Paper Title

Abstract - Healthcare in the US went through radical changein the past decade. Not only is the quality of healthcare a main concern, but customer satisfaction now finds a dominant place in the medical field. The federal government has placed a customer satisfaction component into their hospital reimbursements for Medicare and Medicaid. An important medical arena that most certainly falls in both quality and customer service is emergency first response. In many cases, this is the introduction of the patient into the healthcare system. The first interaction is critical for patient care and has definite customer satisfaction ramifications. Therefore, Emergency Medical Services (EMS) managers find themselves in a critical position. They have finite resources but must respond to emergencies within set time ranges. The number and dispersal of ambulances directly address this issue. Artificial Neural Networks (ANN) have been introduced as an engine to forecast EMS call volume. To be used in real-time, and not just for aggregated capacity planning, these forecasts must have a fine level of granularity regarding time and space. If the time bucket or special component are too big, then the ability of these forecasts to improve ambulance deployment models sufferers. A sensitivity analysis of the major components of ANN will be conducted in an attempt to determine the architecture of the most accurate forecasting model. Components of the neural network, such as the input layer, processing layer(s), output layer, activation function, and the training, validation, testing ratio, will be adjusted to maximize or simply improve forecast accuracy. Keywords - Emergency Medical Services, Neural Networks, EMS Calls, Forecasting, Ambulance Service Demand.