UTTechMedCHOIRResearch & DevelopmentProjectsPublic access defribillators and cardiac arrest

Public access defribillators and cardiac arrest

In this project we collaborate with Amsterdam UMC location AMC and the Dutch Heart Foundation. This project funds our PhD student Robin Buter, is led by project leader dr. Derya Demirtas, and also involves dr.ir. Erik Koffijberg and prof.dr.ir. Erwin Hans as promotors.

To prepare for emergencies, strategic resources or facilities are often proactively deployed without knowing exactly when and where the next emergency will be. Volunteer citizens play a vital role in rapid response to natural disasters and medical emergencies. However, little attention has been given to including uncertainty regarding responders in deciding on locations of facilities/resources. These uncertainties cause delays in response times and result in unacceptably poor outcomes. Successfully hedging against them is essential for not only survival, but also for cost effective utilization of (scarce) resources

This research develops models for locating emergency facilities under both demand and supply uncertainty, while minimizing adverse health outcomes. In particular, an important public health problem is addressed: strategic location of automated external defibrillators (AEDs) to be used by volunteers in case of an out-of-hospital cardiac arrest (OHCA). Due to the recent development of a civilian response app that dispatches nearby registered volunteers to OHCA victims, data on availability and location of volunteers is available. The goal is to find the optimal locations of new AEDs such that health outcomes of OHCA, measured by Quality-Adjusted Life Years, is maximized. This allows for analysis of cost effectiveness of current and additional optimally placed AEDs.

For finding the optimal locations, optimization techniques from the Operations Research (OR) discipline are used and supplemented with metaheuristics and data science. First, models will analyze the spatiotemporal trends in OHCAs and availability of volunteers, and estimate response rates of volunteers to alerts using historical data. Then, using literature and our own data, we will predict survival depending on response time, which will enable health outcome assessment. Afterwards, locations of AEDs are optimized using metaheuristics and neighborhood search.