Development of a predictive model for severe hyperlipidemic acute pancreatitis based on LASSO regression: a...

In recent years, the incidence of hyperlipidemic acute pancreatitis (HLAP) has been increasing. Identifying the risk factors associated with severe HLAP and developing a predictive model are crucial for early detection and intervention, thereby alleviating the disease burden. This study aimed to investigate the risk factors associated with severe HLAP and develop a predictive model.

Data on HLAP treated in Taixing People's Hospital Affiliated to Yangzhou University from January 1, 2020, to June 30, 2023, were retrospectively collected and divided into a mild group (N = 296) and a moderate severe/severe group (N = 60). Univariate analysis and Least Absolute Shrinkage and Selection Operator (LASSO) regression were used to select variables, and the selected variables were incorporated into logistic regression to analyze the risk factors of severe disease. A logistic regression model was constructed. The receiver operating characteristic (ROC) curve and the area under the curve (AUC) were used to evaluate model differentiation, and the Hosmer-Lemeshow goodness-of-fit test and calibration curve were used to evaluate model consistency.

The univariate analysis showed statistically significant differences in 50 variables between the mild and moderately severe/severe groups. LASSO regression identified the following variables: D-dimer, blood calcium, cholesterol, standard bicarbonate (SB), total carbon dioxide, and C-reactive protein-albumin ratio (CAR). The constructed logistic regression model included D-dimer, blood calcium, and cholesterol, with an AUC of 0.8341 (95% CI [0.7724-0.8958]). The model's calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test (χ ² = 6.8383, P = 0.5542), and the calibration curve demonstrated that the model's predictions closely aligned with observed outcomes.

The risk factors of severe HLAP include D-dimer elevation, calcium depletion and cholesterol elevation. The predictive model established by logistic regression has good performance, which is helpful for early identification and intervention by clinicians.