APPLICATION OF TIME SERIES MODELS TO PREDICT DAILY EMERGENCY DEPARTMENT VISITS

Abstract

With the country’s growing population, the demand for healthcare services has increased, resulting in overcrowding in hospital Emergency Departments (EDs). Such overcrowding can adversely affect the healthcare system by diminishing patient satisfaction and quality of care, extending patient length of stay, and placing additional strain on hospital resources. To mitigate these challenges, researchers have emphasized the importance of forecasting ED patient arrivals—on hourly, daily, monthly, or yearly scales—to enable efficient resource planning and management. This study focuses on forecasting daily patient arrivals at the ED of District Headquarters (DHQ) Hospital Charsadda, with the aim of supporting hospital management in enhancing operational efficiency. Data on daily ED arrivals from January 2024 to January 2025 were collected from the hospital records. A Seasonal ARIMA (SARIMA) model was applied, with SARIMA (1, 0, 1) (2, 1, 3) _7 identified as the best-fit model based on the lowest values of the Akaike Information Criterion (AIC), Bayesian Information Criterion (BIC), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE). Model adequacy was confirmed through the Ljung-Box test for autocorrelation and the Jarque-Bera test for normality of residuals, validating the assumptions of the time series model. The selected SARIMA model produced an MAE of 13 and a MAPE of 2.15, demonstrating high forecasting accuracy. These results indicate that SARIMA is an effective tool for predicting daily ED arrivals and can assist hospital management in optimizing staff scheduling and resource allocation.