A Machine Learning Approach for Identifying Gene Biomarkers Guiding the Treatment of Breast Cancer

Area of Science:

• Genomics
• Bioinformatics
• Machine Learning in Oncology

Background:

• Genomic profiles offer insights into breast cancer cell biology and treatment responses.
• Personalizing cancer therapy based on gene expression is a key goal in oncology.

Purpose of the Study:

• To develop a hierarchical machine learning system for predicting 5-year breast cancer survivability.
• To identify potential gene biomarkers associated with specific treatment outcomes and patient survival.

Main Methods:

• A hierarchical machine learning model was developed, classifying patients based on survivability and therapy type (hormone therapy, radiotherapy, surgery).
• The model utilizes a tree-based structure with five nodes, classifying one class against others at each node.
• Standard classifiers, feature selection, and prediction methods were applied to genomic and clinical data from 347 breast cancer patients.

Main Results:

• The machine learning system achieved high performance in predicting patient classes.
• Identified genes at each node may serve as potential biomarkers for targeted breast cancer treatments.
• Literature analysis confirmed the strong association of some identified biomarkers with breast cancer and general cancer survivability.

Conclusions:

• The developed machine learning model effectively predicts 5-year breast cancer survivability.
• The identified gene biomarkers hold promise for the development of personalized treatment strategies.
• This approach can contribute to improving therapeutic decisions and patient outcomes in breast cancer care.