Mathematical prediction patterns for pre-hospital emergencies
BACKGROUND
The mathematics predictions of different occurrences are no longer science fiction. Complex mathematical models are used to help to anticipate and optimize different industries. Designing mathematical models which describe the past, we are able to anticipate the future. Our aim is to use this into pre-hospital field.
Today staffing and capacitating method for SMURD Romanian EMS is rather static. Our goal is to use a multidisciplinary approach to improve this process by finding a pattern and predict the needs.
Patterns can be identified everywhere. We came upon a predictable pattern to demonstrate this in pre-hospital to understand when o this extra staff and ambulances are needed.
The objective of this study is to prove a mathematical model based on historical contextual factors and apply this for forecasting. Prediction models may provide an efficient way of ambulance distribution, based on historical needs.
METHODS
The study investigates the number, type and gravity of emergencies, correlating them with days of the week and types of cases. Data source is SMURD Sibiu county database on a period of 8 years (2010 - 2017). The number emergencies addressed by EMS is calculated for different pathologies, time intervals and type of ambulances.
An auto-regressive integrated moving average (ARIMA) mathematical model is chosen as this model works with time series data in order to predict future points in the series (forecasting).
RESULTS
The dynamic of 112 calls changes specifically by time and location. We create a prediction curve to anticipate the most probable occurrence in the future. During national holidays, the number of emergencies increases regardless the pathology. Intoxication cases increase by 40% (possibly because of ethanol abuse) and trauma by 11%.
Also we are able to create prediction curves to anticipate crowding and more demanding periods of time.
CONCLUSIONS:
This mathematical model is able to predict the pre-hospital emergency calls. We are able to forecast a rise of medical incidents based on contextual and social events.
The old planning techniques, which assume that the demand of ambulances is known upfront, are proven not reliable in certain circumstances because there is a need of a dynamical and contextual tuning. Emergency call patterns tend to be highly busty and time and location dependent. A scientific approach bring the experience and competence of other fields to emergency medicine to simply save lives
