Predicting A Pattern of Patient Arrival at Emergency Department by Using Data Mining Technique and Neural Network Model

Emergency department (ED) is the first place for providing diagnostic and therapeutic services to emergency patients. Due to importance of speed and accuracy in providing services; the proper allocation of resources، the department must consider this matter in a particular way. Planning Emergency resources implements regardless of patient overcrowding which occurs at different times. In conclusion the emergency department may faces lack of resources and it results in delay of providing services، a whole mess in functions and decreasing in quality of services. This study is aimed to overcome these problems by suggesting a model for predicting the number of arrival patients at ED. The number of arrival patients is predicted based on the data colleted by counting arrival patients and using the data mining technique and neural network model (Multi-layer Perceptron). The number of arrival patients during whole days of a week and 24 hours a day were calculated by sorting out 1، 2 and 3 priorities. The highest number of arrival patients counted was for Saturdays and the lowest for Fridays. Holidays and non-holidays` number of arrival patients differ. The number of arrival patients on formal holidays was similar to Fridays. The highest number of arrivals was between 9 am and 11 and also between 20 pm and 23 pm and the lowest arrivals was between 2 am and 7 am. prediction the number of ED arrival patients can be used for estimating required sources and distributing them appropriately and improving quality of services.