The number of patients using emergency medical services (EMS) through ambulance dispatch has been increasing annually in Taiwan. Due to limited medical resource allocation, patients may not get on-time admission after they are sent to a hospital Emergency Department. This study aimed to construct a forecasting system to predict the availability of ED and ICU beds. A simulation-based forecasting system integrated with Google Maps is proposed to provide ED recommendations for ambulance dispatch. The web crawler technique continuously collects open data from the emergency information systems. An arrival transfer mechanism was proposed to convert the raw data into a simulation input. The results were then integrated with hospital assessment and routing distance in Google Maps to provide the most appropriate hospital ED to which a patient should be sent. The results provided forecast accuracy for bed availability for the next 20, 40, and 60 min in 10 selected hospitals in central Taiwan. In most hospitals, the forecasting accuracy is high. For example, the ED and ICU bed availability accuracies in the next 20, 40, and 60 min are [100%, 88.7%], [100%, 90.5%], and [100%, 92.1%]. The two scenarios also showed hospital recommendations based on the available beds, routing distance, and eight-dimensional assessments. Previous EMS research usually did not consider future bed availability as the research target. This study predicted future bed availability and recommended the most appropriate hospitals for EMS dispatchers.

Emergency department crowding has been one of the main issues in the health system in Taiwan. Previous studies have usually targeted the process improvement of patient treatment flow due to the difficulty of collecting Emergency Department (ED) staff data. In this study, we have proposed a hybrid model with Discrete Event Simulation, radio frequency identification applications, and activity-relationship diagrams to simulate the nurse movement flows and identify the relationship between different treatment sections. We used the results to formulate four facility layouts. Through comparing four scenarios, the simulation results indicated that 2.2 km of traveling distance or 140 min of traveling time reduction per nurse could be achieved from the best scenario.