Diagnosis of triple negative breast cancer using expression data with several machine learning tools

Area of Science:

• Oncology
• Bioinformatics
• Artificial Intelligence

Background:

• Breast cancer is a leading global cancer, with Triple Negative Breast Cancer (TNBC) subtypes exhibiting poor prognosis.
• TNBC is characterized by the absence of oestrogen receptors, progesterone receptors, and HER2, complicating early detection and treatment.
• Developing accurate diagnostic tools for TNBC is crucial for improving patient outcomes.

Purpose of the Study:

• To evaluate the efficacy of AI-based neural models for the differential diagnosis of TNBC.
• To compare the diagnostic accuracy of Binary Logistic Regression, Multi-Layer Perceptron (MLP), and Radial Basis Functions (RBF) for TNBC detection.
• To identify significantly upregulated genes in TNBC compared to healthy controls using microarray data.

Main Methods:

• Signal intensity data from microarray experiments were analyzed.
• AI models including Binary Logistic Regression, MLP, and RBF were employed for classification tasks.
• Gene expression patterns differentiating TNBC from normal samples were investigated.

Main Results:

• The Multi-Layer Perceptron (MLP) model demonstrated the highest accuracy at 93.4% in classifying TNBC.
• Binary Logistic Regression achieved 90.0% accuracy, while the Radial Basis Functions (RBF) model showed 74% accuracy.
• Specific gene expression profiles were identified as significantly upregulated in TNBC samples.

Conclusions:

• AI-based neural networks, particularly MLP, offer a highly accurate approach for the differential diagnosis of TNBC.
• These AI models can aid in the early detection of TNBC, potentially improving patient prognosis.
• Microarray data combined with machine learning provides a powerful strategy for identifying and diagnosing breast cancer subtypes.