Highway Accident Severity Prediction for Optimal Resource Allocation of Emergency Vehicles and Personnel

Traffic accidents could have significant impacts on people's lives. Roughly half a million highway traffic accidents occurred in 2018 in the USA. Intelligent Transportation Systems (ITSs) and Vehicular Ad Hoc Networks (VANETs) have great potential to enhance highway traffic safety and improve emergency response. This paper uses real-life traffic and accident data for a Florida highway to build prediction models to predict traffic accident severity. Accurate severity prediction is beneficial for both the responders and the drivers. First responders are in high demand, and the pandemic has made the situation worse. When an accident occurs, the emergency center dispatches a random number of emergency vehicles. Unfortunately, this number exceeds the number of vehicles needed most of the time, leaving fewer resources to respond to simultaneous accidents in different locations. Also, an increased number of emergency vehicles could introduce secondary accidents. In fact, for every ten accidents, one of them is a secondary accident. Our research gives an accurate prediction for the number of emergency vehicles needed based on the accident severity. We have used historical traffic and accident data obtained from the Florida Department of Transportation District 4 (FDOT-D4) to predict an accident and its severity. This can be used to esetimate the right number of emergency vehicles to respond to the accident. Our real-time prediction models can help reduce highway traffic accidents and congestion as well. Our prediction results demonstrate promising accuracy results and computation cost to support ITS applications.