Predicting cognitive status in stroke survivors from driving performance.

Introduction: This study aimed to determine whether simulated driving performance can reliably predict cognitive impairment in stroke survivors.

Methods: Cognitively impaired (n = 35) and normal (n = 54) stroke survivors completed a simulated driving course with reactive, distracted, and route-planning sections. Performance was assessed using lane departures, average speed, brake reaction time, task completion time, and route accuracy.

Results: Logistic regression models correctly distinguished cognitive status in 77.5% of cases for reactive and distracted driving, and 80.9% for route planning. Notably, the route planning task also achieved the highest classification rate of cognitively impaired participants (∼70%). Receiver operating characteristic (ROC) analyses on the strongest predictors from each driving section revealed significant areas under the curve (AUCs), with optimal cutoffs identifying cognitively impaired participants at 70%-80% accuracy.

Discussion: These findings provide a critical foundation for developing simulator-based assessments as practical, functionally relevant screening tools for identifying cognitive impairment and determining driving readiness post-stroke.

Highlights: Stroke survivors were tested on simulated driving tasks.Driving metrics were lane departures, speed, reaction time, and route accuracy.Cognitive status was predicted with greater than 75% accuracy.Simulators may be a clinical tool for assessing post-stroke driving readiness.