Construction of an infection risk prediction model for breast cancer patients undergoing postoperative chemotherapy
ZHONG Yanlan
ZHANG Qing
PENG Yun
Abstract：Objective:To construct an infection risk prediction model for breast cancer patients undergoing postoperative chemotherapy.Methods:A total of 368 patients who underwent postoperative chemotherapy for breast cancer and admitted to the thyroid and breast surgery department of Ganzhou People's Hospital from October 2020 to June 2023 were selected as the study subjects.Based on Logistic regression analysis,classification regression tree,and back propagation neural network algorithms,risk prediction models for infection in breast cancer patients undergoing postoperative chemotherapy were constructed.The predictive value was analyzed by comparing the receiver operating characteristic curves of the prediction models.Results:Infections occurred in 62 breast cancer patients undergoing postoperative chemotherapy,primarily affecting the respiratory tract.Multivariate Logistic regression analysis results showed that bone marrow suppression,C-reactive protein,and procalcitonin were independent influencing factors for infections in breast cancer patients undergoing postoperative chemotherapy(P<0.05).The classification regression tree model showed that C-reactive protein,procalcitonin,drainage duration,and diabetes were influencing factors for infections.The back propagation neural network model showed that the importance of factors affecting infections in breast cancer patients undergoing postoperative chemotherapy was ranked as follows:C-reactive protein>procalcitonin>combined diabetes>length of hospital stay>bone marrow suppression>drainage duration>serum albumin>chemotherapy cycles.Among the three models,the back propagation neural network model demonstrated the best predictive performance.The area under the receiver operating characteristic curve was 0.996.The sensitivity was 1.000.The specificity was 0.931.Conclusions:The influencing factors of infection risk in breast cancer patients undergoing postoperative chemotherapy include C-reactive protein,procalcitonin,diabetes,length of hospital stay,and bone marrow suppression,etc.The infection risk prediction model for breast cancer patients undergoing postoperative chemotherapy constructed based on machine learning algorithms all exhibited good performance,with the back propagation neural network model demonstrating the best predictive performance.
Keywords：machine learningbreast cancerpostoperativechemotherapyinfectionpredictive modelLogistic regression analysisclassification regression treeback propagation neural networkinfluencing factor
Publication Date：2025-12-10
Online Publishing Date：2025-12-08(First online date of this platform, not the publication date of the document)
Pages：8( 3934-3941 )