Insights & Publications
The AI for Humanity Lab produces scholarly articles, reports, and thought leadership pieces that advance the responsible development and governance of artificial intelligence.
Advancing Malaria Prediction in Uganda through AI and Geospatial Analysis Models
Authors: Maria A. Komugabe | R. Caballero | Itamar Shabtai | S.P. Musinguzi
Published: Journal of Geographic Information System, Vol. 16, pp. 115–135, April 2024
This study explores how integrating Gregor’s Type IV theory with Geographic Information Systems (GIS) improves our understanding of malaria transmission patterns in Uganda. By combining data-driven algorithms, artificial intelligence, and geospatial analysis, the research identifies the most reliable predictors of malaria incident rates and assesses the impact of climate and preventive factors on transmission.
Key Finding: The Random Forest model outperformed all other models with an R² of approximately 0.88. Antimalarial treatment was identified as the most influential factor, while mosquito net access was associated with significant reduction in incident rates and higher temperatures correlated with increased rates.
Methods: Linear Regression, K-Nearest Neighbor, Neural Network, and Random Forest predictive modeling with geospatial analysis.
Geospatial and Path Analysis for Enhancing Malaria Control and Primary Healthcare Delivery in Low-Income Nations: A Case Study of Uganda
Authors: Maria Assumpta Komugabe | Richard Caballero | Itamar Shabtai | Zhaoxia Yi | Zachary Dodds
Published: American Journal of Epidemiology and Infectious Disease, Vol. 12, No. 3, pp. 44–54, September 2024
This study investigates the use of geospatial and path analysis to enhance malaria control and primary healthcare delivery in Uganda. Using generalized linear regression, ordinary least squares regression, and spatial autocorrelation (Moran’s I), the research identifies key factors influencing malaria incidence rates including mean temperature, antimalarial treatment, mosquito net access, total population, and number of health centers.
Key Finding: Mean temperature showed the strongest direct effect on malaria incidence (β = 0.658), while expanding health center networks in underserved regions was identified as critical to improving healthcare outcomes. Spatial autocorrelation revealed significant geographic clustering of malaria rates, highlighting the need for targeted interventions.
Methods: Generalized Linear Regression (GLR), Ordinary Least Squares (OLS), Path Analysis, Spatial Autocorrelation (Moran’s I), Bivariate Color Mapping.