基于机器学习构建结直肠癌患者术后并发吻合口漏风险的决策树预测模型OA

Objective To establish a decision tree prediction model for anastomotic leakage(AL)after radical surgery for colorectal can-cer(CRC)based on a logistic regression model,so as to provide scientific evidence for clinical healthcare professionals to identify high-risk patients early and formulate individualized therapeutic intervention strategies,thereby optimizing the prevention,diagnosis,and treatment process for AL.Methods A total of 305 patients undergoing radical surgery for CRC at the First Affiliated Hospital of Jinzhou Medical University from January 2022 to December 2024 were retrospectively enrolled.According to whether AL occurred,the patients were divided into the AL group(n=42)and the non-AL group(n=263).Univariate analysis and multivariate logistic regression were used to analyze the influencing factors for postoperative AL in CRC patients,and a corresponding decision tree prediction model was established by machine learning.Results Univariate analysis showed that there were statistically significant differences between the two groups in age,smoking,alcohol consumption,hypertension,diabetes mellitus,TNM stage,distance of the lower edge of the tumor from the anal verge,preoperative body mass index(BMI),preoperative serum albumin level,and preoperative white blood cell count(all P＜0.05).Multivariate logistic re-gression analysis showed that age,diabetes mellitus,TNM stage,distance of the lower edge of the tumor from the anal verge,preoperative BMI,and preoperative serum albumin level were independent influencing factors for AL after radical surgery in CRC patients(all P＜0.05).A decision tree model was established based on these factors.Receiver operating characteristic curve showed that the area under the curve of the model was 0.91.Conclusions The decision tree model constructed based on logistic regression is helpful for medical staff to formu-late multidimensional treatment plans and achieve early warning and effective prevention and control of AL,improving the prognosis and quality of life of the patients.