Predicting micronutrient deficiency with publicly available satellite data

Micronutrient deficiency (MND), which is a form of malnutrition that can have serious health consequences, difficult to diagnose in early stages without blood draws, are expensive and time-consuming collect process. It even more at public scale seeking identify regions higher risk MND. To provide data widely frequently, we propose an accurate, scalable, low-cost, interpretable regional-level MND prediction system. Specifically, our work the first use satellite data, such as forest cover, weather, presence water, predict micronutrients iron, Vitamin B12, A, directly from their biomarkers. We real-world, ground truth biomarker collected four different across Madagascar for training, demonstrate viable predicting MND, surprisingly exceeding performance baseline predictions based only on survey responses. Our method could be broadly applied other countries where available, potentially create high societal impact if these used by policy makers, officials, or healthcare providers.