Qiu Yanyu, Ji Guangnian, Wu Jinsheng
Department of General Surgery, The Affiliated Huaian No.1 People's Hospital of Nanjing Medical University, Huai'an, Jiangsu , China.
Obes Surg. 2025 Jul 25. doi: 10.1007/s11695-025-08100-3.
This study aims to comprehensively investigate the factors influencing weight reduction outcomes one year after bariatric surgery and construct a Nomogram.
A retrospective study analyzed 546 patients who underwent bariatric surgery at the bariatric center from 2015 to 2021. They were randomly divided into a 7:3 ratio for a training set (382 cases) and a testing set (164 cases). Univariate logistic regression and two machine learning techniques (LASSO, best subset regression) were employed for variable selection. The optimal model was derived via stepwise backward regression, Akaike Information Criterion (AIC), and Area Under the Curve (AUC). Receiver operating characteristic (ROC) curve analysis, calibration curve analysis, and Hosmer-Lemeshow test were employed to graphically evaluate and validate the performance of the model, while decision curve analysis (DCA) was utilized to assess its clinical value.
The predictive factors in the final nomogram included hip circumference, the surgical procedure and T2DM. Utilizing these three independent risk factors, a nomogram prediction model was developed, demonstrating robust discriminative ability with an area under the curve (AUC) of 0.742 (95% CI: 0.672-0.813) for the training set and 0.726 (95% CI: 0.607-0.845) for the test set. Furthermore, the model exhibited high accuracy, as evidenced by the non-significant Hosmer-Lemeshow test (P > 0.05) for both the validation and test sets. The decision curve analysis further confirmed the model's effectiveness in accurately predicting one-year weight loss outcomes following bariatric surgery.
The nomogram prediction model based on hip circumference, surgical procedure, and T2DM reasonably predicts one-year weight loss after bariatric surgery.
本研究旨在全面调查影响减重手术后一年体重减轻结果的因素，并构建列线图。
一项回顾性研究分析了2015年至2021年在减重中心接受减重手术的546例患者。他们以7:3的比例随机分为训练集（382例）和测试集（164例）。采用单因素逻辑回归和两种机器学习技术（LASSO、最佳子集回归）进行变量选择。通过逐步向后回归、赤池信息准则（AIC）和曲线下面积（AUC）得出最优模型。采用受试者工作特征（ROC）曲线分析、校准曲线分析和Hosmer-Lemeshow检验对模型性能进行图形化评估和验证，同时利用决策曲线分析（DCA）评估其临床价值。
最终列线图中的预测因素包括臀围、手术方式和2型糖尿病。利用这三个独立危险因素，开发了列线图预测模型，训练集的曲线下面积（AUC）为0.742（95%CI：0.672-0.813），测试集为0.726（95%CI：0.607-0.845），显示出强大的判别能力。此外，该模型表现出高准确性，验证集和测试集的Hosmer-Lemeshow检验均无统计学意义（P>0.05）。决策曲线分析进一步证实了该模型在准确预测减重手术后一年体重减轻结果方面的有效性。
基于臀围、手术方式和2型糖尿病的列线图预测模型能够合理预测减重手术后一年的体重减轻情况。
