AI Phenotyping Links Cardiovascular, Kidney Disease to Rapid MASH Progression, Higher Costs

An artificial intelligence (AI) phenotyping model built on real-world data from more than 14,000 patients with metabolic dysfunction-associated steatohepatitis (MASH) found that cardiovascular and kidney diagnoses were the characteristics most consistently linked to rapid liver scarring, major long-term complications, and the highest health care costs, according to a study published in Therapeutic Advances in Gastroenterology.¹

AI Analysis Reveals Predictors of Fibrosis Progression, High Costs

MASH, the more severe and inflammatory form of metabolic dysfunction-associated steatotic liver disease (MASLD), develops in roughly a quarter of the estimated 30% of adults worldwide who have MASLD. About 1 in 5 of these patients go on to experience rapid fibrosis progression, but the underlying drivers have remained unclear, making it difficult to identify those in need of closer monitoring.

Liver biopsy remains the standard for staging fibrosis but is invasive, costly, and impractical for widespread use. As a result, clinical practices are looking toward noninvasive tools such as the Fibrosis-4 index (FIB-4). That score, however, has documented accuracy limitations in patients both older than 65 years and younger than 35 years, as well as those with obesity or diabetes, leaving a gap in reliable early identification. Consequently, investigators conducted a study using AI phenotyping to identify factors associated with rapid fibrosis progression, long-term clinical outcomes, and high health care costs.

They used the OM1 Real-World Data Cloud, an electronic health record and claims dataset covering more than 340 million people in the US, to identify 14,707 patients with a MASH diagnosis and at least 1 FIB-4 score within 90 days of diagnosis, spanning January 2013 to September 2022. Based on OM1's PhenOM AI platform, diagnoses, procedures, medications, lab values, and demographic data from the 12 months before diagnosis, the investigators grouped patients into thematically related "phenotypic signals," which were then tested against 3 outcomes: rapid fibrosis progression, long-term liver- and cardiovascular-related clinical outcomes, and health care costs in the top 10% of the cohort.

Cardiometabolic Phenotypes Associated With Worse MASH Outcomes

Among 1795 patients evaluated for fibrosis trajectory, a history of anemia and thrombocytopenia, including iron deficiency anemia and general thrombocytopenia, showed the largest absolute difference between patients who progressed rapidly and those who did not. Cardiovascular diagnoses, including atrial fibrillation, coronary artery disease, and congestive heart failure, showed the strongest relative association. Heart failure–related procedures, a history of hospitalization, and kidney diagnoses such as acute kidney failure and chronic kidney disease (CKD) also distinguished patients with rapid progression.

Among 13,880 patients assessed for longer-term outcomes, chronic diseases including diabetes and hypertension, kidney diagnoses, cardiac diagnoses, and inpatient hospitalization were associated with the development of cirrhosis or other liver-related complications. In a separate analysis limited to patients without prior cardiovascular disease, cardiac diagnoses, kidney diagnoses, and chronic conditions such as diabetes were linked to subsequent heart failure, myocardial infarction, stroke, or unstable angina.

In addition, among 10,133 patients with sufficient claims data, the strongest cost signal combined heart failure–related procedure codes with infection and sepsis procedure codes, as well as cardiac diagnoses and diagnostic testing. Hospitalization procedure codes paired with kidney diagnoses, as well as gastrointestinal diagnoses combined with abdominal imaging, also distinguished the top 10% of patients by cost.

Across all 3 analyses, cardiovascular and kidney-related diagnoses consistently emerged as the strongest phenotypic signals, a pattern the authors said may reflect shared risk factors between cardiovascular disease and MASH, as well as the bidirectional relationship between type 2 diabetes and MASH that also raises CKD risk.

The study’s findings build on prior research showing that MASH's economic burden is driven primarily by disease complications rather than medication costs and that untreated MASH with type 2 diabetes can push costs up 5-fold.² Per-patient MASH costs can exceed $32,000 over a lifetime, with each 1-unit rise in FIB-4 corresponding to a 3.4% increase in mean annual costs.³

Findings Support More Targeted Monitoring in MASH

At the same time, the authors acknowledged several of their study’s limitations, including the incomplete or inconsistent medical records that come with using real-world data. They also acknowledged the reliance on diagnosis codes rather than biopsy or imaging confirmation as a key limitation. With these in mind, the investigators suggested areas for further research.

“A better understanding of patient characteristics associated with rapid fibrosis progression, the development of long-term clinical outcomes, and high health care costs could enable the identification of patients requiring more regular monitoring and suggest new approaches for management,” they concluded.

References

1. Mangla KK, Gbadamosi SO, Semeniuta D, et al. Using artificial intelligence to identify characteristics associated with clinical and economic outcomes in MASH (FOCUS-MASH). Ther Adv Gastroenterol. 2026;19:17562848261462398. doi:10.1177/17562848261462398
2. Joszt L, Younossi Z. Untreated MASH with T2D to drive costs 5-fold: Zobair Younossi, MD. AJMC^®. May 28, 2026. Accessed July 17, 2026. https://www.ajmc.com/view/untreated-mash-with-t2d-to-drive-costs-5-fold-zobair-younossi-md
3. Management strategies for metabolic dysfunction–associated steatotic liver disease (MASLD). AJMC. November 7, 2024. Accessed July 17, 2026. https://www.ajmc.com/view/management-strategies-for-metabolic-dysfunction-associated-steatotic-liver-disease-masld-