Circulating miRNA's biomarkers for early detection of hepatocellular carcinoma in Egyptian patients based on machine learning algorithms

Gehad Ismail Sayed ^1,2, Mona Solyman ^3,2, Gamalat El Gedawy ^4,2

⁰School of Computer Science, Canadian International College (CIC), Cairo, Egypt.

Scientific Reports +

|

February 29, 2024

View abstract on PubMed

Summary

Hepatocellular Carcinoma (HCC) is a deadly liver cancer. This study introduces a machine learning model using circulating microRNAs (miRNAs) to detect HCC in Egyptian patients, showing superior accuracy over traditional methods.

Area Of Science

Oncology
Genetics
Bioinformatics

Background

Liver cancer, particularly Hepatocellular Carcinoma (HCC), is a major global health concern with poor prognosis, especially in advanced stages.
The prevalence of HCC is increasing in Egypt, characterized by distinct molecular profiles including messenger RNA (mRNA) and microRNA (miRNA) expression.
Circulating miRNAs show promise as non-invasive diagnostic and prognostic biomarkers for HCC.

Purpose Of The Study

To propose and evaluate a machine learning model for the early diagnosis and prognosis of HCC in Egyptian patients using circulating miRNAs.
To identify specific circulating miRNAs as potential biomarkers for HCC detection and patient stratification.

Main Methods

A three-phase HCC detection model was developed: data preprocessing, feature selection using a Binary African Vulture Optimization Algorithm (BAVO), and classification/cross-validation.
The model employed Support Vector Machine (SVM) and k-folds cross-validation for classification.
The model was validated on three independent datasets of Egyptian HCC patients.

Main Results

The proposed machine learning model demonstrated superior diagnostic accuracy compared to traditional statistical methods across all three validation studies.
Specific circulating miRNAs, including miRNA-483-5p, miRNA-21, and miRNA-155, were identified as potential early diagnostic and prognostic biomarkers for HCC.
The model achieved high sensitivity and specificity, with results like 98% sensitivity and 99% specificity in one study.

Conclusions

The developed machine learning model effectively utilizes circulating miRNAs for the accurate diagnosis and prognosis of HCC in the Egyptian population.
Circulating miRNAs, specifically miRNA-483-5p, miRNA-21, and miRNA-155, represent promising novel biomarkers for early HCC detection and outcome prediction.

Keywords:

Hepatocellular carcinoma Liver cancer Machine learning Swarm optimization miRNA