There's a Fatty Liver Epidemic. AI Could Help Get Ahead of It
More than 1.66 billion people are living with excess fat in their liver, and researchers are betting machine learning can catch the disease before it becomes deadly.
Written by OutOfToken AI
August 13, 2026 · 4 min read · Synthesized from reporting by Wired · How this works
Fatty liver disease has quietly become one of the largest public health burdens on the planet, affecting well over a billion people worldwide. Most don't know they have it, because early-stage disease rarely produces symptoms. Now a growing coalition of researchers, hospitals, and startups is betting that AI can find the disease before it finds patients in an emergency room.
A Silent Pandemic, By the Numbers
The condition once known as nonalcoholic fatty liver disease, now clinically termed MASLD (metabolic dysfunction-associated steatotic liver disease), has been described by researchers as a 'silent' global pandemic. Its more severe form, MASH, involves inflammation and scarring that can progress to fibrosis, cirrhosis, or liver cancer. Because early stages are largely asymptomatic, the disease often isn't caught until significant damage has already occurred.
Machine Learning Enters the Clinic
Researchers have identified three areas where AI is already making measurable progress: imaging analysis, risk stratification, and biomarker discovery. Deep learning models trained on medical scans are improving the accuracy of non-invasive diagnosis, reducing reliance on invasive liver biopsies. Machine learning models are also being used to stratify patient risk with more precision than traditional screening methods allow.
Repurposing Scans You Already Have
One of the more striking applications comes from Nanox.AI, which built an algorithm that analyzes liver density from abdominal and chest CT scans, imaging many patients already undergo for unrelated reasons. Low liver density on these scans can flag fatty liver disease that would otherwise go unnoticed. The approach effectively turns routine imaging into an opportunistic screening tool, catching disease in patients who were never being tested for it in the first place.
"MASLD affects an estimated 1.66 billion people globally, according to the Fatty Liver Foundation — and most cases go undiagnosed until significant liver damage has occurred."
Patients, Records, and the NLP Angle
Detection isn't limited to imaging. At the University of Washington, a research team led by Dr. Rotonya Carr and resident researcher Dr. Ariana Stuart has been testing an iterative natural language processing algorithm designed to identify MASLD diagnoses buried in clinical notes and medical records. That kind of tool could help health systems retroactively surface patients who already show signs of the disease in their charts but were never formally flagged. Meanwhile, the Fatty Liver Foundation has partnered with Secure AI Labs on a patient-facing risk stratification tool, aimed at people who may never set foot in a specialist's office. The tool uses machine learning to help individuals assess their own risk and understand what lifestyle changes might slow or halt disease progression, part of a broader push to find asymptomatic, undiagnosed patients before their disease advances.
The Obstacles Are Real
None of this is a finished product. Researchers studying AI's role in liver disease point to persistent challenges, including data heterogeneity across hospitals and populations, and bottlenecks in translating promising algorithms into actual clinical workflows. Predicting how the disease will progress in any individual patient also remains difficult, since the path from simple fat accumulation to cirrhosis or cancer varies widely and depends on multiple metabolic risk factors.
The tools being built now, from CT-scan algorithms to patient-facing risk calculators, suggest a future where fatty liver disease is caught during a routine scan or a five-minute questionnaire rather than after a cirrhosis diagnosis. Whether that future arrives quickly depends on messier problems than the algorithms themselves: getting hospitals to adopt new workflows, standardizing data across health systems, and reaching the asymptomatic majority who have no reason to suspect anything is wrong. If researchers can clear those hurdles, AI may end up doing for liver disease what it has already started doing for cancer screening — turning a silent epidemic into a manageable one.
Editorial Note
The research sources comprehensively corroborate all major factual claims in the article: the global prevalence figures, the characterization of MASLD as a silent pandemic, disease progression pathways, and specific AI applications from companies like Nanox.AI and research institutions. Source 2 directly validates the three AI application areas mentioned (imaging analysis, risk stratification, biomarker discovery). The sources confirm the epidemiological claims and technical capabilities described.
Claim Tracker
AI-assessed
Source 1 states 'more than 1.66 billion people around the world' are affected by NAFLD/NASH. Source 3 indicates 'global prevalence of MASLD is 25%' which aligns with this magnitude.
Source 3 explicitly states it 'is considered a "silent" global pandemic.' Source 6 confirms 'This disease progresses silently.'
Source 4 states 'MASH represents the advanced inflammatory form...where fat accumulation in the liver triggers fibrosis and progressive liver injury.' Source 6 confirms progression 'to fibrosis, cirrhosis, and, in some cases, lead to liver cancer.'
Source 2 confirms 'deep learning-based imaging analysis has significantly enhanced non-invasive diagnostic accuracy for fatty liver disease.'
Source 3 states 'Nanox.AI's unique and pioneering algorithm highlights liver density measurements using abdominal and chest CTs. Low liver density is suggestive of fatty liver disease.'
Ask AI about this story
// discussion
sign in to join the discussion