving Assessment System with Self‐Reported Driving Behavior in a Sample of Cognitively Normal Older Adults

Abstract

participated). Harlie was programmed to randomly choose one topic at a time out of a repertoire of six total topics, including the specialised topic and five generic topis, such as 'music' or 'good memories'. Participants were seated individually and shown how to chat with Harlie on a smartphone. Headphones were used and each participant spoke to Harlie for approximately 5 minutes. Audio and text recordings were securely saved to an online server for analysis by the research team and participants were encouraged to share feedback on the experience during a group discussion. Results:The feedback on the experience of using Harlie was positive in both groups. While the knitting module was successful in prolonging conversational engagement, the woodworking module was not eliciting longer engagement for this specialised topic. Conclusions: All community members were able to successfully use the Harlie application, feedback was positive and differential chat preferences could be identified. This is a promising finding with potential future applications for using this chatbot with isolated older adults including those with dementia. By following the approach described here, individual chat preferences can be explored for each potential user, which could further enhance the experience of chatting with Harlie. Background: Changes in driving behavior among older drivers has become an increasing area of public and research interest as older adults continue to drive longer. Controlled conditions like on-theroad tests and driving simulators evaluate driving ability in clinical populations and older adults, but are limited in understanding daily driving behavior. Existing research that measure naturalistic driving behavior are often expensive, obtrusive, require extensive modifications to a vehicle and involve cumbersome download procedures. To address this gap in research technology, we developed a new methodology including, data acquisition, curation and archiving procedures, and wrote custom computer codes (using R and ArcGIS) to modify a commercial off the shelf global positioning system (GPS) solution for our own use. We then conducted a pilot study to assess the degree to which our naturalistic driving system was associated with self-reported driving behavior among older adults with and without preclinical Alzheimer's disease (AD). Methods: Participants (N¼20) with normal cognition (Clinical Dementia Rating, 0), aged 65 years and older were recruited from the Knight Alzheimer Disease Research Center. Participants completed a clinical assessment and psychometric battery, brain amyloid imaging, and cerebrospinal fluid (CSF) collection. Driving behavior was evaluated using the self-report, Driving Habits Questionnaire (DHQ) and by our system, which incorporated, a GPS device plugged into participant's vehicle. We examined correlations between five main outcomes collected by both methods: speeding, driving space, total miles driven, total number of trips, and number of days-per-week driven. Group differences were also exami...