Scholarlink

Purpose <p>This study aims to compare the clinical and patient-reported outcomes (PROs) between oncoplastic breast-conserving surgery (OBCS) and conventional breast-conserving surgery (CBCS) in early breast cancer (EBC) patients. Additionally, we applied machine learning (ML) to explore clinicopathologic variables associated with satisfaction with breasts, providing exploratory insights into individualized and patient-centered care in breast-conserving surgery (BCS).</p> Methods <p>A retrospective analysis was conducted on 350 breast cancer patients who underwent either OBCS (<i>n</i> = 182) or CBCS (<i>n</i> = 168). Clinical data, including patient characteristics, surgical details, and complications, were collected and compared between the two groups. Patient satisfaction was measured using the BREAST-Q questionnaire. Disease-free survival (DFS) was analyzed using Kaplan–Meier survival curves. ML models were developed to predict satisfaction with their breasts. SHapley Additive exPlanations (SHAP) values were employed to determine the importance of variables in the optimal model.</p> Results <p>Patients in the OBCS group had significantly larger tumor sizes (23.0 ± 9.2&#xa0;mm vs. 17.5 ± 6.2&#xa0;mm, <i>p</i> &lt; 0.001) and longer operative times (179 ± 50&#xa0;min vs. 90 ± 43&#xa0;min, <i>p</i> &lt; 0.001) compared to the CBCS group. Despite these differences, complication rates were comparable between the groups. The OBCS group reported higher satisfaction with breasts (78 vs. 65, <i>p</i> &lt; 0.001) and better psychosocial well-being (84 vs. 71, <i>p</i> &lt; 0.001). Kaplan–Meier analysis showed no significant difference in DFS between the two groups during the available follow-up period (<i>p</i> = 0.267). ML model performance, evaluated through accuracy, ROC curves and calibration curves, identified the neural network (NN) (AUC: 0.76 and 0.77) as the optimal model in both the training and validation cohorts. SHAP analysis indicated that surgery type, menopause status, location of carcinoma and BMI were among the variables associated with breast satisfaction.</p> Conclusion <p>OBCS was associated with higher breast satisfaction and psychosocial well-being than CBCS in this cohort, with no significant difference in long-term outcomes during the available follow-up period. ML and SHAP analysis provided exploratory insights into factors associated with breast satisfaction, highlighting the potential value of personalized approaches in BCS.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

An exploratory study of patient-reported outcomes in oncoplastic and conventional breast-conserving surgery using explainable machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年4月28日

查看全文
  • Jiahui Ren,
  • Daxue Li,
  • Junge Gong,
  • Hongyuan Li,
  • Xiang Zhang

摘要

Purpose

This study aims to compare the clinical and patient-reported outcomes (PROs) between oncoplastic breast-conserving surgery (OBCS) and conventional breast-conserving surgery (CBCS) in early breast cancer (EBC) patients. Additionally, we applied machine learning (ML) to explore clinicopathologic variables associated with satisfaction with breasts, providing exploratory insights into individualized and patient-centered care in breast-conserving surgery (BCS).

Methods

A retrospective analysis was conducted on 350 breast cancer patients who underwent either OBCS (n = 182) or CBCS (n = 168). Clinical data, including patient characteristics, surgical details, and complications, were collected and compared between the two groups. Patient satisfaction was measured using the BREAST-Q questionnaire. Disease-free survival (DFS) was analyzed using Kaplan–Meier survival curves. ML models were developed to predict satisfaction with their breasts. SHapley Additive exPlanations (SHAP) values were employed to determine the importance of variables in the optimal model.

Results

Patients in the OBCS group had significantly larger tumor sizes (23.0 ± 9.2 mm vs. 17.5 ± 6.2 mm, p < 0.001) and longer operative times (179 ± 50 min vs. 90 ± 43 min, p < 0.001) compared to the CBCS group. Despite these differences, complication rates were comparable between the groups. The OBCS group reported higher satisfaction with breasts (78 vs. 65, p < 0.001) and better psychosocial well-being (84 vs. 71, p < 0.001). Kaplan–Meier analysis showed no significant difference in DFS between the two groups during the available follow-up period (p = 0.267). ML model performance, evaluated through accuracy, ROC curves and calibration curves, identified the neural network (NN) (AUC: 0.76 and 0.77) as the optimal model in both the training and validation cohorts. SHAP analysis indicated that surgery type, menopause status, location of carcinoma and BMI were among the variables associated with breast satisfaction.

Conclusion

OBCS was associated with higher breast satisfaction and psychosocial well-being than CBCS in this cohort, with no significant difference in long-term outcomes during the available follow-up period. ML and SHAP analysis provided exploratory insights into factors associated with breast satisfaction, highlighting the potential value of personalized approaches in BCS.