Motor vehicle crashes lead causes of adolescent mortality, making this a major health threat facing US adolescents. Given that 95.6% of novice driver crashes are due to driver error, improved skill should reduce crash incidence. Research supports this: novice driver crash rates peak immediately following licensure and decline steeply with experience gained over the months following licensure. However, it is not known which critical driving skill deficits present at licensure predict crashes. The long-term goal of this research is to develop interventions prior to licensure that can lead to the safest independent driving post licensure. With this R21 project, we will identify deficits in driving skills that are (a) associated with major predictors of early crash risk (age at licensure and sex) and that (b) can predict differences in crash rates within the first year of independent licensure. We will leverage our exclusive access to an innovative, newly available data source for young driver research. To enable study of safety critical driving skill deficits in novice drivers, results from a new (as of July 2017) virtual driving test (VDT) - developed by the PI, adopted by the State of Ohio and delivered immediately prior to the on-road licensing examination - will be linked to Ohio licensing and crash data. We propose these aims:
Aim 1 : identify skill clusters using multivariate clustering, regression and machine learning techniques to test the hypothesis that these clusters are related to age at licensure, sex, and time spent in the learner period;
Aim 2 a: use data on driving skills linked with reported crash data in Ohio to determine whether the skill clusters identified in Aim 1 are associated with crash risk;
and Aim 2 b: determine if those skill clusters from Aim1 are predictive of crashes in rural teen drivers. With a large sample of new driver applicants (>25,000), this study will examine, for the first time, differences in skill, starting at an age (adolescence) and a time (within the first year of licensure) when crash risk is high. We will create a normative dataset of novice driver skills; classify skill deficits by age, sex and time in learner period; and quantify the relationship between skill deficits and crash rates during the first 2 months of licensure (when crash rates are highest for young drivers) and the following 10 months as experience accumulates. If the use of the VDT proves successful in predicting crash risk, we will be able to identify novice drivers with elevated crash risk before granting a license. This R21 will enable the following lines of research: (1) Develop new interventions that address skill deficits through training, individualized to the needs of the novice driver (in partnership with Ohio). (2) Develop or adapt advanced driver assistance systems (ADAS) or self-driving vehicle technology to mitigate identified skill deficits (as part of Dr. Winston?s National Science Foundation Industry/University Cooperative Research Center). (3) Conduct a prospective study (R01) of the interacting trajectories of driving skill acquisition and cognitive development that lead to individual differences in skills at the time of licensure and the ensuing crash risk.
We will identify safety-critical driving skill deficits immediately prior to licensure that predict crashes, starting at an age (adolescence) and a time (within the first year of licensure) when crash risk is high. We will leverage exclusive access to newly available data on >25,000 novice drivers: state licensing and crash data linked to results from a new (as of July 2017), well-validated virtual driving test that identifies driving skill deficits (adopted by Ohio). If this R21 finds the VDT successful in predicting crash risk, we plan to: (1) conduct a future study of the interacting trajectories of driving skill acquisition and cognitive development that lead to individual differences in skills at the time of licensure and the ensuing crash risk and (2) develop new interventions (training and vehicle technology) to address this risk.
