Meta Partners With Five US Universities to Accelerate Medical and Scientific Breakthroughs
Meta partners with five US universities to fund $2 million in research using its AI models for medical and scientific breakthroughs, including infant tracheostomy monitoring and dementia care.
Meta announced partnerships with California Institute of Technology, The Ohio State University College of Medicine, Louisiana State University, Indiana University, and UTHealth Houston to accelerate research using its AI models. The collaboration includes $2 million in funding to support projects in medicine, physics, chemistry, biology, computer science, and mathematics. Researchers will focus on leveraging Meta’s models, glasses, and mixed reality headsets to address gaps in healthcare and scientific discovery. The initiative aims to bridge academia and industry to drive practical applications of AI in real-world settings.
Dr. Michael Dunham at Louisiana State University is leading the Eyes-On project, which uses computer vision to detect tracheostomy tube displacement in infants—a life-threatening emergency—where continuous monitoring is challenging. The system employs Meta’s open-weight SAM models for tracking and integrates with bedside cameras and edge computing. A proof-of-concept has been published, with further development planned.
Dr. David Crandall at Indiana University is developing AI methods to analyze conversations involving older adults with dementia, aiming to improve conversational agents and social robots. The research focuses on adapting interactions to individual users with cognitive differences, addressing limitations of standard conversational AI systems. The project tests whether AI can become sensitive to unique communication patterns in dementia care.
At The Ohio State University College of Medicine, Dr. Yuan Xue, Dr. Yuchi Han, and Dr. Xia Ning are investigating how Meta’s open-weight models can predict heart failure with preserved ejection fraction (HFpEF) risk using abdominal CT scans, medical records, and ECG data. The team designs the model to handle missing data, reflecting real-world clinical practice where complete inputs are rare. The project seeks to enable earlier detection of HFpEF, a condition often diagnosed late due to overlapping symptoms.