基于列线图构建乳腺肿块微创旋切术后病灶残留相关影响因素的预测模型

Construction of a predictive model based on columnar maps for factors influencing lesion residuals after minimally invasive rotary mastectomy for breast masses

目的探讨乳腺肿块微创旋切术后病灶残留相关影响因素,并采用列线图构建预测模型。方法选取临沂市人民医院乳腺外科2018年6月至2020年6月乳腺肿块患者228例作为研究对象,均行微创旋切术,采用单因素、多因素分析术后病灶残留的影响因素,并构建列线图风险预警模型,给予评价与验证。结果228例术后随访6个月,失访3例,225例完成术后随访,其中成功切除病灶有185例,占比为82.22%;病灶残留有40例,占比为17.78%。单因素、Logistic回归分析显示,肿瘤直径(t=15.52,P<0.001)、病灶形态(t=15.52,P<0.001)、病灶边界(χ^(2)=7.71,P=0.006)、超声引导(χ^(2)=12.69,P<0.001)、微创旋切系统(χ^(2)=10.64,P=0.001)是乳腺肿块微创旋切术后病灶残留的影响因素。基于以上因素构建乳腺肿块微创旋切术后病灶残留的列线图模型,模型受试者工作特征曲线(ROC)下面积为0.974(95%CI:0.954~0.994),表明校准曲线与实际曲线基本吻合;模型外部验证显示,所得ROC下面积为0.962(95%CI:0.933~0.991),表明该模型具有较高的校准度。结论肿瘤直径、病灶形态、病灶边界、超声引导、微创旋切系统是乳腺肿块微创旋切术后病灶残留的影响因素,基于以上因素构建的术后发生病灶残留的列线图模型预测具有较好的可信度。

Objective To investigate the influencing factors associated with lesion residual after minimally invasive rotary mastectomy for breast masses and to construct a predictive model using columnar plots.Methods Two hundred and twenty-eight patients with Breast lumps in Linyi People’s Hospital from Jun.2018 to Jun.2020 were selected as study subjects,all of whom underwent minimally invasive rotational resection,and the influencing factors of postoperative lesion residual were analyzed using univariate and multifactorial analysis,and a column line graph risk warning model was constructed and given for evaluation and validation.Results 228 patients were followed up for 6 months after surgery,3 cases were lost,225 cases completed postoperative follow-up,among which 185 cases(82.22%)had successfully resected lesions;40 cases(17.78%)had residual lesions.Univariate,logistic regression analysis showed that tumor diameter(t=15.52,P<0.001),lesion morphology(t=15.52,P<0.001),lesion boundary(χ^(2)=7.71,P=0.006),ultrasound guidance(χ^(2)=12.69,P<0.001),and minimally invasive spinotomy system(χ^(2)=10.64,P=0.001)were the factors influencing lesion residual after minimally invasive spinotomy of breast masses.Based on the above factors to construct a column line graph model of lesion residual after minimally invasive rotational mastectomy for breast lumps,the area under the working characteristic curve(ROC)for model subjects was 0.974(95%CI:0.954-0.994),indicating that the calibration curve basically matched the actual curve;external validation of the model showed that the area under the obtained ROC was 0.962(95%CI:0.933-0.991),indicating that the model has a high degree of calibration.Conclusion Tumor diameter,lesion morphology,lesion boundary,ultrasound guidance,and minimally invasive spinotomy system are the influencing factors of lesion residual after minimally invasive spinotomy for Breast lumps,and the column line graph model prediction of lesion residual occurring after surgery constructed based on the above factors has good confidence.