Construction and validation of a risk prediction model for cancer-related anorexia in breast cancer patients undergoing chemotherapy
JIA Xuebin
BAI Jinxin
MA Suhui
BAI Yifan
WANG Xing
BAO Chenlong
Abstract：Objective:To explore risk factors for breast cancer chemotherapy patients with cancer-related anorexia,to build and verify Logistic regression and random forest model prediction and generalization performance.Methods:A non-matched group case-control study was adopted.A Logistic regression model and a random forest model were constructed for 412 breast cancer patients undergoing chemotherapy in tertiary grade A hospitals in Hebei Province.External validation was conducted using 128 patients from another two tertiary grade A hospitals to compare the performance of the two risk prediction models.Results:Low pain tolerance,high economic toxicity,introversion,nausea and vomiting,suspected insomnia,and insomnia were the common predictors of the two predictive models,while the random forest model increased the fatigue factor.The AUC of internal validation of Logistic regression and random forest model was 0.840 and 0.899 respectively,and the AUC of external validation was 0.946 and 0.987.There were statistically significant differences in the comparison of AUC between internal and external validation(Z=6.121,P<0.001);Z=2.588,P<0.05.Conclusions:Both models have good predictive performance and clinical applicability for the risk of cancer anorexia in breast cancer patients undergoing chemotherapy.The random forest model is superior to the Logistic regression model.
Keywords：breast cancerchemotherapycancer-related anorexiaprediction modelrandom forest
Publication Date：2025-05-15
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 1638-1643 )