Abstract
BACKGROUND: Multiple Organ Dysfunction Syndrome (MODS) is a lethal complication of acute pancreatitis (AP), making early identification of high-risk patients crucial for improving outcomes. AIM: To develop and validate a clinically applicable nomogram for predicting MODS risk in acute pancreatitis patients based on clinical and laboratory variables collected within the first 24 hours of admission. METHODS: We conducted a single-center retrospective cohort study using routinely collected electronic health records from Shanxi Bethune Hospital (Taiyuan, China), including 693 adult patients with acute pancreatitis admitted between January 1, 2019 and December 31, 2021. From 29 candidate indicators obtained within 24 hours of admission, key predictors were selected via LASSO regression, and three machine-learning models - generalized linear model (GLM), random forest (RF), and support vector machine (SVM) - were constructed and compared. RESULTS: Among the candidate models, the GLM-based nomogram showed the best overall performance. Using a parsimonious variable selection strategy, we derived a final 7-variable model. This model demonstrated good discrimination, with areas under the curve of 0.829 in the training cohort and 0.846 in the validation cohort, and showed satisfactory calibration, a high negative predictive value, and clinical net benefit. CONCLUSION: Our internally validated 7-variable nomogram, based on a generalized linear model, showed good discrimination and a high negative predictive value for early prediction of multiple organ dysfunction syndrome in patients with acute pancreatitis, and may assist clinicians in early risk stratification and resource allocation. Further external validation and prospective studies are warranted to confirm its generalizability and clarify its impact on clinical decision-making and patient outcomes.