A Machine Learning Model for Predicting Breast Cancer Recurrence and Supporting Personalized Treatment Decisions...

Area of Science:

• Oncology
• Biostatistics
• Machine Learning

Background:

• Accurate prediction of breast cancer recurrence is crucial for personalized treatment.
• Existing models may lack the precision and interpretability needed for clinical decision support.

Purpose of the Study:

• To evaluate a machine learning model integrating LASSO feature selection and ensemble learning for breast cancer recurrence risk prediction.
• To support personalized treatment decisions in breast cancer patients.

Main Methods:

• Utilized clinical data from 1,131 breast cancer patients.
• Applied LASSO for feature selection and developed an ensemble learning model.
• Employed SHAP (Shapley additive explanations) for model interpretability.

Main Results:

• The ensemble model achieved an AUC of 0.817, outperforming single models (AUC 0.711).
• Identified six key predictors: lymph node positivity, ER status, Ki-67, lymphovascular invasion, tumor size, and age.
• SHAP analysis quantified feature contributions, enhancing clinical understanding.

Conclusions:

• The LASSO-enhanced ensemble model improves breast cancer recurrence prediction accuracy and interpretability.
• The model provides individualized recurrence risks, aiding data-driven clinical decision-making.
• Potential as a clinical decision support tool for personalized breast cancer management.