Prediction of Driving Ability in Persons with Brain Disorders using Sensory-Motor and Cognitive Tests

Brain lesions and degeneration can lead to a decreased ability to perform the physical and cognitive functions necessary for safe driving. A battery of computerized sensory-motor and cognitive tests (SMCTeststrade) has been developed to quantify sensory-motor and cognitive dysfunction, with particular application to the assessment of driving abilities in patients with neurological disorders. SMCTests and an on-road driving assessment were applied to 50 subjects with brain lesions referred to the Driving and Vehicle Assessment Service at Burwood Hospital, Christchurch (36 males, 14 females; age range 43-85 years, mean age 71.3 years; 35 stroke, 4 traumatic brain injury, 4 Alzheimer´s disease, 7 with other diagnoses). Two techniques were used to build model equations for prediction of on-road driving ability based on SMCTests performance - binary logistic regression and nonlinear causal resource analysis (NCRA). Binary logistic regression correctly classified 94% of referrals as on-road pass or fail, while NCRA correctly classified 90% of referrals. Leave-one-out cross-validation analysis estimated that binary logistic regression would correctly predict the classification of 86% of an independent referral group as on-road pass or fail, while NCRA would correctly predict 76%. Results indicate that the predictive model based on binary logistic regression would be slightly more accurate than NCRA at predicting on-road pass or fail in an independent subject group. Conversely, the NCRA model also provides valuable information on the extent to which a subject would pass or fail an on-road driving assessment and identifies the deficit which most limits their driving ability