A team of researchers from USC, Microsoft’s AI for Good Lab, Amref Health Africa, and Kenya’s Ministry of Health has developed an AI model that can forecast acute child malnutrition in Kenya up to six months in advance with 86% accuracy.

By combining clinical data from over 17,000 Kenyan health facilities with satellite imagery on crop health, the model outperforms traditional methods that rely on historical trends alone. It is especially effective in regions with unpredictable surges in malnutrition.

“Malnutrition is a public health emergency in Kenya. Children are dying unnecessarily,” said Laura Ferguson of USC’s Institute on Inequalities in Global Health.

Kenya’s current tools are based largely on expert judgment, making it difficult to predict emerging hotspots. The AI model, detailed in a PLOS One study out May 14, 2025, offers a more data-driven approach and has already been integrated into a prototype dashboard for early intervention.

“This is a game-changer,” said USC computer science professor Bistra Dilkina. “It captures complex relationships to predict more accurately.”

Roughly 350,000 children in Kenya under five suffer from acute malnutrition, with regional rates as high as 25%. With the AI framework built on widely used systems like DHIS2, researchers believe the model can be adapted globally to help fight malnutrition in other vulnerable countries.