Lasso-Logistic regression analysis and nomogram prediction model construction of the influencing factors of postoperative malnutrition in patients with hepatocellular carcinoma
Objective:To analyze the influencing factors of postoperative malnutrition in patients with liver cancer by Lasso-Logistic regression,and to construct a nomogram prediction model to provide reference for the adjustment of postoperative nutritional status in patients with liver cancer.Methods:A total of 460 patients undergoing surgery for liver cancer in our hospital from January 2021 to June 2023 were selected and divided into a training cohort(322 cases)and a validation cohort(138 cases)according to a ratio of 7:3.The nutritional status of patients after sur-gery was evaluated according to the patients'subjective global rating(PG-SGA)scale,and the patients were divided into a normal nutrition group and a malnutrition group.Lasso-Logistic regression was used to analyze the influencing factors of postoperative malnutrition of liver cancer.The nomogram prediction model was constructed by rms pack-age in the training cohort,and the predictive efficiency of the model was verified by receiver operating characteristic(ROC)curve,calibration curve and decision curve(DC A)in the validation cohort.Results:There was no statisti-cally significant difference in general clinical data between the training queue and the validation queue(P＞0.05).Lasso-Logistic regression analysis showed that age,preoperative body mass index(BMI),preoperative nutrition risk screening score(NRS2002),anemia,diabetes,tumor stage,albumin(ALB),glucagon like peptide-1(GLP-1),total bile acid(TBA),and glycocholic acid(CG)were all independent influencing factors for postoperative malnu-trition of liver cancer patients in the training cohort(P＜0.05).The ROC curve showed that the predicted area un-der the curve of the column chart prediction model for postoperative malnutrition in liver cancer surgery patients in the training and validation queues was 0.873 and 0.902,respectively.The calibration curve showed that the col-umn chart model predicted the risk of postoperative malnutrition in liver cancer surgery patients in the training and validation queues,which was basically consistent with the actual risk status.Conclusion:The constructed nomo-gram model has high predictive value and good predictive utility for predicting postoperative malnutrition in patients with liver cancer surgery,but further validation and evaluation are needed.
NETL
NSTL
万方数据
