Machine Learning-based Classification of Rheumatoid Arthritis and Osteoarthritis Using Gene Expression Data

Bakhyt S. Atzhaxynova

2026

Full text Open access: Hybrid Metadata (CSL-JSON)
DOI 10.4103/wkpj.wkpj_24_26
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

A BSTRACT Background: Accurate classification of related diseases remains a major challenge in biomedical research, particularly when based on high-dimensional molecular data. Machine learning (ML) approaches offer promising solutions for improving disease discrimination and identifying relevant biomarkers. Materials and Methods: Gene expression data from synovial tissue samples, including rheumatoid arthritis (RA) and osteoarthritis (OA), were analyzed using supervised ML models. Logistic Regression and Random Forest classifiers were implemented and evaluated. Data preprocessing included variance filtering and dimensionality reduction. Model performance was assessed using accuracy, precision, recall, and area under the receiver operating characteristic (ROC) curve (AUC). Feature importance analysis was performed to identify key gene probes contributing to classification. Results: The Random Forest model demonstrated superior performance, achieving near-perfect classification on the test dataset (accuracy, precision, recall, and AUC = 1.000), while Logistic Regression showed moderate performance (accuracy = 0.714, AUC = 0.750). Despite identical ROC AUC values (1.000) in graphical analysis, confusion matrices revealed classification errors in the Logistic Regression model. Feature importance analysis identified several gene probes, including 209374_s_at and 203786_s_at, as major contributors to disease discrimination. Conclusion: These findings highlight the effectiveness of Random Forest for classifying RA and OA based on gene expression data and suggest that identified gene probes may serve as potential molecular markers. The study supports the application of ML in biomedical research and emphasizes the need for validation in larger, independent cohorts.