# The emerging role of machine learning-based methods in cancer classification using microRNA

## TL;DR

Machine learning models using microRNA data can improve cancer classification and enable non-invasive diagnostics.

## Contribution

This paper reviews how machine learning methods enhance cancer classification by leveraging microRNA biomarkers.

## Key findings

- ML models using miRNA data can distinguish cancerous from normal tissues.
- Techniques like Random Forest and SVM improve breast cancer subtype classification.
- Deep learning aids in kidney cancer analysis using miRNA data.

## Abstract

Early detection and accurate classification of cancer are crucial to improving patient outcomes. Diagnosis and classification of tumors using conventional methods remains challenging. MicroRNAs (miRNAs) are potential biomarkers for accurate tumor classification and differentiation of tumor subtypes. In cancer progression, miRNAs act as oncogenes or tumor suppressors to regulate gene expression. As a result of their stability in bodily fluids such as blood, urine, and saliva, they are ideal for non-invasive diagnostic procedures. Machine learning (ML) models can identify discriminative miRNAs for various cancers, such as breast, lung, colorectal, and kidney cancers. The integration of ML with miRNA data has demonstrated significant potential for differentiating cancerous tissues from normal tissues and identifying clinically relevant biomarkers. For instance, techniques such as feature engineering and selection, including recursive ensemble selection and miRNA-mRNA network analysis, have been shown to enhance both model accuracy and interpretability. Methods based on Random Forest (RF) and Support Vector Machines (SVM) have successfully classified breast cancer subtypes, and miRNA signatures from fecal samples have been highly effective in diagnosing colorectal cancer. Furthermore, deep learning and neuro-fuzzy systems support kidney cancer analysis, highlighting miRNA-driven ML's role in cancer diagnostics and personalized treatment. This review illustrates the transformative potential of miRNA-driven ML models for advancing cancer diagnostics and enabling personalized treatment strategies.

•Early cancer detection and precise classification are key to better outcomes.•Standard diagnostics still struggle to distinguish cancer types and subtypes.•MicroRNAs influence cancer progression and are detectable in body fluids.•MicroRNAs show strong potential as biomarkers for tumor classification.•Machine learning helps identify microRNAs that differentiate cancer types.

Early cancer detection and precise classification are key to better outcomes.

Standard diagnostics still struggle to distinguish cancer types and subtypes.

MicroRNAs influence cancer progression and are detectable in body fluids.

MicroRNAs show strong potential as biomarkers for tumor classification.

Machine learning helps identify microRNAs that differentiate cancer types.