EURO 2024 Copenhagen
Abstract Submission

EURO-Online login

1705. Enhancing Emergency Response Efficiency via Deep Reinforcement Learning: A Novel Model for Dynamic Dispatching

Invited abstract in session TA-55: ML & OR Applications in Transport Modelling, stream Transportation.

Tuesday, 8:30-10:00
Room: S02 (building: 101)

Authors (first author is the speaker)

1. Felix Rauschert
Big Data Analytics in Transportation, TU Dresden
2. Matthias Ribesmeier
Chair Of Railway Operations, TU Dresden
3. Pascal Kerschke
Big Data Analytics in Transportation, TU Dresden & ScaDS.AI

Abstract

Emergency Medical Services (EMS) are essential for providing timely pre-hospital care. Efficient EMS dispatching, a critical yet complex decision-making process, involves not just quick responses to emergency calls but also optimal resource allocation amidst unpredictable demands. While traditional dispatch strategies typically prioritize sending the nearest vehicle, research has shown this approach does not yield the best outcomes. We introduce a comprehensive solution that integrates factors like scalability, uncertainties, dynamic environments, and adaptive strategies, thereby enhancing the EMS system's ability to respond more effectively and efficiently.

We propose a novel approach using Deep Reinforcement Learning (DRL) to refine EMS response efficiency. By integrating key metrics such as emergency severity and response time, we offer a strategic decision-making framework that significantly improves operational outcomes. Utilizing publicly available data from San Francisco, we highlight our DRL model's effectiveness in surpassing traditional dispatching methods. Our approach uniquely leverages a larger state-action space for DRL and incorporates crucial performance indicators, making EMS deployment more nuanced and efficient than in previous studies. This results in faster, more effective incident responses than traditional strategies, showcasing the potential of advanced machine learning in emergency dispatch optimization.

Keywords

Status: accepted

Back to the list of papers