Breast Cancer Subtypes Classification with Hybrid Machine Learning Model.

Area of Science:

• Oncology
• Medical Informatics
• Artificial Intelligence

Background:

• Breast cancer is a heterogeneous disease with distinct molecular subtypes.
• Machine learning (ML) techniques are advancing cancer detection and prognosis.
• Accurate tumor detection is crucial for effective clinical management.

Purpose of the Study:

• To develop a computer-driven diagnostic system for improved breast tumor detection accuracy.
• To classify triple-negative breast cancer (TNBC) and non-triple-negative breast cancer (non-TNBC) patients.
• To leverage clinicopathological features for enhanced breast cancer classification.

Main Methods:

• Proposed a hybrid ML model combining genetic algorithm and support vector machine (GA-SVM).
• Classified patients based on clinicopathological features from multiple tertiary care hospitals.
• Compared GA-SVM with other SVM-based models (SVM-RFE, LASSO-SVM, Grid-SVM, linear SVM).

Main Results:

• The GA-SVM hybrid model demonstrated superior classification performance compared to other models.
• Model validation was performed using two distinct hospital-based datasets from the North West of the African subcontinent.
• Performance was rigorously evaluated using 10-fold cross-validation and metrics like MSE, logarithmic loss, F1-score, AUC, and precision-recall curves.

Conclusions:

• The developed hybrid ML model shows promise for accurate breast cancer subtype classification.
• This tool can assist medical practitioners in optimizing treatment strategies.
• Improved classification can lead to better patient outcomes in breast cancer care.