Predicting the burden of acute malnutrition in drought-prone regions of Kenya: a statistical analysis
medRxiv
Download
Abstract: This study investigates the potential to predict acute malnutrition in Kenya’s arid and semi-arid regions using publicly available data. Covering 2015–2019, researchers analyzed 44,218 child records from SMART surveys and tested statistical models like random forests and generalized linear models. Key predictors included climate, food prices, health services, and conflict data. While models—especially for global acute malnutrition—showed moderate performance, none were highly accurate. The study concludes that improved, broader datasets may enhance predictive capability, offering a cost-effective tool to supplement ground surveys and support timely, targeted responses to nutrition crises worsened by climate-related droughts.