Residual control chart monitoring for autocorrelated epidemiological data: an application to monthly dengue cases

Abstract

Dengue fever remains a major public health threat in tropical and subtropical regions, particularly in densely populated urban areas. An effective surveillance system is essential to detect early surges in case numbers and prevent widespread outbreaks. However, conventional Statistical Process Control (SPC) techniques typically assume that observations are independent and identically distributed, an assumption often violated in epidemiological time series data due to temporal autocorrelation. This study proposes a hybrid monitoring framework that integrates time series modelling with SPC to address the autocorrelation structure in monthly dengue case data. First, an autoregressive integrated moving average (ARIMA) model is employed to capture the temporal dependencies in the data. The residuals from the ARIMA model—assumed to be approximately independent—are then analysed using an Individual Moving Range (IMR) control chart. The proposed approach is applied to monthly dengue case data from Makassar, Indonesia, covering January 2013 to December 2024. The results demonstrate that residual-based control charts are more effective in identifying out-of-control signals that align with recorded dengue outbreaks, compared to traditional SPC methods applied directly to raw data. This method provides a statistically robust and practical tool for enhancing early warning systems in dengue surveillance.