Prediction of acute pancreatitis severity using a nomogram based on clinical features and body composition.

This study aimed to construct a new model based on quantitative computed tomography (QCT) body composition and clinical features for early prediction of acute pancreatitis (AP) severity. The clinical features and body composition of patients with clinical first-onset AP between January 1, 2024, and May 30, 2024, were analyzed. Concurrently, 100 healthy physical examination patients were included to collect the clinical characteristics and QCT parameters. AP was divided into mild AP (MAP, n = 66), moderate severe AP (MSAP, n = 18), and severe AP (SAP, n = 21), according to the revised Atlanta classification (RAC), subsequently, the patients were divided into the SAP (n = 21) and non-severe AP (NSAP; N = 84) groups. Clinical features and body composition parameters were used to determine risk factors for SAP using univariate and multivariate logistic regression methods. Efficacy was assessed using calibration curves, receiver operating characteristic (ROC) curves, and a decision curve analysis (DCA). A total of 105 patients with AP and 100 healthy individuals undergoing physical examinations were included in this study. Except for subcutaneous adipose tissue (SAT), all other body parameters showed statistically significant differences between the 2 groups (P < .05). Univariate and multivariate logistic regression analyses revealed that alcoholic etiology, C-reactive protein (CRP), total adipose tissue (TAT), skeletal muscle area (SMA) were independent predictive factors for SAP, and a model was derived. For the training cohort, the nomogram predicted SAP with area under the curve (AUC) of 0.87 (95% CI: 0.78-0.95), sensitivity of 0.80 (95% CI: 0.69-0.92), and specificity of 0.80 (95% CI: 0.64-0.96). For the validation cohort, the AUC was 0.81 (95% CI: 0.65-0.96), sensitivity was 0.56 (95% CI: 0.33-0.79), and specificity was 0.79 (95% CI: 0.57-1.00), indicating that the model had high discriminative power. The Hosmer-Lemeshow test P-value was .628, indicating that the nomogram performed well in calibration. Finally, the DCA demonstrated the clinical applicability of the model. The present study demonstrated that alcoholic etiology, CRP level, TAT, and SMA are independent risk factors for predicting SAP. The developed nomogram has good discrimination, calibration, and clinical applicability.