AI algorithm accurately detects early-stage metabolic-associated steatotic liver disease

Liver disease, which is treatable when discovered early, often goes undetected until late stages, but a new study revealed that an algorithm fueled by artificial intelligence can accurately detect early-stage metabolic-associated steatotic liver disease (MASLD) by using electronic health records. The study was scheduled for presentation today at The Liver Meeting, hosted by the American Association for the Study of Liver Diseases.

A significant proportion of patients who meet criteria for MASLD go undiagnosed. This is concerning because delays in early diagnosis increase the likelihood of progression to advanced liver disease.”

Ariana Stuart MD, resident at University of Washington Internal Medicine Residency Program and lead author of the study

Researchers used an AI algorithm to analyze imaging findings in electronic health records from three sites within the University of Washington Medical System to identify patients who met the criteria for MASLD, the most common form of liver disease, affecting 4.5 million adults in the United States. While 834 patients met the criteria, only 137 actually had an official MASLD-associated diagnosis in their record. This left 83% of patients undiagnosed even when data in their electronic health record showed they met the criteria for MASLD.

“People should not interpret our findings as a lack of primary care training or management,” Stuart said. “Instead, our study shows how AI can complement physician workflow to address the limitations of traditional clinical practice.”

MASLD occurs when fat isn’t managed properly in the liver and is often associated with other common diseases such as obesity, Type-2 diabetes, and abnormal cholesterol levels. Early diagnosis of MASLD is key because it can quickly progress to more severe forms of liver disease, but many individuals in this early stage are asymptomatic, making diagnosis challenging.

Ariana Stuart, MD, will present the study, “Artificial Intelligence for Early MASLD Identification in the Electronic Medical Record,” abstract 2360, on Saturday, Nov. 16 at 8 a.m. PST.