Malaria outbreak prediction at the sub-district level in Zambia using remote sensing satellite data

Area of Science:

• Epidemiology
• Environmental Science
• Public Health

Background:

• Seasonal malaria surges strain health systems, causing shortages of diagnostics and antimalarial drugs.
• Accurate outbreak forecasts are crucial for efficient health resource planning.

Purpose of the Study:

• To develop and validate a predictive model for malaria outbreaks in a low-transmission area of southern Zambia.
• Utilize remote sensing data for temperature and rainfall prediction.

Main Methods:

• Collected weekly malaria case data from health facilities over 15 years in Choma District, Zambia.
• Used Climate Hazards Group InfraRed Precipitation with Station data for rainfall and MODIS/Terra satellite for land surface temperature.
• Employed lagged correlation analysis and negative binomial regression, training on 2010-2016 data and validating from 2017-2024.

Main Results:

• Mean nighttime temperature (Nov-Jan) and mean daily rainfall (Dec) were optimal predictors.
• Malaria cases significantly correlated with higher nighttime temperatures and increased precipitation.
• The model accurately predicted the 2020 outbreak, with 4% difference from observed cases, using data from four months prior.

Conclusions:

• A simple, weather-driven model accurately forecasts malaria outbreaks up to four months ahead in low-transmission settings.
• Predictions can guide targeted stock management and resource mobilization, mitigating shortages during outbreaks.