Forecasting weekly dengue incidence in Sri Lanka: Modified Autoregressive Integrated Moving Average modeling approach

Area of Science:

• Epidemiology
• Public Health
• Time Series Analysis

Background:

• Dengue presents a significant public health challenge in Sri Lanka, necessitating effective surveillance and control strategies.
• Accurate incidence forecasting is vital for proactive disease management and resource allocation.

Purpose of the Study:

• To develop and validate a statistical model for predicting weekly dengue cases in the Colombo district, Sri Lanka.
• To assess the suitability of an Autoregressive Integrated Moving Average (ARIMA) model for dengue incidence forecasting.

Main Methods:

• Utilized weekly dengue fever data from January 2015 to August 2020 for model development.
• Selected and calibrated an ARIMA (2,1,0) model with an autoregressive component of order 16.
• Validated the model using independent data and assessed residual diagnostics for randomness, constant variance, and normality.

Main Results:

• The selected ARIMA model demonstrated a close match between forecasted and observed dengue incidence.
• Model performance was evaluated using parameter significance, AIC, and SBIC criteria.
• Residual analysis confirmed the model's statistical assumptions, indicating suitability for forecasting.

Conclusions:

• The developed ARIMA model provides a valuable tool for public health decision-makers in dengue surveillance and control in Colombo.
• External factors, such as COVID-19 disruptions, can impact forecasting accuracy, highlighting the need for adaptive surveillance systems.