Modeling and forecasting monthly patient volume at a primary health care clinic using univariate time-series analysis

• Published in: Computer methods and programs in biomedicine Vol. 56; no. 3; pp. 235 - 247
• Main Authors: Abdel-Aal, R.E, Mangoud, A.M
• Format: Journal Article
• Language: English
• Published: Shannon Elsevier Ireland Ltd 01.06.1998; Elsevier Science
• Subjects: Ambulatory Care; Analysis of Variance; ARIMA analysis; Biological and medical sciences; Computer Simulation; Computerized, statistical medical data processing and models in biomedicine; Extrapolation; Forecasting; Health care; Health management; Health Services Needs and Demand; Mathematical Computing; Mathematical models; Medical sciences; Medical statistics; Modeling; Patient Admission - statistics & numerical data; Patient volume; Polynomials; Primary Health Care - statistics & numerical data; Regression analysis; Saudi Arabia; Time series analysis; Saudi Arabia; Patient volume; Health management; Time-series analysis; ARIMA analysis; Modeling; Forecasting; Statistical method; Time series; Patient; Autoregressive model; Performance; Comparative study; Monthly variation
• ISSN: 0169-2607; 1872-7565
• DOI: 10.1016/S0169-2607(98)00032-7

Two univariate time-series analysis methods have been used to model and forecast the monthly patient volume at the family and community medicine primary health care clinic of King Faisal University, Al-Khobar, Saudi Arabia. Models were based on nine years of data and forecasts made for 2 years. The optimum ARIMA model selected is an autoregressive model of the fourth order operating on the data after differencing twice at the nonseasonal level and once at the seasonal level. It gives mean and maximum absolute percentage errors of 1.86 and 4.23%, respectively, over the forecasting interval. A much simpler method based on extrapolating the growth curve of the annual means of the patient volume using a polynomial fit gives the better figures of 0.55 and 1.17%, respectively. This is due to the fairly regular nature of the data and the lack of strong random components that require ARIMA processes for modeling.