AI tools could spot hidden fatty liver disease before it becomes deadly

Researchers say artificial intelligence can scan electronic health records and routine scans to identify people with early-stage fatty liver, a condition affecting about 30% of adults worldwide.

More than a billion people worldwide carry excess fat in their livers, a silent condition that can progress to fibrosis, cirrhosis and heightened cardiovascular risk. Because symptoms rarely appear early, three-quarters of diagnoses occur only after life-threatening damage. Scientists including Jeffrey Lazarus and Jonathan Dranoff argue that AI can retrospectively sift through massive health-record databases and routine imaging to pinpoint those most likely to have dangerous liver fat levels.

By automating Fib-4 calculations and employing models that read chest x-rays, AI could alert primary-care doctors to refer patients for further testing or treatment with drugs such as semaglutide, resmetirom, or AI-guided therapies like ALADDIN. Start-ups such as Evido have already commercialized AI algorithms that outperform traditional scores in large cohorts, and partnerships with firms like Roche are expanding their reach. While still largely in research phases, these technologies could enable earlier intervention, improve patient adherence, and reduce costly liver transplants.