AI Tools Target Fatty Liver in 30% of Adults, Flagging Disease at 82% Accuracy
Updated
Updated · WIRED · Aug 13
AI Tools Target Fatty Liver in 30% of Adults, Flagging Disease at 82% Accuracy
3 articles · Updated · WIRED · Aug 13
Summary
More than 1 billion people may benefit from new AI screening efforts as researchers and startups build tools to detect fatty liver disease earlier from routine records, blood tests and even chest X-rays.
82% accuracy was reported by a Japanese model that found fatty liver on routine chest X-rays, while clinicians say AI could also automatically calculate Fib-4 scores from standard bloodwork and flag high-risk patients for referral.
470,000 middle-aged people were included in testing of Evido’s LiverPRO, commercialized with Roche, which outperformed the widely used Fib-4 score in predicting serious liver risk; another model, ALADDIN, also beat Fib-4 in identifying candidates for resmetirom.
75% of cirrhosis cases are diagnosed only when life-threatening, even though early liver damage can often be reversed through weight loss, lower alcohol intake, exercise and some drug treatments, making earlier detection both a clinical and cost-saving priority.