Linear Regression Method to Model and Forecast the Number of Patient Visits in the Hospital

Abstract

The results accurate prediction on a census of patients in the hospital unit is very important for patient safety, health results and resource planning. The limitations of the ability to control the census and clinical tests in Cilacap area General Hospital became the reason for the importance of census forecasts at the hospital. The alternative of divination using the census of the average remains from the previous year on clinical practices have limited because of the variation of the census. The purpose of this research is to : (i) analyzing the census RSUD Cilacap every month in patients outpatient hospitalization, emergency and to develop models of divination the census, (ii) to evaluate the level of accuracy of the model compared with the average census approach remains. The data used in this study are the five years of census data at Cilacap Regional Public Hospital per month retrospectively for model development (January 2011 – December 2015) and two years of data for validation (January 2016 – December 2017). Simple linear regression method and Random Forest (RF) is used to make a forecast model for the number of inpatients, outpatient, and emergency patient visits. The model obtained was evaluated using MAPE. Based on the results obtained, the linear regression algorithm has the better performance compared to random forest algorithms in forecasting the number of inpatients, outpatients, and emergency patients visits.