Predicting the dengue cluster outbreak dynamics in Yogyakarta, Indonesia: a modelling study

Area of Science:

• Epidemiology
• Public Health
• Spatial Analysis

Background:

• Dengue transmission is influenced by human mobility and climate, but their precise roles in local disease clustering are unclear.
• Understanding these factors is crucial for predicting and managing dengue outbreaks in urban environments.

Purpose of the Study:

• To investigate the impact of human mobility and weather conditions on dengue risk and spatiotemporal clustering in Yogyakarta, Indonesia.
• To evaluate the predictive performance of a Bayesian spatiotemporal model using different spatial structures.

Main Methods:

• A Bayesian spatiotemporal model was developed for neighborhood dengue outbreak prediction.
• Human mobility proxies from social media, weather data, and past dengue cases were used as predictors.
• Model performance was assessed using geographical proximity versus human mobility-based adjacency matrices.

Main Results:

• Incorporating human mobility proxies into the spatial covariance structure significantly improved model fit and prediction accuracy.
• The model demonstrated high accuracy in predicting dengue case clusters, with minimal deviations from observed data.
• Weather conditions did not provide additional predictive value for dengue clustering at this urban scale.

Conclusions:

• Social media-derived human mobility data, combined with disease surveillance, enables accurate prediction of within-city dengue dynamics.
• These findings support proactive and timely dengue outbreak management strategies.
• The study highlights the utility of novel data sources for epidemiological modeling.