Key predictive factors of breast cancer based on race using machine learning models
Shuning Yin1, Gaurav Nanda1, Raji Sundararajan1
1School of Engineering Technology, Purdue University, West Lafayette, IN 47907, USA.
Annals of Epidemiology
|March 30, 2026
Summary
Machine learning identified key breast cancer risk factors like biopsy history and age, revealing significant racial disparities in incidence rates across different age groups.
Area of Science:
- Oncology
- Data Science
- Public Health
Background:
- Breast cancer is a significant global health concern.
- Understanding risk factors and racial disparities is crucial for targeted prevention and treatment.
Purpose of the Study:
- To identify key predictors of breast cancer risk using machine learning (ML).
- To investigate racial differences in breast cancer risk factors and incidence.
- To apply explainable AI (XAI) for model interpretability.
Main Methods:
- Utilized Breast Cancer Surveillance Consortium (BCSC) data (2005-2017).
- Applied Naïve Bayes, Logistic Regression, and Extreme Gradient Boosting models.
- Employed variable importance and SHapley Additive exPlanations (SHAP) for factor identification.
- Stratified analyses by six racial groups.
Main Results:
- History of biopsy (50.04%) and age group (25.85%) were the strongest predictors.
- Menopausal status, breast density, and age at first childbirth were also significant.
- Racial disparities observed: White women had highest incidence (esp. >65), Black women higher in younger groups, Native American women in certain older groups, Asian/Pacific Islander and Other/Mixed groups generally lower.
Conclusions:
- ML and XAI effectively identified breast cancer risk predictors.
- The study highlights significant racial disparities in breast cancer risk factors and incidence.
- Findings underscore the need for race-specific breast cancer risk assessment and interventions.
