Refining the ambulance travel model for more accurate transport time estimation

Abstract

This paper refines the ambulance travel model, which is used to determine the optimal route from an emergency site to a hospital, to enhance the accuracy of travel time estimation. The model provides driving rules regarding the acceleration and velocity of the ambulance under the assumption that it moves along a route within a road network consisting of nodes and links. The conventional model, proposed in 2019, has a drawback: it cannot accurately estimate travel time in two specific situations - when the route includes road sections where traffic signals are synchronously controlled, and when there are significant hourly variations in traffic volume. To address this drawback, two new rules are incorporated into the conventional model, resulting in a modified version. To account for hourly variations in traffic volume, the Directions API of Google Maps Platform is utilized to adjust the maximum speed of the ambulance. Based on 90 ambulance transport records collected in Hiroshima City, the model parameters are optimized using the genetic algorithm by minimizing three objective functions related to travel time, a patient’s blood pressure fluctuations, and the compressive force exerted on the patient’s back, in terms of Pareto optimality. As a result of the modification, the estimation accuracy of travel time improved by an average of 4.2%. Especially for the transport route with road sections where traffic signals are placed at short intervals, which the conventional model struggled to handle effectively, the accuracy improved by 97.8%.