Machine Learning to Predict Remission Between 6 and 24 Months in Rheumatoid Arthritis: Insights From JAK, an International Registry Collaboration
Authors' conclusion
Using routinely collected baseline data across 11 registries, prediction of remission showed limited discrimination and was best suited to ruling out remission. Performance was similar for a 10-variable model and logistic regression, indicating limited incremental value of model complexity without richer predictors.
Study context
- Population
- Patients with rheumatoid arthritis (RA) initiating TNFi, JAKi, IL-6 inhibitors, abatacept or RTX; 11 international registries (JAK-pot collaboration)treatment courses: 21,675 training, 5,418 internal, 1,807 external validation
- Comparison
- Full XGBoost model vs simplified 10-predictor model vs logistic regression
- Outcome
- Remission (CDAI ≤2.8) within 6 to 24 months after treatment initiation or switching
- Follow-up
- 6 to 24 months
Methodological detailsmachine learning (XGBoost), 11 international registries, external validation (Switzerland), logistic regression benchmark
- Machine learning (XGBoost)
- 11 international registries
- External validation (Switzerland)
- Logistic regression benchmark
Performance
As reported in the abstract
Prediction of remission, external validation
| Test or model | AUC | Sensitivity | Specificity | PPV | NPV |
|---|---|---|---|---|---|
| Full model | 0.797 (95% CI 0.774-0.821) | 0.804 | 0.650 | 0.454 | 0.902 |
| Simplified model | 0.802 | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract |
| Logistic regression | 0.809 | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract | — not stated in the abstract |
AUC
Full model0.797 (95% CI 0.774-0.821)
Simplified model0.802
Logistic regression0.809
Sensitivity
Full model0.804
Simplified model— not stated in the abstract
Logistic regression— not stated in the abstract
Specificity
Full model0.650
Simplified model— not stated in the abstract
Logistic regression— not stated in the abstract
PPV
Full model0.454
Simplified model— not stated in the abstract
Logistic regression— not stated in the abstract
NPV
Full model0.902
Simplified model— not stated in the abstract
Logistic regression— not stated in the abstract
— not stated in the abstract
Further findings
Key predictors
- Patient global assessment, 28-tender joint count, previous b/tsDMARD exposure and HAQ-DI score
Original publication
Machine Learning to Predict Remission Between 6 and 24 Months in Rheumatoid Arthritis: Insights From JAK, an International Registry Collaboration. Arthritis Rheumatol. Published 7 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →