Emergency Demand Forecast And Physician Staffing

The emergency department is for patients requiring emergency care for serious and life-threatening conditions.Nowadays,public emergency departments around the world struggle daily with congestion and increasing demand.To make sure that patients can receive therapy in time,accurate triage and demand-based physician scheduling are needed.The paradox between the over-crowded problem in rush hours and resource redundancy in spare hours is considered.Meanwhile,the waiting time targets of patients with different emergency level is also discussed.As a whole,the demand forecast containing triage classification prediction and patient arrival prediction is established at the first stage.Then,the demand-based physician scheduling model considering different patients’waiting time targets and physicians’cost is studied.Firstly,the Decison Tree algorithm is applied to emergency triage classification.Input variables are conducted from physical data,patient complaint and triage basis.Four levels given by triage nurses are considered as output variables.The synthetic minority oversampling technique is used to solve imbalance categories of the samples.As a result,useful rules are extracted and the classification result is greater than new nurses.Secondly,multiple single machine learning algorithms are used to predict the hourly,daily and weekly arrival of patients,based on variables such as emergency arrival data,weather,air quality.Then ensemble learning is established and achieves more accurate prediction,which is the foundation of doctor scheduling.Finally,a daily scheduling model is built for emergency doctors.The goal is to minimize the cost of doctors and the overtime cost of different level patients.Integer programming is established and estimated by the genetic method with a 100-day simulation.Numerical experiments invalid its practical meaning.