This study utilizes Earth observation data to predict infectious disease incidence, focusing on malaria in Nigeria—a country with a high burden of transmission. The dataset, derived from satellite imagery, includes temperature, normalized difference vegetation index (NDVI), rainfall, nighttime lights, and distance to water bodies, covering the years 2000 to 2017. Malaria incidence rates were obtained from the Demographic Health Surveys (DHS). Two deep learning models—Long Short-Term Memory (LSTM) and Transformer—were trained on data from 2000 to 2016 and internally evaluated on 2017 data. The results indicate that the LSTM model achieved superior predictive performance, with a lowest root mean squared error (RMSE) of 1.0%, compared to 5.0% for the best-performing Transformer configuration. These findings underscore the importance of temporal modeling and feature integration for accurate disease prediction. The proposed methodology is generalizable and may support data-driven decision-making for public health interventions in other malaria-endemic regions. All data and code will be made publicly available upon acceptance of the paper.