摘要
目的:构建非轻症急性胰腺炎（NMAP）患者死亡风险的列线图预测模型，并验证其预测效能和临床应用价值，同时分析其对于其他评分系统的优势。方法:纳入大型重症监护数据库MIMIC-Ⅲ中的606例NMAP患者临床资料，按7∶3比例随机分为训练集和验证集。采用LASSO-Cox回归分析构建NMAP患者死亡风险列线图预测模型，并通过受试者工作特征（ROC）曲线、校准曲线以及决策曲线分析（DCA）对列线图模型进行评估。然后比较列线图模型与急性胰腺炎严重程度床边指数（BISAP）、序贯器官衰竭评分（SOFA）、快速序贯器官功能衰竭评分（qSOFA）、急性生理评分Ⅲ（APS Ⅲ）及牛津急性疾病严重程度评分（OASIS）对NMAP患者死亡风险的预测效能。结果:LASSO-Cox回归分析结果表明，年龄以及入院24 h内的收缩压、红细胞分布宽度（RDW）、血清白蛋白、尿素氮（BUN）、总胆红素和国际标准化比值（INR）是与NMAP患者死亡风险相关的独立危险因素（P<0.05），以此建立的列线图预后模型预测训练集NMAP患者14、30、60、90 d内死亡风险的ROC曲线下面积（AUC）分别为0.76（95% CI：0.67~0.83）、0.79（95% CI：0.72~0.83）、0.83（95% CI：0.77~0.87）、0.83（95% CI：0.78~0.88），验证集对应AUC分别为0.85（95% CI：0.76~0.94）、0.83（95% CI：0.76~0.91）、0.86（95% CI：0.79~0.93）、0.87（95% CI：0.81~0.93）。校准曲线显示训练集和验证集模型预测概率与实际生存率之间均具有较好的一致性，DCA曲线显示列线图模型阈值在0.2~0.8时具有明显的临床净获益。时依ROC曲线显示列线图模型的预测效能优于BISAP、SOFA、qSOFA、APSⅢ及OASIS评分（均P<0.05）。结论:基于年龄以及入院24 h内的收缩压、RDW、血清白蛋白、BUN、总胆红素和INR建立的列线图预测模型简单便捷，可早期评估NMAP患者的死亡风险，且准确性较高。
Abstract
Objective:To construct a nomogram prediction model for early prediction of mortality risk in patients with non-mild acute pancreatitis (NMAP) and analyze its clinical application effect and advantages over other scoring systems.Methods:Clinical data of 606 patients with NMAP from the large medical information mart for intensive careⅢ database (MIMIC-Ⅲ ) were selected. The patients were randomly divided into training and validation sets in a 7∶3 ratio. LASSO-Cox regression analysis was performed to construct a nomogram prediction model for mortality risk in NMAP patients. The model’s performance was assessed through receiver operating characteristic (ROC) curve, calibration curves, and decision curve analysis (DCA). Additionally, the predictive efficacy of the nomogram model was compared with BISAP, SOFA, qSOFA, APS Ⅲ, and OASIS scores.Results:LASSO-Cox regression analysis identified age, systolic blood pressure within 24 hours of admission, red blood cell distribution width (RDW), serum albumin, blood urea nitrogen (BUN), total bilirubin, and international normalized ratio (INR) as independent risk factors associated with mortality in NMAP patients (P<0.05). A nomogram prognostic model was developed based on these factors. The area under the curve (AUC) for the nomogram model was 0.76 (95% CI: 0.67-0.83), 0.79 (95% CI: 0.72-0.83), 0.83 (95% CI: 0.77-0.87), and 0.83 (95% CI: 0.78-0.88), respectively, for predicting mortality at 14, 30, 60, and 90 days in NMAP patients. The validation set demonstrated AUC values of 0.85 (95% CI: 0.76-0.94), 0.83 (95% CI: 0.76-0.91), 0.86 (95% CI: 0.79-0.93), and 0.87 (95% CI: 0.81-0.93), respectively. Calibration curves indicated excellent agreement between predicted and observed probabilities of mortality in both the training and validation sets. The DCA curve indicated that the nomogram had significantly positive net benefit when the threshold probability ranged from approximately 0.2 to 0.8. The ROC curve revealed superior prediction efficiency of the nomogram model compared to BISAP, SOFA, qSOFA, APSⅢ, and OASIS scores (P<0.05).Conclusion:The nomogram model, incorporating age, systolic blood pressure within 24 hours of admission, RDW, serum albumin, BUN, total bilirubin, and INR offers a simple and convenient tool for accurate prediction of death risk in NMAP patients early.
万方数据