Modelling and Predicting Daily Arrival of Patients at the Accident and Emergency Department of the University of Cape Coast Hospital, Ghana

Abstract

Background: Modelling and forecasting demand for future emergency healthcare services is increasingly gaining wide attention in the emergency healthcare industry worldwide. This aids hospital managers in looking into various options to appropriately plan and allocate available scarce resources for optimal and swift performance. Despite its importance, our knowledge of daily patient flow into the Accident and Emergency Department (AED) of the University of Cape Coast Hospital is incomprehensive, and even the model that best explains its movements remains unknown. Methods: Using daily periodicity of 517 time-series observations on daily patient arrivals sourced from the AED register over January 2020 through May 2021 the autoregressive integrated moving average (ARIMA) of the classical Box-Jenkins methods of time series analysis was used to analyse the data. Results: This study revealed twenty-five non-seasonal candidate models for the hospital AED and ARIMA (0, 1, 2) emerged as the best fitting model. The study results showed that the daily patient arrivals at the AED section of the University Hospital witnessed a 50% decline on average within the study period and a further 33% decline in the forecast region. The findings also revealed very high volatility in daily patient arrivals with an average of eight patient arrivals per day. Conclusion Non-seasonal ARIMA (0, 1, 2) was identified as the best model. Thus, for policy, intervention, and future research direction, it is recommended that steps are taken to investigate the highly volatile nature of patient arrivals as well as the steady downward trending of the daily patient flow at the Accident and Emergency Department of the University Hospital