PI: Alexandre Jacquillat
University: Carnegie Mellon University
Emergency Medical Services (EMS) have been instrumental to reduce mortality and improve public health. At the same time, the capabilities of EMS systems are constrained by limited vehicle fleets, which may prevent them from responding to emergencies within adequate time windows.
At the same time, the growth of ride-sharing platforms (e.g., Uber, Lyft) creates opportunities to augment EMS systems with on-demand, distributed transportation resources. This can take place through two mechanisms. First, on-demand transportation can transport low-priority patients to medical care facilities in order to free up EMS capacity to respond to higher-priority emergencies. This strategy leverages ride-sharing as a substitute to EMS. Second, ride- sharing vehicles can deliver life-saving equipment to emergency locations to enhance first response. This strategy leverages ride-sharing as a complement of EMS.
The proposed project will develop mathematical models and computational algorithms to optimize the utilization of available EMS vehicles and on-demand ride-sharing services to respond to emergencies. The model will be formulated as a continuous-time Markov decision process which optimizes the dispatch of EMS vehicles to emergency locations and ride-sharing requests as substitutes or complements of EMS vehicles, with the objectives of maximizing the health outcomes and system efficiency associated with emergency care. Tailored algorithms from the field of approximate dynamic programming will be developed to implement the model efficiently, and derive near-optimal policies in reasonable computational times. Using real-world data provided by the University of Pittsburgh Department of Emergency Medicine, the model will be implemented to simulate the impact of the proposed strategies on EMS systems and evaluate their potential benefits in a realistic environment. The project will conclude with a deployment plan for the resulting systems and technologies in the City of Pittsburgh to assess their impact on patient wait times and resulting care delivery.