Machine learning identifies a profile of inadequate responder to methotrexate in rheumatoid arthritis
Authors' conclusion
Our study highlights the usefulness of machine learning in unveiling subgroups of inadequate responders to MTX to guide new therapeutic strategies.
Study context
- Population
- RA patients fulfilling the 2010 ACR/EULAR criteria starting MTX, ESPOIR and Leiden EAC cohorts (training) and tREACH (external validation)493 + 239 + 138 therapeutic sequences
- Outcome
- EULAR response to MTX (prediction of inadequate response)
- Follow-up
- 9 months (± 6 months)
Methodological detailsmachine learning, Shapley values, external validation cohort (tREACH)
- Machine learning
- Shapley values
- External validation cohort (tREACH)
Performance
As reported in the abstract
Model AUROC for EULAR inadequate response, external validation set
AUROC on external validation set
0.72 [95% CI (0.63, 0.80)]
Responders to MTX in external validation set
70%
Chance to respond in patients predicted as inadequate responders
38% [95% CI (20%, 58%)]
Inadequate-response rate using the algorithm to decide MTX initiation (vs 30%)
23% [95% CI: (17%, 29%)]
Further findings
Biomarker in moderate or high activity (DAS28 > 3.2)
- Lymphocyte count superior to 2000 cells/mm³: significantly less likely to respond
Original publication
Machine learning identifies a profile of inadequate responder to methotrexate in rheumatoid arthritis. Rheumatology (Oxford). Published 23 November 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →