Time Series Approaches for Modeling and Forecasting Hospital Inpatient Flow

Time Series Approaches for Modeling and Forecasting Hospital Inpatient Flow

Samira Rasouli1 Hamed Tabesh2 Kobra Etminani3

1) MSc of Medical Informatics, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran Email:
2) Assistant Professor, Biostatistics, Medical Informatics Department, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran Email:
3) Assistant Professor, Medical Informatics, Medical Informatics Department, Faculty of Medicine, Mashhad University of Medical Sciences, Mashhad, Iran Email:

IOI Code : XABC-BKACZ
Publication : International Conference on Hospital Management(ihmc9.ir)
Abstract :
    Introduction: The study and analysis of hospital processes have become a necessity for overall performance improvement of the hospital systems through developing improved policies and decisions. In this context, management and analysis of patient numbers is an important factor. The goals of this paper are to explore and evaluate the use of multiple time series forecasting methods to predict monthly hospital inpatient numbers at six different hospitals and to compare the accuracy performance of these methods. Methods: Monthly inpatient admissions at six hospitals were collected for the period March 2004 through March 2016. We evaluated the use of Holt-winters, Seasonal Autoregressive Integrated Moving Average (SARIMA), Multilayer Perceptron (MLP) and Generalized Regression Neural Networks (GRNN) models to forecast monthly inpatient numbers at each hospital for horizons ranging from 1 to 3 months in advance. The performance of the models was measured based on the Mean Absolute Percentage Error (MAPE) accuracy measure. Results: In this study, Holt-Winters method provides optimal forecasting performance for four hospitals. Totally, best forecasting models of all six hospitals had a MAPE from 2.13% to 4.12%, indicating an accurate prediction. Discussion & Conclusion: Time series analysis is shown to provide an appropriate, readily available tool for predicting the number of hospital inpatient numbers. As presented in this study, given the unique characteristics of different hospitals, optimal predictive models can vary from hospital to hospital. However, applied methods in this study may assist other hospitals to improve their resource allocation and strategic planning.

Keywords : forecasting inpatient numbers holt winters sarima neural network