The Rhode Island Foundation is directing nearly $650,000 toward 26 medical research projects across the state, as part of an effort to help early-career researchers build a track record so they can go after larger federal dollars.
The grants, announced Monday and capped at $25,000 each, will support a wide range of medical research, from improving patient adherence to GLP-1 weight loss medications to training artificial intelligence to more accurately diagnose breast cancer.
Recipients include researchers at the University of Rhode Island, Miriam Hospital, Rhode Island Hospital, Providence College and Johnson & Wales University.
“Through the generosity of our donors, we are able to provide the crucial seed funding that enables local researchers to pursue promising medical advances,” said David N. Cicilline, the Foundation’s president and CEO.
“Although the grants are fairly modest, they can lead to big discoveries that will spark substantial new investments in the state’s research sector as well as create healthy communities across our state,” Cicilline said in a press release announcing the grants.
The funded projects include work by Dr. Alina Jade Barnett, an assistant professor of computer science and statistics at URI, who is developing AI technology designed to help radiologists diagnose breast cancer.
Barnett’s project addresses a subtle but serious problem with existing AI tools: they sometimes arrive at the right answer for the wrong reasons. She says an AI might correctly flag a patient as having breast cancer, but base that conclusion on the patient’s age and medical history, rather than on the appearance of the tissue itself.
“AI models are very powerful predictors but are also prone to hidden errors,” Barnett said. “One way we can make these safer is to design AI that can explain its reasoning. If we know how the AI makes its decision, we can see when it’s making a mistake.”
Her project goes one step further — not just identifying those flawed reasoning patterns, but correcting them, a process she calls “model steering.”
Barnett’s work will draw on a cohort of 400 previous patients and their prior diagnoses. About $20,000 of the $25,000 grant will go toward supporting a PhD student who will handle the primary coding, model development and data analysis.
Barnett is careful to frame her technology as a supplement to medical expertise, not a replacement for it. “It’s designed to assist a radiologist,” she said. “It works as a medical collaborator, not as a replacement for high quality medical care.”
She also sees the grant as a foundation for something larger. As a first-year professor at URI, she hopes the preliminary results this funding makes possible will position her lab to compete for federal grants from agencies like the National Institutes of Health and the National Science Foundation, both of which she says have interest in this type of work.
“This work would not be possible without the Rhode Island Foundation’s support,” Barnett said. “It’s really important to get these funding streams up and running to be able to start these projects and show that now at the University of Rhode Island, we can do this sort of advanced AI work.”
Other funded projects include a URI study on postpartum depression and suicide risk in new mothers; a URI pilot testing non-invasive brain stimulation for teens with lingering concussion symptoms; a Miriam Hospital investigation into why most patients abandon GLP-1 weight loss medications within a year; and a Johnson & Wales project to build a diabetes prevention program designed around the specific needs of Latina adolescents.
A panel of scientists and physicians helped the Foundation evaluate the proposals.
