A new study suggests that advanced machine learning models could help Bangladesh predict dengue outbreaks and mount faster responses, with temperature emerging as the most critical warning signal.
Researchers from the University of Missouri–Kansas City, the University of Texas at Dallas, and Comilla University compared several forecasting models using data from January 2022 to December 2023 across five divisions. They found that an algorithm called XGBoost produced the most accurate results, outperforming traditional statistical models such as SARIMA and other machine learning methods.
The study identified temperature as the strongest predictor of dengue incidence, followed by humidity and rainfall. Because mosquito breeding and viral transmission intensify in warmer conditions, researchers said temperature-based early warning systems could allow authorities to act before outbreaks escalate.
“Advanced machine learning, especially XGBoost, can deliver reliable forecasts even with limited surveillance data,” the authors wrote. They said such models could strengthen public health planning in resource-constrained settings like Bangladesh, enabling hospitals and local governments to prepare for seasonal surges.