Machine learning-based prognostic model for triple-negative breast cancer with axillary lymph node metastasis
Ruyi Huang1, Tianlu Jiang1, Xidong Lv1
1Department of Breast Surgery, The Affiliated Wuxi People's Hospital of Nanjing Medical University, Wuxi People's Hospital, Wuxi Medical Center, Nanjing Medical University, Wuxi, Jiangsu, China.
Frontiers in Oncology
|July 25, 2026
Summary
This study developed a machine learning model for triple-negative breast cancer (TNBC) with axillary lymph node metastasis (ALNM), identifying key prognostic factors like tumor grade and radiotherapy to improve risk prediction for this high-risk group.
Area of Science:
- Oncology
- Biostatistics
- Machine Learning
Background:
- Triple-negative breast cancer (TNBC) with axillary lymph node metastasis (ALNM) signifies a high-risk patient group with a poorer prognosis.
- Currently, no validated risk prediction model exists specifically for TNBC patients with ALNM.
- Machine learning offers a potential avenue for improved risk stratification by integrating diverse clinical and pathological data.
Purpose of the Study:
- To develop and validate the first machine learning-based prognostic model for TNBC patients with ALNM.
- To identify key independent prognostic factors influencing survival in this population.
- To assess the performance and clinical utility of various machine learning survival models.
Main Methods:
- Retrospective analysis of 19,289 TNBC patients with ALNM from the SEER database (2015-2020).
- Development and comparison of five machine learning survival models (CoxPH, RSF, ERST, GBSA, ST).
- Evaluation using C-index, time-dependent AUC, Brier scores, calibration curves, decision curve analysis, and SHAP for interpretability.
Main Results:
- Multivariable Cox regression identified 13 independent prognostic factors, including demographics, tumor features, and treatment modalities.
- The Extremely Randomized Survival Trees (ERST) model demonstrated robust performance with a C-index of 0.7494.
- SHAP analysis highlighted tumor grade, N stage, and radiotherapy as the most influential prognostic factors.
Conclusions:
- A novel, internally validated machine learning prognostic model (ERST) was developed for TNBC patients with ALNM.
- The model exhibits strong discriminatory ability, excellent calibration, and favorable clinical utility.
- This tool aids in personalized risk assessment, potentially informing treatment decisions and improving prognostic counseling for this high-risk group.
