Breast Cancer Type Classification Using Machine Learning.

Area of Science:

• Genomic research
• Precision medicine
• Computational biology

Background:

• Breast cancer is molecularly diverse, necessitating precise classification for effective treatment.
• Distinguishing triple negative breast cancer (TNBC) from non-TNBC is crucial due to TNBC's aggressive nature and poor prognosis.
• Gene expression data holds potential for classifying breast cancer subtypes.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for classifying triple negative breast cancer (TNBC) and non-TNBC patients.
• To identify key gene expression features that differentiate between TNBC and non-TNBC.
• To compare the efficacy of different ML algorithms for breast cancer subtype classification.

Main Methods:

• Analysis of RNA-Sequence data from The Cancer Genome Atlas (TCGA) including 110 TNBC and 992 non-TNBC samples.
• Feature selection based on gene expression thresholds to identify discriminative genes.
• Development and evaluation of four ML models: Support Vector Machines (SVM), K-nearest neighbor, Naïve Bayes, and Decision Tree.
• Independent validation of models using external gene expression datasets.

Main Results:

• The Support Vector Machine (SVM) algorithm achieved the highest accuracy in classifying triple negative breast cancer (TNBC) and non-TNBC.
• SVM exhibited lower misclassification errors compared to K-nearest neighbor, Naïve Bayes, and Decision Tree models.
• Selected gene expression features effectively differentiated between the two breast cancer subtypes.

Conclusions:

• Machine learning algorithms, particularly SVM, are effective tools for classifying breast cancer into triple negative and non-triple negative subtypes.
• ML-based classification using gene expression data can aid in the precision management of breast cancer.
• This approach addresses the unmet need for accurate distinction between aggressive TNBC and other breast cancer types.