Machine Learning Tools for Predicting the Outcome of Pedestrian Crashes: Preliminary Findings in the Metropolitan City of Rome

reating cities and roads at a human scale is the main challenge of a not-distant future. The spread of new modes of transport reduces the distances from this future at a high speed and changes in vulnerable user urban mobility pose new questions in terms of sustainability a safety issues. Actions to promote these two vital conditions constitute the main prerequisites to be ensured by experts in the context of policies, economics, planning, and design.Road accidents are among the top ten causes of death worldwide, ranking seventh overall. On the other hand, they occupy the first place as regards deaths between the ages of 15 and 29. More than 1.2 million people in a single year die and most of them suffer from non-fatal injuries that may often impair quality of life. Although the damages to the human body are unfortunately well known, thanks also to the current scientific literature, accidents may cause a large economic cost, and knowing and understanding this macroeconomic burden could give a strong direction in an effective policy-making decision process. In this paper, following a general review of pedestrian accidents that occurred in the city of Rome, several machine learning models are proposed to predict the outcome of an accident involving a pedestrian.