An Optimized ARIMA Model for Emergency Medical Services Time Series Demand Forecasting Using Bayesian Methods

Abstract

Predicting future demand for emergency services through time series forecasting is a useful tool for emergency medical services (EMS). Accurate forecasting of emergency needs is critical to EMS success and efficiency. Spatial management can be improved by better transportation before incidents, leading to significant improvements in response time, prehospital care, better outcomes, and survival quantitative Autoregressive Integrated Moving Average (ARIMA) models are popularly used for time series forecasting. A systematic approach used a grid search to find the parameter space (p, d, q). Bayesian optimization was used to improve our model by identifying the best over-parameters of the ARIMA model, resulting in improved prediction performance f guarantees Our results suggest automatic and heuristic approaches to state together can be effective for optimizing EMS time-series forecasting, to provide valuable information to optimize EMS availability and resource management