Accelerating progress toward Sustainable Development Goal 2 (SDG 2)—ending global hunger and undernutrition—requires access to granular, low-cost, and near-real-time data on children’s nutritional status. However, current
data collection efforts are often fragmented, infrequent, costly, and inaccessible, particularly during crises or in remote areas, limiting their effectiveness in driving timely intervention and targeted support. The Artificial Intelligence for Monitoring Malnutrition (AIMM) project introduces a low-cost, household-operated tool for monitoring undernutrition using image-based AI for real-time nutrition classification—without the need for physical scales or measuring tapes.
Implemented in the Indian state of Maharashtra in collaboration with the Center for Artificial Intelligence (CAI) at FLAME University in Pune, India, AIMM has the following key objectives:
- Enhance accuracy and predictive capacity: Refine deep learning models that estimate key anthropometric indicators—such as weight, height, MUAC, and weight-for-height z-scores (WHZ)—from 2D and infrared (IR) images. The use of IR imaging further improves measurement accuracy while maintaining privacy.
- Provide a low-cost, real-time alternative: Overcome the high costs and logistical barriers of traditional pen-and-paper surveys and existing AI tools. AIMM offers a breakthrough in child nutrition monitoring, especially in hard-to-reach areas or during crises, by prioritizing accessibility, reducing costs, and promoting long-term adoption.
- Empower caregivers: Enable caregivers, particularly in rural and under-served areas, to self-collect and monitor children’s nutritional status using a non-invasive, user-friendly, AI-powered mobile application. This will improve monitoring and facilitate timely interventions.
- Increase adaptability: With its open-source design, AIMM will foster uptake across diverse contexts and facilitate integration into global, regional, and national health systems.
By harnessing AI to democratize the collection of nutrition data, AIMM not only advances more effective and inclusive global health governance but also offers a blueprint/case for integrating granular, near real-time data into the monitoring of a range of SDGs and as international early warning systems in a variety of domains. The project empowers communities to generate evidence from below, enhancing local agency, accountability, and trust in health systems. At the global level, AIMM offers a blueprint for integrating near real-time data into international early-warning systems, fostering transparent decision-making in nutrition programming, food aid allocation, and Sustainable Development Goal monitoring.
AIMM builds on previous SNSF- and FCDO-funded projects that resulted in the following publications:
- Bhavnani, R. and N. Link. (2025). “Autonomous Nutrition Reporting by Households: Piloting an Open-Source Smartphone App in Rural Kenya.” Under Review
- Bhavnani et al. (2023). “Trajectories of Resilience to child undernutrition in the Kenyan Drylands.” Frontiers in Sustainable Food Systems, 7 (July).
- Bhavnani et al. (2023). “Household Behavior and Vulnerability to child undernutrition in Kenya.” Nature, Humanities & Social Sciences Communications, 10: 63 (February).
- MERIAM. (2021). “Simulating child undernutrition Toolkit (SAMT).” Interactive Online Dashboard for the MERIAM Project.
Timeline: January 2026 - December 2029.
Funding organisation:
