Meta gives $2 million to five US universities for medical and science AI research

Meta is expanding its academic partnerships by awarding a combined $2 million to five universities in the United States. The funds will support research projects applying artificial intelligence models to fields including medicine, physics, chemistry, biology, computer science, and mathematics.

The institutions receiving support are the California Institute of Technology, The Ohio State University College of Medicine, Louisiana State University, Indiana University, and UTHealth Houston. Alongside funding, Meta is providing access to its models, smart glasses, and mixed reality headsets, paired with direct collaboration with company researchers.

Monitoring vulnerable infant patients

At Louisiana State University, researchers are developing a system to monitor infants with tracheostomies. When a tracheostomy tube shifts out of place, it creates an immediate life-threatening emergency. Because these infants cannot cry out, continuous oversight is vital but not always possible for hospital staff.

The project, named Eyes-On, uses Meta’s open-weight Segment Anything Model (SAM) alongside bedside cameras and edge computing. The goal is to track tube placement and spot signs of infant distress automatically. A proof-of-concept for the setup has already been published.

Dementia support and heart failure prediction

Indiana University researchers are studying conversational systems for older adults with dementia. Standard conversational tools often fail to handle the communication patterns of people with cognitive changes. The team is building multimodal systems that can adapt interactions to individual patients, with potential applications for social robots.

At The Ohio State University College of Medicine, another team is working to detect heart failure with preserved ejection fraction earlier. This condition makes up about half of all heart failure cases, but early diagnosis is difficult because symptoms resemble other conditions such as obesity. Researchers are using Meta’s open-weight models to assess risk by analyzing abdominal CT scans, patient records, and ECG data, while designing the tool to function even when medical records have missing information.