Machine Learning Tool Improves Noninvasive Identification of F2 Fibrosis in MASH
A new machine learning–based calculator may improve identification of patients with metabolic dysfunction–associated steatohepatitis (MASH) who meet fibrosis thresholds for resmetirom therapy, according to results from the ALADDIN study. The tool was developed to address the lack of a vibration-controlled transient elastography (VCTE)–based algorithm specifically targeting significant fibrosis (≥F2).
Investigators analyzed biopsy-confirmed metabolic dysfunction–associated steatotic liver disease across multiple centers, using a training cohort of 827 patients, a test cohort of 504 patients, and an external validation cohort of 1,299 patients. Five machine learning algorithms were evaluated, with the top 3—random forest, gradient boosting machines, and XGBoost—combined into an ensemble model.
In external validation, the ALADDIN-F2-VCTE model, which integrates routine laboratory parameters with VCTE, achieved an area under the curve (AUC) of 0.791. This performance exceeded VCTE alone (AUC 0.745), FibroScan–aspartate aminotransferase score (0.710), and the Agile-3 model (0.740), with significant differences in discrimination, calibration, and decision curve analyses. The authors noted that ALADDIN-F2-VCTE “uniquely supports a refined noninvasive approach to patient selection for resmetirom without the need for liver biopsy.”
A laboratory-only version, ALADDIN-F2-Lab, which does not require VCTE, achieved an AUC of 0.706 and outperformed Fibrosis-4, steatosis-associated fibrosis estimator, and LiverRisk scores. The authors stated that ALADDIN-F2-Lab “offers an effective alternative when VCTE is unavailable.”
For gastroenterologists and hepatologists, the key takeaway is that accurate noninvasive identification of F2 fibrosis is increasingly important with the approval of resmetirom. The ALADDIN models may enhance patient selection using routinely available data, potentially reducing reliance on biopsy while improving precision in therapeutic decision-making. Further implementation studies may clarify how these tools perform in community settings and integrate into routine clinical workflows.
Reference
Alkhouri N, Cheuk-Fung Yip T, Castera L, et al. ALADDIN: A machine learning approach to enhance the prediction of significant fibrosis or higher in metabolic dysfunction-associated steatotic liver disease. Am J Gastroenterol. 2026;121(2):362-374. doi:10.14309/ajg.0000000000003432
