An Integrated Machine Learning Scheme for Predicting Mammographic Anomalies in High-Risk Individuals Using...

Area of Science:

• Oncology
• Biostatistics
• Machine Learning

Background:

• Mammographic screening is crucial for early breast cancer detection.
• Identifying high-risk individuals can optimize screening protocols.
• Machine learning offers advanced tools for analyzing complex health data.

Purpose of the Study:

• To develop an integrated machine learning (ML) scheme for predicting positive mammographic findings.
• To identify significant demographic and obstetric/gynecological risk factors for mammographic anomalies.
• To combine Least Absolute Shrinkage and Selection Operator (Lasso) logistic regression with Extreme Gradient Boosting (XGB) for enhanced prediction.

Main Methods:

• Retrospective analysis of questionnaire data from a national mammographic screening program (2017-2020).
• Utilized an integrated ML scheme combining Lasso logistic regression and XGBoost.
• Analyzed data from 21,107 women, correlating 18 breast-cancer-related risk factors with mammographic findings.

Main Results:

• The integrated ML scheme, particularly Lasso models with XGB-generated variable combinations, demonstrated effective prediction.
• Top predictors for positive mammography included younger age, breast self-examination, older age at first childbirth, nulliparity, and recent mammography history (within 2 years).

Conclusions:

• The proposed integrated ML approach enhances the prediction of positive mammographic findings.
• Specific risk factors highlight the need for targeted and timely mammographic screening interventions.
• This study provides valuable insights for refining breast cancer screening strategies.