Scholarlink

Breast cancer remains one of the leading causes of cancer deaths, requiring advanced methods to assess treatment efficacy and improve survival predictions. This study aims to predict overall survival (OS) and progression-free survival (PFS) between 6 and 36 months in patients with advanced breast cancer receiving ribociclib therapy. Survival analysis was performed using two datasets, RIBECCA and RIBANNA, which were derived from German Phase III clinical and non-interventional studies, to assess the survival outcomes in patients with advanced breast cancer under ribociclib therapy. The best OS results were obtained at month 12 with the Cox Proportional Hazards model, achieving a concordance index (C-index) of 0.720. The best PFS predictions were achieved at month 6 with the GBSA model, with a C-index of 0.728. In addition, Shapley Additive Explanation (SHAP) values were used to identify the most influential features and explain model predictions. Key predictors included liver metastases, treatment regimen, prior treatment, and quality of life scores. This study highlights the potential of survival machine learning models, offering valuable insights towards data-driven improvements in clinical decision-making.

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

Machine Learning Predictions of Overall and Progression-Free Survival in Advanced Breast Cancer

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Tatiana Merzhevich,
  • Alexandros Tanzanakis,
  • Emmanuelle Salin,
  • Claudia Quiering,
  • Christoph Kurz,
  • Benjamin Gmeiner,
  • Bjoern M. Eskofier

摘要

Breast cancer remains one of the leading causes of cancer deaths, requiring advanced methods to assess treatment efficacy and improve survival predictions. This study aims to predict overall survival (OS) and progression-free survival (PFS) between 6 and 36 months in patients with advanced breast cancer receiving ribociclib therapy. Survival analysis was performed using two datasets, RIBECCA and RIBANNA, which were derived from German Phase III clinical and non-interventional studies, to assess the survival outcomes in patients with advanced breast cancer under ribociclib therapy. The best OS results were obtained at month 12 with the Cox Proportional Hazards model, achieving a concordance index (C-index) of 0.720. The best PFS predictions were achieved at month 6 with the GBSA model, with a C-index of 0.728. In addition, Shapley Additive Explanation (SHAP) values were used to identify the most influential features and explain model predictions. Key predictors included liver metastases, treatment regimen, prior treatment, and quality of life scores. This study highlights the potential of survival machine learning models, offering valuable insights towards data-driven improvements in clinical decision-making.