A NEW study presented at The Liver Meeting by the American Association for the Study of Liver Diseases in California, USA, reveals that artificial intelligence (AI) can identify early-stage metabolic-associated steatotic liver disease (MASLD) using electronic health records (EHRs). This breakthrough highlights the potential of AI to improve the early detection of a condition that often goes unnoticed until it progresses to more severe stages.
MASLD, the most common form of liver disease, affects approximately 4.5 million adults in the United States. It occurs when fat is not effectively processed in the liver and is frequently linked to conditions such as obesity, Type-2 diabetes, and abnormal cholesterol levels. While treatable in its early stages, MASLD often remains undiagnosed because many patients show no symptoms until the disease has advanced.
Researchers applied an AI algorithm to analyse imaging data from EHRs across three sites within the University of Washington Medical System. Of the 834 patients identified as meeting MASLD criteria, only 137 had an official diagnosis recorded. This indicates that 83% of cases were left undiagnosed, despite sufficient data being available in their medical records.
This study demonstrates the value of AI in addressing gaps in traditional diagnostic processes. By leveraging data that already exists in EHRs, AI algorithms can uncover patterns and conditions that might otherwise be overlooked. In the case of MASLD, early detection is critical to preventing disease progression to advanced liver damage, including cirrhosis or liver failure.
As MASLD often coexists with other health conditions, such as metabolic syndrome, early diagnosis could also provide an opportunity to manage associated diseases more effectively. The integration of AI into clinical workflows may offer a transformative approach to managing not only MASLD but a range of underdiagnosed conditions, potentially improving outcomes for millions of patients.
Reference
Stuart A et al. AI finds undiagnosed liver disease in early stages. Abstract 2360. AASLD Annual Meeting, 15-19 November 2024.
