AI Model Predicts High-Risk Crash Locations to Improve Road Safety
-
Researchers developed a machine learning model to predict high-risk crash locations based on road features
-
The model was trained on a dataset of 15,000 km of roads across 7,000 locations in Greece
-
Key predictors of crashes road design issues, pavement damage, incomplete signage/markings
-
Model could be used for real-time monitoring and recommendations on fixing high-risk roads
-
Authors eager to apply similar AI tools to other societal problems by combining math and real-world data