EsoDetect: computational validation and algorithm development of a novel diagnostic and prognostic tool for dysplasia in Barrett's esophagus

  • 0Ophiomics, Lisbon, Portugal.

|

|

Summary

This summary is machine-generated.

Researchers identified gene expression biomarkers to diagnose low-grade dysplasia in Barrett's esophagus (BE) and predict progression to esophageal adenocarcinoma (EAC). Machine learning algorithms using these biomarkers achieved high accuracy for early diagnosis and prognosis.

Area Of Science

  • Gastroenterology
  • Oncology
  • Bioinformatics

Background

  • Barrett's esophagus (BE) is a precursor to esophageal adenocarcinoma (EAC).
  • Early diagnosis and prognosis of BE progression are critical for patient outcomes.
  • Current diagnostic and prognostic tools require improvement.

Purpose Of The Study

  • Identify biomarkers for diagnosing low-grade dysplasia (LGD) in BE.
  • Identify biomarkers for predicting BE progression to EAC.
  • Develop diagnostic and prognostic algorithms for BE management.

Main Methods

  • Analyzed gene expression data from public databases (RNAseq and microarray).
  • Utilized thresholding functions and F1-scores for gene ranking.
  • Trained and selected machine learning classifiers (SVM) for diagnostic and prognostic applications.

Main Results

  • Identified 18 diagnostic and 15 prognostic genes.
  • Developed a linear SVM (diagnostic) and RBF SVM (prognostic) algorithm, each using 10 genes.
  • Both algorithms achieved recall and specificity scores exceeding 0.90.

Conclusions

  • Novel gene expression biomarkers and SVM algorithms show potential for early BE diagnosis and prognosis.
  • The developed algorithms demonstrate superior performance compared to existing literature.
  • Further clinical validation is necessary for integration into patient management.