[目的]分析乳腺影像报告与数据系统(breast imaging reporting and data system,BI-RADS)4～5类女性中乳腺癌发病风险因素,构建乳腺癌风险预测模型.[方法]收集2019年6月至2023年3月山西省晋城市泽州县妇幼保健院和长治医学院附属和平医院337例BI-RADS 4～5类35岁以上女性临床及影像资料,经单因素分析后以P＜0.05的变量结合最优子集法进行变量筛选,以赤池信息准则(akaike informa-tion criterion,AIC)最低的组合构建Logistic回归模型并进行Bootstrap内部验证,采用C指数、Calibration校准曲线、决策曲线(decision curve analysis,DCA)评估预测模型的临床有效性.[结果]体质指数(body mass index,BMI)、生产次数＞2次,患有乳腺良性疾病史、致密型乳腺、肿块不规则为乳腺癌的危险因素.Logit(P)=-6.618+0.24BMI+0.606乳腺良性疾病史+0.306致密型乳腺+1.059×生育次数(＞2次)+1.814肿块不规则.该模型的C指数为0.811(95％CI:0.731～0.891),H-L拟合优度检验P=0.205(P＞0.05),结合校准曲线证明此模型的区分度和校准度良好.[结论]成功构建乳腺癌高危人群风险评估模型,且具有良好的预测价值,有利于识别乳腺癌患者,具备临床参考价值.
Construction of Risk Prediction Model for BI-RADS Category 4～5 Breast Cancer
[Objective]To investigate the risk factors of patients with BI-RADS(breast imaging reporting and data system)4～5 category breast cancer,and to construct a risk prediction model.[Methods]The clinical and imaging data of 337 patients over 35 years old with BI-RADS 4～5 admitted in Zezhou County Maternal and Child Health Hospital and Peace Hospital Affiliated to Changzhi Medical College from June 2019 to March 2023 were collected and analyzed.The risk factors for BI-RADS 4～5 breast cancer was analyzed with univari-ate analysis,the variables with P＜0.05 were screened with the optimal subset method,and the Logistic regres-sion model was constructed with the lowest combination of akaike information criterion(AIC)for Bootstrap internal verification.C-index,calibration calibration curve,and decision curve analysis(DCA)were used to evaluate the clinical effectiveness of the prediction model.[Results]Body mass index,history of benign breast disease,dense breast,number of births(＞2 times)and irregular masses were the independent risk factors for breast cancer.Logistic regression model was established with five variables:Logit(P)=-6.618+0.24×BMI+0.606 × history of benign breast disease+0.306 xdense breast+1.059 x number of births(＞2 times)+1.814x irregular masses.The C index of the model was 0.811(95％CI:0.731～0.891),and the H-L goodness of fit test P=0.205.Com-bined with the calibration curve,it was proved that the discrimination and calibration of this model were good.[Conclusion]A risk assessment model of BI-RADS 4～5 breast cancer has been constructed with good predic-tive value,which is helpful to identify breast cancer patients and had clinical reference value.
breast cancerBI-RADS 4～5Logistic regressionpredictive model
NETL
NSTL
万方数据
