Construction and verification of a risk prediction model for postoperative pulmonary complications following laparoscopic hepatectomy

CHEN Can-hui; WANG Xiao-zhen; LIANG Han-biao; LIAO Le-yi; WANG Kai; LI Chuan-jiang

doi:10.19538/j.cjps.issn1005-2208.2025.05.20

Abstract：Objective To investigate the risk factors for postoperative pulmonary complications(PPCs)following laparoscopic hepatectomy and develop a risk prediction model for PPCs.Methods A retrospective analysis was conducted on the clinical case information of 500 patients who underwent laparoscopic hepatectomy at the Hepatobiliary Surgery Department of Nanfang Hospital,Southern Medical University,betweem January 2023 and May 2024.These patients were categorized into a control group(406 cases)and a PPCs group(94 cases)based on whether PPCs occurred postoperatively.The enrolled patients were divided into two datasets at a ratio of 7∶3,with the first 70％serving as the training set and the remaining 30％as the validation set.The risk factors for PPCs were screened and analyzed using LASSO regression,and a prediction model was constructed based on LASSO-logistic regression.Results The independent risk factors of PPCs finally included in the prediction model were history of chronic lung disease,difficult hepatectomy,extent of hepatectomy,operation duration,intraoperative colloid infusion,postoperative fluid infusion within 24 hours,postoperative ambulation time,and postoperative tracheal intubation removal time.The independent protective factor of PPCs was postoperative preventive nebulization therapy.The C-index of the model was 0.898(95％CI 0.852-0.944),and the Hosmer-Lemeshow goodness of fit test yielded a P value of 0.2785.The LASSO-logistic regression model demonstrated good discrimination and calibration in the validation set[with an area under the receiver operating characteristic curve of 0.863(95％CI 0.796-0.929,P＜0.001)].In the clinical decision curve(DCA),the threshold probability range for the training set was 0.02-0.90,and for the validation set,it was 0.03-0.88.Conclusion The calibration curve fit of the nomogram model constructed based on LASSO-logistic regression is ideal,demonstrating good consistency between the predicted and actual incidence of PPCs.The model exhibits strong predictive power,yet it requires further external validation for confirmation.