What began as a student’s exploration with data has since turned into a tool that may predict patients’ risk of advanced liver fibrosis.
Eri Osta, a radiology resident at the Dell Medical School, is pioneering HepaticMD, an AI-based tool that uses data from routine labs, like blood tests, to warn patients with metabolic dysfunction-associated steatotic liver disease who are at risk of developing advanced liver damage, according to a Dell Med announcement.
Osta, who is currently prototyping HepaticMD, said the tool will be able to be implemented onto electronic medical records, which have patient’s health information, where it can flag patients who may be at risk. For him, the tool offers a low-cost, preemptive solution.
“What is the value of a tool that just lives in a paper?” Osta said. “Let’s put this tool to work. Let’s actually think, ‘What do we need to do to bring this kind of technology into the hands of the people who need it?’”
Jack Virostko, director of research for the Diagnostic Radiology Residency and Osta’s principal investigator, said a strength of HepaticMD’s AI-driven model is its transparency. The tool’s ability to show which variables inform its predictions may mean physicians and patients will be more willing to use it, despite skepticism toward AI in health care, Virostko said.
“There’s always been this fear in medicine and in science in general that AI is this black box,” said Virostko, an associate professor of diagnostic medicine. “(Physicians) don’t want to be told, ‘Hey, I’m taking your job away from you. Just listen to me. I’m the black box.’ They want to understand how the model is coming to that decision.”
Nicole Clark, who leads operations at Texas Health Catalyst, a Dell Med program supporting researchers’ product development, said the program has seen a growing number of applications in recent years that use AI-driven tools.
“AI is here, I think, to stay,” Clark said. “Now, it’s about finding ways to leverage AI in an ethical and responsible way to help improve the outcomes of folks.”
HepaticMD also comes amid growing concern over increases in metabolic dysfunction-associated steatotic liver disease. A study published in January from the Journal of the American Medical Association found that liver disease rates are projected to rise from 33.7% in 2020 to 41.4% in 2050. Hirsh Trivedi, director of the Metabolic Liver Program at Dell Med, said detecting the liver disease early on may prevent further complications, including liver cancers, liver decompensation and liver failure.
“Picking up (the liver disease) early, identifying the stage at where it’s at promptlyand using that information to guide and inform subsequent care is extremely important to try to optimize outcomes for these patients,” Trivedi said.
Osta said HepaticMD represents an opportunity where clinicians and patients can learn to trust AI in the treatment process.
“I hope that (HepaticMD) starts a trend for holding AI tools accountable on what role they play on our clinical decision making,” Osta said. “Ultimately, that will build trust with the clinical user and hopefully patients.”
