Scholarlink

<p>Rheumatoid arthritis (RA) is a complex autoimmune disorder with a strong genetic basis. Early and accurate diagnosis remains challenging due to overlapping symptoms with other joint diseases. This research presents a machine learning-based framework for classifying RA using single nucleotide polymorphism data derived from genome-wide association studies. Datasets were collected for RA, osteoarthritis, psoriatic arthritis, juvenile idiopathic arthritis, and gout, and preprocessed. Random forest, extra trees, XGBoost, and bagging classifiers were trained and integrated through an optimized soft voting ensemble. The final model achieved an accuracy of 97.89% and an AUC of 99.46%. Feature importance analysis highlighted the role of intergenic and intron variants especially within chromosome 6 supporting known associations with the HLA region. Additional genes like PTPN22, LINC02789, MTCO3P1 and TSBP1-AS1 also emerged as significant contributors. The proposed system demonstrates high predictive performance and provides valuable insights into the genetic architecture of RA.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning-Enabled Rheumatoid Arthritis Classification Using GWAS Data: A Performance Analysis

  • 期刊论文

  • 发表日期：

    2025年10月13日

查看全文
  • A. Ezhil Grace,
  • R. Thandaiah Prabu,
  • S Rimlon Shibi,
  • R. Saravanakumar

摘要

Rheumatoid arthritis (RA) is a complex autoimmune disorder with a strong genetic basis. Early and accurate diagnosis remains challenging due to overlapping symptoms with other joint diseases. This research presents a machine learning-based framework for classifying RA using single nucleotide polymorphism data derived from genome-wide association studies. Datasets were collected for RA, osteoarthritis, psoriatic arthritis, juvenile idiopathic arthritis, and gout, and preprocessed. Random forest, extra trees, XGBoost, and bagging classifiers were trained and integrated through an optimized soft voting ensemble. The final model achieved an accuracy of 97.89% and an AUC of 99.46%. Feature importance analysis highlighted the role of intergenic and intron variants especially within chromosome 6 supporting known associations with the HLA region. Additional genes like PTPN22, LINC02789, MTCO3P1 and TSBP1-AS1 also emerged as significant contributors. The proposed system demonstrates high predictive performance and provides valuable insights into the genetic architecture of RA.