Driving simulator predicts which teens are more likely to crash, CHOP study finds

Handing the car keys to a newly minted teenage motorist is one of the more white-knuckle moments of parenthood, because teen drivers crash nearly four times as often as drivers aged 20 and up.

Some of those fender-benders could be prevented if teenagers were tested first on a virtual driving simulator, a new study by Children’s Hospital of Philadelphia researchers suggests.

The authors measured the driving skills of 16,914 newly licensed Ohio teenagers and young adults on the simulator, then tracked one year’s worth of police reports to count how many got in actual car accidents.

Those who’d been classified as safe drivers on the simulator were 10% less likely than the average study participant to get in a real-world crash, the CHOP team reported in the journal Pediatrics.

Young drivers who were flagged by the simulator as having “major issues with dangerous behavior,” on the other hand, were 11% more likely than average to get in an accident. The real-world crash rate was especially high — 24% above average — for a subset of the “major issues” group whom the simulator had classified as “risky, no control.”

These riskiest drivers were prone to aggressive acceleration, tailgating, and poor lane control, among other problem behaviors, when tested on the simulation software, said CHOP cognitive neuroscientist Liz Walshe, the study’s lead author. Now that researchers can identify which drivers are at greater risk of crashing, the next step is to give them extra on-road training, the authors say.

“We can stress-test your driving skills,” Walshe said. “We can say these are the drivers who still need to practice some, and here’s some feedback on how to do that.”

» READ MORE: At the doctor's office, Philly-area teens are tested for risky driving behaviors like tailgating

Most of the study participants were age 16 to 18. Just under 3,900 participants were aged 19 to 24.

The simulation software, based on years of research by CHOP pediatrician-scientist Flaura Koplin Winston and her colleagues, was developed by a CHOP spinout firm called Diagnostic Driving Inc., based in Philadelphia.

The software is loaded into a standard computer with steering wheel and pedals attached. During the 15-minute virtual assessment, users must navigate suburban and city streets shown on the computer screen, keeping alert for other motorists and sudden distractions, such as pedestrians darting across a crosswalk at the last minute.

For the study, Walshe and other CHOP scholars collaborated with researchers at the University of Pennsylvania, Drexel University, and the University of Michigan. The research was done in cooperation with the Ohio Traffic Safety Office, a division of that state’s public safety department.

The study findings were welcomed by Emily Davidson, the safety office’s executive director.

“We applaud the work they’re doing to help make teen drivers safer behind the wheel,” she said in an email. “We are still reviewing the results of the recent study, but overall, they seem to confirm what we expected, that teen drivers who exhibit risky behaviors on virtual driving assessments are more likely to be involved in a future crash.”

Even drivers whom the simulator classified as having “no issues” with risky behavior were still at risk of a real-world crash. These safest drivers averaged about one-tenth of a crash per year, reflecting the fact that most had no crashes at all.

Drivers in the “major issues with dangerous behavior” group averaged 0.13 crashes per year, meaning that many of them also had no accidents.

Still, by the end of the study, those fractional crash risks added up to some significant damage. The CHOP researchers tracked participants for up to three years, finding that 2,301 of the 16,914 eventually got in a crash.

For Philly-area parents and guardians who want to test their kids, the software is installed in computers at 26 CHOP outpatient primary-care clinics.

More than 4,000 area teens have taken the virtual test in the past two years. The assessment is free to CHOP patients, funded with a $4 million grant from NJM Insurance Group, a New Jersey-based insurer that sells auto coverage.