Predictive Modeling of Severity of Injuries in Motor Vehicle Crashes

• Activity Number: 297
• Type: Topic Contributed
• Date/Time: Tuesday, August 2, 2016 : 8:30 AM to 10:20 AM
• Sponsor: Government Statistics Section
• Author(s): Aditi Pradeep Sharma* and Michael Wierzbicki and Gaurav Sharma
• Companies: University of Maryland Baltimore County and The EMMES Corporation and The EMMES Corporation
• Keywords: Traffic Safety ; Motor Vehicle Crashes ; Injury Severity ; Variable Selection ; Machine Learning ; Predictive Modeling

Abstract

According to the Centers for Disease Control and Prevention, motor vehicle crash deaths in 2013 resulted in $44 billion in medical and work loss costs. The National Highway Traffic Safety Administration reports that over 32,000 people die in motor vehicle crashes each year. An additional 2.3 million are injured or disabled. Therefore, it is important to identify and explore associations between injury severity and crash-level, vehicle-level and person-level characteristics. Using the Department of Transportation's General Estimates System (GES) data collected over the past 5 years, multiple characteristics of motor vehicle crashes will be explored, potential causal predictors of injury severity resulting from motor vehicle crashes will be determined, and a model that would accurately predict severity will be developed. In addition, statistical evaluation of the sensitivity of results to the particular variable selection and predictive modeling approach will be presented.