Abstract
Purpose :
To demonstrate that driving performance under a challenging environment shows a strong decrease beyond a certain level of visual field loss in glaucoma patients.
Methods :
Thirty eight glaucoma and 30 age-matched healthy patients drove a winding one lane wide (3.6 meters) country road with a limited preview at 30 meters and a reduced contrast (25% between on-road and off-road surface) as experienced in fog on unmarked roads. Performance was quantified using the inverse of the straight heading time-to-line-crossing (InvTLC) defined as the time it would take the car’s center of gravity to cross a lane boundary if its heading were maintained. InvTLC quantifies the safety margin available to the driver before heading needs to be changed. Standard performance metrics such as standard deviation of lateral position are inconsistent with the fact that drivers naturally cut curves and thus experience high lateral positions when driving expediently on a winding road. We developed a measure called behavioral entropy that first establishes for every road point the distribution of InvTLCs observed across healthy drivers and then uses those distributions to compute the entropy of each particular InvTLC profile observed in heathy and glaucoma patients. If a driver shows many outliers along the drive, then the entropy of its drive will be high. Different models that map the standard automated perimetry (SAP) mean deviation (MD) of better eye to the observed entropy of the InvTLC profile were established to determine whether driving performance is protected up to a certain SAP MD of better eye after which it deteriorates.
Results :
On average patients with glaucoma had significantly degraded performance when driving under fog conditions (mean InvTLC was 1.00 vs. 1.34; P=005). A two-linear-segment model and an exponential model show that the breakpoint falls around a SAP MD of better eye of -12 dB. It is also shown that such a clear result cannot be obtained by using standard metrics, thus showing the power of the behavioral entropy approach.
Conclusions :
The study demonstrates that the communication theoretic metric of behavioral entropy of InvTLC profiles offers a greater sensitivity to the effects of glaucoma on driving performance than standard metrics. Behavioral entropy is a sensitive metric for any time series behavioral data for which normative data is available.
This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.