Assessing dengue fever risk in Costa Rica by using climate variables and machine learning techniques

TL;DR

This study uses climate data and machine learning models like GAMLSS and Random Forest to predict dengue fever risk in Costa Rica, aiding health officials in resource planning before outbreaks.

Contribution

It introduces a combined approach of climate variables and advanced machine learning techniques for dengue risk prediction in Costa Rica.

Findings

Reliable predictions of dengue cases achieved

Uncertainty in predictions quantified

Models can assist in resource allocation

Abstract

Dengue fever is a vector-borne disease mostly endemic to tropical and subtropical countries that affect millions every year and is considered a significant burden for public health. Its geographic distribution makes it highly sensitive to climate conditions. Here, we explore the effect of climate variables using the Generalized Additive Model for location, scale, and shape (GAMLSS) and Random Forest (RF) machine learning algorithms. Using the reported number of dengue cases, we obtained reliable predictions. The uncertainty of the predictions was also measured. These predictions will serve as input to health officials to further improve and optimize the allocation of resources prior to dengue outbreaks.