TIME SERIES MODELLING OF HYPERTENSION CASES IN BORNO STATE: A COMPARATIVE STUDY OF ARIMA AND SARIMA

Abstract

Hypertension remains a major public health challenge in Nigeria, especially in conflict-affected regions such as Borno State, where disruptions in healthcare delivery hinder effective disease monitoring. Accurate forecasting of hypertension cases is essential for planning, resource allocation, and early intervention. This study applied Autoregressive Integrated Moving Average (ARIMA) and Seasonal ARIMA (SARIMA) models to monthly hypertension data recorded from major hospitals in Maiduguri between January 2015 and December 2024 (N = 120). Data preprocessing involved handling missing values, log transformation, stationarity checks, and model identification using ACF and PACF diagnostics. Competing models were evaluated using Akaike Information Criterion (AIC), Mean Absolute Percentage Error (MAPE), and Mean Squared Error (MSE). Results showed that the SARIMA model significantly outperformed the non-seasonal ARIMA model across all forecasting metrics, demonstrating its ability to capture the strong seasonal fluctuations in the series. SARIMA produced lower forecasting errors, narrower confidence intervals, and more stable predictions for 2025. While the study relied on hospital-based data that may not fully represent community-wide burden, the findings highlight the effectiveness of seasonal forecasting methods for strengthening public health planning in Maiduguri. Future work should incorporate exogenous predictors such as environmental variables, displacement trends, and socioeconomic conditions, as well as explore hybrid time-series and machine-learning models.