FORECASTING DAILY EMERGENCY DEPARTMENT ARRIVALS USING TIME SERIES MODELING

Abstract

With the increasing overall population of the country, the demand for healthcare services is also rising, leading to overcrowding in the Hospital Emergency Department (ED). Overcrowding can significantly impact the healthcare system by reducing patient satisfaction and quality of treatment, increasing the length of stay, and straining hospital resources. To address this issue, many researchers have suggested forecasting ED patient arrivals on an hourly, daily, monthly, or yearly basis to help hospital management plan resources efficiently. This study aims to forecast the daily patient arrivals at the ED of District Headquarters (DHQ) Hospital Charsadda to support the hospital’s management in improving operating efficiency. The data of daily patient arrivals to the ED from January 2024 to January 2025 were obtained from the Emergency Department of DHQ Hospital Charsadda. A seasonal ARIMA (SARIMA) model was applied to ED data. The SARIMA (1, 0, 1) (2, 1, 3)₇ model was identified as a best fit model based on the lowest Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC), as well as Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). The model's suitability was assessed using the Ljung-Box correlation and the Jarque-Bera normality tests, confirming that the residual terms met the assumptions for an appropriate time series model. The selected SARIMA (1, 0, 1) (2, 1, 3)₇ model yielded an MAE of 13, while the MAPE was 2.15, indicating high forecasting accuracy. Thus, it is concluded that the SARIMA is a suitable model to provide daily forecasts for ED arrivals and may assist hospital management in optimizing staff scheduling and resource allocation.