A genetic algorithm for identifying spatially-varying environmental drivers in a malaria time series model

Area of Science:

• Epidemiology
• Environmental Science
• Computational Biology

Background:

• Malaria forecasting models utilize environmental factors like temperature and precipitation.
• Malaria-environment relationships are context-dependent, varying by ecological and social factors.

Purpose of the Study:

• To optimize a spatiotemporal malaria model by identifying clusters with similar environmental sensitivities.
• To improve the fit of environmentally-driven disease models through spatial stratification.

Main Methods:

• A genetic algorithm was employed to aggregate districts into clusters based on environmental sensitivities.
• The model was tested using seven years of weekly Plasmodium falciparum data and remote sensing data in Ethiopia's Amhara Region.

Main Results:

• The genetic algorithm identified six distinct clusters of districts.
• Districts within each cluster exhibited unique responses to environmental predictors such as temperature and precipitation.

Conclusions:

• Spatial stratification significantly enhances the performance of environmentally-driven malaria models.
• Genetic algorithms offer an effective method for identifying spatial clusters to improve disease modeling.