GPS-based data driven modeling of ambulance travel times: The case of Žilina region

Abstract

Due to the crucial role of emergency medical service vehicles in the healthcare system, the ability to more precisely represent, simulate and predict their operation will be always invaluable. This objective sets a considerable challenge to researchers worldwide, especially to those who are dealing with areas where the frequency of accident occurrences is significant. One way to quantitatively address this goal is by modeling their travel time and routing considering GPS based data. We illustrate how the data-driven model, considering spatiotemporal variables, can improve upon the state-of-the-art models. The modeling of travel time is performed for different types of origin-destination pairs. We define the problem not only for station-to-patient trips as is typically addressed by others, but also we extend the modeling to other journeys, i.e., patient-to-hospital, hospital-to-station, and patient-to-station. The consideration of these layers (different spatiotemporal variables and various trip parts) in the analysis proved to noticeably improve the predictability power.