Abstract

Seasonal Autoregressive Moving Average (SARIMA) models are an extension of ARIMA models that specifically address the presence of seasonality in time series data. By incorporating both non-seasonal and seasonal components, SARIMA models capture long term trends, lagged values within seasons. The SARIMA model was applied to TB data obtained in the north health zone of Malawi from January 2013 to September, 2020. The Box Jenkin seasonal ARIMA approach was used to identify the best model for forecasting. We used the auto.arima function in R to identify the best model to predict future trends in TB case notifications. The winter multiplicative method of exponential smoothing was used to forecast future trends in TB case notifications. For model selection, we used the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Quarterly TB case notifications were analyzed, stratifying the data by disease site, HIV status, sex, and patient age group. A cyclic pattern of TB case notifications was observed, with peaks during the rainy season and at the end of the cold season. The best model for predicting future trends in TB case notifications was determined to be SARIMA (0, 1, 2) (1, 0, 0)4 (the lower AIC and BIC values, 240.81 and 246.41, respectively) Additionally, a higher proportion of TB incidence was found among males across all age groups. The study’s findings indicate an increasing trend in predicted TB incidence in the near future, accompanied by a seasonal pattern. Forecasting of PTB incidence between the years 2021 and 2024 showed a slightly increasing trend. The implications of this study highlight the importance of health education, timely medical care seeking, and proactive service planning to accommodate higher service utilization during high TB risk periods.

More details

School	: School of Natural and Applied Sciences
Issued Date	: 2023

Download full document