More efficient ambulance planning with new mathematical models

In serious life-threatening situations a fast ambulance response is required. Late arrivals can have a serious impact on the well-being of the patients and can also have consequences regarding policy making. In practice, providing a fast ambulance response time is not always possible. PhD candidate Martin van Buuren of CWI has developed mathematical models that can help to realize fast response times.

Publication date: 15-05-2018

In serious life-threatening situations a fast ambulance response is required. Late arrivals can have a serious impact on the well-being of the patients and can also have consequences regarding policy making. In practice, a fast ambulance response time is not always possible. PhD candidate Martin van Buuren of CWI has developed mathematical models that can help to realize fast response times.

Urban versus rural areas

The dispersion of ambulances in regions with combinations of urban and rural areas often produces a field of tension. Current ambulance allocation models are more likely to position ambulances in urban areas due to the large call volume, while in practice ambulance service providers position bases and ambulances in rural areas as well in order to fulfill local needs. Blindly applying existing models to these ‘mixed’ regions,  generally leads to overstaffing, and thus unnecessarily high costs. Van Buuren now developed a so called ‘adjusted queuing approach’ taking into account the different characteristics of urban versus rural areas, resulting in more realistic ambulance planning. Van Buuren calculated for example that in the Purmerland region the occupation of ambulance could be reduced from 3 to 1. With this approach the so called demand projection (great demand in urban areas projected to more quiet rural areas, in this case from the Amsterdam Noord region) is eliminated.

Fastest route not always the best

Dispatchers usually relocate ambulances by the fastest route. But sometimes the fastest relocation is not the best one, as the coverage while driving over the fastest route can be relatively low. In some cases a route through villages can lead to better coverage than a route via a highway with little exits. Van Buuren has studied the effect of taking alternative routes in ambulance relocations. The research results for four actual ambulance regions show that this so-called dynamic routing can help at an operational level to obtain a more fair distribution of ambulance coverage, while keeping the fraction of late arrivals over the entire ambulance region stable.

Pro-active relocation

Pro-active relocation is a powerful means to boost ambulance efficiency. It means ambulances go to a location at which they can provide a better coverage to the region compared to assigning each vehicle a fixed base. In 2015 Van Buuren has run a pilot with relocation algorithms developed by CWI in a study at research partner GGD Flevoland. These algorithms help to give advice to the dispatchers about when, where and which ambulance to relocate. The pilot was successful: the effectiveness of relocations improved when using a dynamic relocation policy, compared to previous years when the relocation algorithms were not used. By using the new algorithms and software GGD Flevoland reached the nationwide target: 95% of the ambulances reached the incident within 15 minutes. Van Buuren processed the adjusted algorithms in a tool and connected them to existing software used by the dispatch center. GGD Flevoland and other dispatch centers adopted the CWI policies for ongoing use and permanent implementation.

Martin van Buuren will defend his PhD thesis 'Efficient Planning of Ambulance Services - Theory and Practice' at 16 May from 9.45 to 13.00 at VU University Amsterdam.

Promotors: Prof.dr. R.D van der Mei (VU and CWI) and prof.dr. S. Bhulai (VU)