IIT–AIIMS Jodhpur Team Uses AI to Revolutionize Childhood Malnutrition Assessment

New Delhi – A collaborative research team from the Indian Institute of Technology (IIT) Jodhpur and the All India Institute of Medical Sciences (AIIMS) Jodhpur has developed an artificial intelligence (AI)–based method to more effectively detect malnutrition in children, one of the world’s most urgent public health challenges.

The study, published in the open-access journal MICCAI, introduces DomainAdapt, an innovative multitask learning framework. Unlike traditional screening tools that rely on manual, time-intensive measurements, DomainAdapt dynamically adjusts task weights using domain knowledge and mutual information. This enables the AI system to estimate key anthropometric indicators such as height, weight, and mid-upper arm circumference (MUAC), while also classifying conditions including stunting, wasting, and underweight.

“By simply analyzing a child’s photo, our framework can provide a nutritional assessment without the need for complex anthropometric measurements,” said Misaal Khan, a doctoral student in medical technology at IIT–AIIMS and lead author of the study. “This approach makes malnutrition screening faster, more accessible, and scalable—particularly in resource-limited settings.”

A critical element of the research is AnthroVision, a pioneering dataset comprising 16,938 multi-pose images of 2,141 children, collected from both clinical settings at AIIMS Jodhpur and community settings such as government schools in Rajasthan. With its diversity of backgrounds, clothing, and lighting, AnthroVision serves as a robust foundation for advancing AI-driven child health assessment.

In extensive testing, DomainAdapt outperformed existing multitask learning models, showing significant improvements in predicting nutritional status and offering a cost-effective tool for global health systems.

“This research marks a key step toward equitable healthcare access,” Khan emphasized. “By combining AI with domain expertise, we can equip healthcare workers with tools that are accurate, affordable, and truly scalable.”

With inputs from IANS