Identification of individuals at high risk of developing rheumatoid arthritis: a balanced random forest model in a cohort of 1544 first-degree relatives
Authors' conclusion
Overall, the results demonstrate that predicting RA by combining genetics, serological biomarkers, epidemiological risk factors and clinical signs is promising, although model generalisation remains challenging.
Study context
- Population
- First-degree relatives of rheumatoid arthritis (RA) patients (RA-FDRs), prospective SCREEN-RA cohort
- Participants
- 1,540 RA-FDRs
- Outcome
- Development of RA
Methodological detailsbalanced random forest models, fivefold cross-validation, classification threshold targeting high sensitivity
- Balanced random forest models
- Fivefold cross-validation
- Classification threshold targeting high sensitivity
Main finding
As reported in the abstract
Prediction of RA development (BRF model) · mean follow-up 7.1 years
Participants who developed RA
27
Balanced random forest (BRF) sensitivity (high), with modest specificity; moderate predictive performance
≥0.85
Observational data: associations, not effects.
Secondary outcomes
Seropositive inflammatory arthritis (IA) (exploratory) · mean follow-up 7.1 years
- Participants who developed seropositive IA
- 126
- Age, gender and anti-RA33 autoantibodies were the main predictor variables
- Exploratory analyses
Original publication
Identification of individuals at high risk of developing rheumatoid arthritis: a balanced random forest model in a cohort of 1544 first-degree relatives. RMD Open. Published 27 November 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →