Time series analysis of patient attendance in Uganda: a case study of Makerere University Hospital

Abstract

Time series analysis and forecasting has become a major tool in many applications in the medical sector. This paper examines the trend at which patients attend the hospital over a period of the study, models Auto Regressive Integrated Moving Average (ARIMA) and forecasts hospital attendance in Makerere university hospital, Uganda. Secondary data with monthly Out Patient Department data (OPD) from January, 2018 to July, 2021 from the Makerere University Hospital was used. Data was analyzed using STATA statistical software. The study therefore shows the usefulness of the statistics to medicine and hence advice has been sought on the analysis and interpretation of medical data. The work is presented in five chapters. The first chapter being the introduction, the second chapter is a review of the related literature where views of several of the various writers on the topic concerned were analyzed. Chapter three is the research methodology, it examines the various research methods used in the data collection. All data collected was analyzed in chapter four. Finally, summaries were made in chapter five for the entire research and conclusions and suggestions were made on how to improve health care to the people. Among the most effective approaches for analyzing time series data is the model introduced by Box and Jenkins, ARIMA. With the analysis of the data, it was found that the data was non-stationary, and it was, therefore. differenced once to make it stationary. Also, with the help of the ACF and PACF plots, tentative models were fit to the data. ARIMA (4,1, 2) was noted to fit the data well. Further adequacy test on the model also confirmed the validity of the selected model. The model was used to forecast. for monthly cases of malaria for the next one year (12 subsequent months)