event
Thursday 19 Nov 2020: Applications of Operational Research in Emergency Health Care
Thomas Monks and Michael Allen - University of Exeter, College of Medicine and Health
Zoom 14:30-15:30
This seminar will discuss two applied research projects in emergency health care currently underway in CMH. The first study is funded by the South West Ambulance Service Trust (SWAST) to build and evaluate forecast accuracy of models to predict daily demand for emergency ambulance services. The study tests the external validity of a benchmark model developed for SWAST in 15 time series from the London, Yorkshire and Welsh Ambulance Services. Results suggest that a combination forecast of Facebook's Prophet method and a seasonal ARIMA model provide a good balance of point forecast accuracy and uncertainty estimation up to 12 weeks ahead. The second study SAMueL (Stroke Audit Machine Learning) is funded by NIHR HS&DR. The study is using a national dataset to identify the levers for improving stroke thrombolysis: a highly time sensitive clot busting procedure used in emergency stroke care. The SAMueL discussion will focus on use of a Neural Networks, Random Forests and ensemble approaches to predict how clinical teams from high performing hospitals would treat patients in a hospital with lower performance.
