Time Series Model Comparative Approaches in Analyzing Accident and Emergency Departmental Demand in Eastern Nigeria

Abstract:

The problem of the inappropriate attendances, morbidity pattern of patients in emergency condition and exhibited staff stress by some Nigerian hospitals motivated this work. The aim of the paper is to address the challenges of hospital accident and emergency departmental demand in Eastern Nigeria using statistical methods on a secondary data collected from January 2007 to December 2017. We extend a class of multivariate Vector-ARMA (VARMA) forecast models to predict the demands envisaged at Bishop Shanahan hospital’s accidents and emergency department and to do a comparative study based on existing research works. In this study we compared both univariate and multivariate time series model approaches to identify the best forecasting model to be used in addressing medical services challenges in Eastern Nigeria. Multivariate Vector Autoregressive Moving Average (VARMA (p,q)) models were compared with the Box and Genkins ARMA(p,q) model and exponential smoothing model by Holt Winter to ascertain their accuracy levels in predicting Accident and Emergency Departmental Demand. The result of the work shows evidence of increasing Accident and Emergency Department for the people within the age range of 20-45 years. The multivariate VARMA (p,q) models outperformed the univariate ARMA(p,q) models and the Holt winter’s model by having the smallest errors for all the error measures (MAE, RMSE, MAPE) considered in the research. In conclusion, we advise healthcare providers at Federal, State and Local Government levels in Nigeria to intensify efforts in the sensitization and campaign against hard drug and alcohol consumption among our youths, since our active workforce of the nation, age (20-45 years) are prone to emergency health related issues now and in the nearest future.