Introduction <p>Metabolic dysfunction-associated steatotic liver disease (MASLD) affects more than 30% of the global population. It is particularly prevalent among individuals with severe obesity. MASLD progression leads to advanced stages of liver fibrosis, which are associated with increased morbidity and mortality. However, current tools for liver fibrosis assessment remain limited. We aimed to identify potential lipid biomarkers and predict significant liver fibrosis through lipidomic analysis.</p> Methods <p>This study included patients with severe obesity who underwent bariatric surgery between 2016 and 2020 at Taipei Medical University Hospital. Preoperative liver samples and both pre- and postoperative serum samples were subjected to mass spectrometry. Statistical analyses were performed to compare lipidomic data among three histological groups and between pre- and postoperative states. Five machine learning models were used to predict significant liver fibrosis on the basis of patient characteristics and preoperative serum lipidome.</p> Results <p>The study cohort comprised patients with normal liver histology (<i>n</i> = 10), metabolic dysfunction-associated steatohepatitis with no or mild liver fibrosis (<i>n</i> = 19), and metabolic dysfunction-associated steatohepatitis with significant liver fibrosis (<i>n</i> = 21). Diradylglycerol 34:1 and triradylglycerol 52:3 emerged as promising biomarkers of significant liver fibrosis. The machine learning models exhibited excellent predictive performance (area under the curve: 0.942–0.983). Key lipid features included phosphatidylinositol 36:2, diradylglycerol 42:2, and phosphatidylcholine O-38:0.</p> Conclusion <p>Our findings indicate serum lipidome as a strong predictor of significant liver fibrosis. The identified lipid biomarkers may potentially be incorporated into current noninvasive clinical tools for patients with severe obesity and MASLD undergoing bariatric surgery.</p>