目的 探讨联合临床及超声多参数构建的列线图模型预测人表皮生长因子受体2（human epidermal growth factor receptor 2，HER-2）阳性乳腺癌的应用价值。 方法 回顾性纳入3个中心343例经病理证实为乳腺癌的患者并分为训练集和验证集。在训练集中运用单因素、LASSO和多因素回归分析，选出独立预测因素并构建列线图模型；同时进行Bootstrap 1 000次抽样验证；采用Hosmer-Lemeshow检验评估模型拟合度；绘制校准曲线评估模型的校准度；绘制受试者工作特征（receiver operating characteristic，ROC）曲线并计算曲线下面积（area under the curve，AUC），及相关指标衡量模型的性能；绘制临床决策曲线评估模型的临床适用性，通过验证集进行验证。 结果 单因素、LASSO和多因素回归分析筛选后显示年龄、TTP和充盈缺损征为HER-2阳性乳腺癌的独立预测因素（均P < 0.05），并构建列线图模型。Bootstrap 1 000次抽样验证显示模型预测效能稳定；Hosmer-Lemeshow检验表明模型拟合度较好；校准曲线显示模型预测概率良好；训练集AUC为0.863（95%CI：0.806 ~ 0.920），验证集AUC为0.846（95%CI：0.764 ~ 0.929），提示模型具有较高的区分能力和泛化能力；临床决策曲线表明模型的临床适用性较好。 结论 临床和多模态超声参数联合构建的列线图模型在预测HER-2阳性乳腺癌中具有潜在应用价值。

Objective To evaluate the predictive value of a nomogram model developed by integrating clinical and ultrasound multiparameters for HER-2-positive breast cancer. Methods This study retrospectively enrolled 343 patients with pathologically confirmed breast cancer from three medical centers and randomly divided them into training and validation cohorts. Univariate analysis， LASSO regression， and multivariate logistic regression were conducted on the training set to identify independent prognostic factors and construct a nomogram model. Bootstrap resampling with 1000 iterations was performed to evaluate the model′s robustness. Model calibration was assessed using calibration curves and the Hosmer-Lemeshow goodness-of-fit test. Receiver operating characteristic （ROC） curves were generated to evaluate model discrimination， and the area under the curve （AUC） along with other performance metrics were calculated. Decision curve analysis was employed to assess the clinical utility of the model， and the validation cohort was used for external validation. Results Univariate， LASSO， and multivariate regression analyses demonstrated that age， TTP （time to peak）， and the presence of a filling defect sign were independent predictors of HER-2-positive breast cancer （all P < 0.05）. Based on these independent predictors， a nomogram model was constructed. Bootstrap validation with 1，000 resamples indicated that the model′s predictive performance was stable. The Hosmer-Lemeshow test confirmed satisfactory model calibration， while the calibration curve illustrated accurate prediction probabilities. The area under the curve （AUC） for the training set was 0.863 （95%CI： 0.806 ~ 0.920）， and for the validation set， it was 0.846 （95%CI： 0.764 ~ 0.929）， indicating strong discriminative and generalization capabilities. Additionally， the clinical decision curve analysis demonstrated favorable clinical utility. Conclusion A nomogram model integrating clinical and multimodal ultrasound parameters demonstrates potential utility in predicting HER-2-positive breast cancer.