Development and validation of methods to predict ground- and helicopter-based emergency medical service travel times

Abstract Background Helicopter-based Emergency Medical Services (EMS) are a costly and finite resource and should only be deployed when they provide substantial benefits to patients. One such benefit is a more rapid response to incidents and transport to hospital compared to Ground-based EMS. This study evaluates methods for estimating travel times for the purpose of optimizing the selection of helicopter vs. ground EMS. Methods Ground EMS travel times were estimated using an open-source routing tool, the estimates of which were adjusted using linear regression and gradient boosting models based on historical travel times. Helicopter travel times were similarly estimated based on great circle distances adjusted by historical data. Models were evaluated using several measures of predictive accuracy to enable comparison with other studies. Results In validation data collected after finalization of the methods, the best performing prediction models for Ground EMS travel times explained a substantial portion of variation in drive times to location and to hospital (R2 0.79 and 0.70 respectively), and less for helicopter flight times (R2 0.46 and 0.38 respectively). Mean average percentage error was 18-20% for all investigated estimation techniques and time intervals. Conclusion Integrating tools based on the open-source code provided with this study has the potential to improve the ability of dispatchers to estimate relative travel times and assess the potential time benefit of directing the most appropriate resources to a given patient.