Early prediction of bone destruction in rheumatoid arthritis through machine learning analysis of plasma metabolites
Performance · Rapid radiographic progressionResult — go to the resultOdds ratio 0.01 95% CI 0.01–0.03Authors' conclusion
The CjBM was developed for early prediction of bone destruction in patients with RA, and the evaluation of BERS emphasizes the significance of metabolite features.
Study context
- Population
- Patients with rheumatoid arthritis (RA); baseline plasma metabolites by liquid chromatograph-mass spectrometer
- Participants
- 60 patients
- Comparison
- High vs low bone erosion risk score (BERS); CjBM vs reported prediction models
- Outcome
- Radiographic progression (bone destruction) by van der Heijde-modified total Sharp score (mTSS)
- Follow-up
- one-year follow-up
Methodological detailsmachine learning, prospective cohort, 1-year follow-up
- Machine learning
- Prospective cohort
- 1-year follow-up
Performance
Rapid radiographic progression
Odds ratio
0.01
95% CI 0.01–0.03
P = 0.003
High BERSvs Lower BERS
OR 0.01 (95% CI 0.01–0.03); P = 0.003
As reported in the abstract
- CjBM (BERS + MTX and CRP) AUC
- 0.800
- Adding BERS improves discrimination of original reported model
- PDeLong=0.035
Further findings
mTSS change over 1-year follow-up · one-year follow-up
- Median increase in mTSS
- 3.50
Plasma metabolites and BERS for rapid radiographic progression
- Metabolites detected
- 774
- Metabolites differing between patients with different outcomes
- 77
- Core metabolites identified by Gaussian Naive Bayes: mangiferic acid, O-acetyl-L-carnitine, an eicosatrienoic acid and 16-methylheptadecanoic acid; BERS built on these features
Comparison with reported prediction models
- Adding BERS improved discrimination of the original reported model
- PDeLong=0.035
- CjBM vs reported models in identifying rapid radiographic progression: near statistical significance
Original publication
Early prediction of bone destruction in rheumatoid arthritis through machine learning analysis of plasma metabolites. Arthritis Res Ther. Published 21 May 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →