Predicting rapid progression in knee osteoarthritis: a novel and interpretable automated machine learning approach, with specific focus on young patients and early disease
Authors' conclusion
Our novel tool's transparency and reliability in predicting rapid knee OA progression distinguish it from conventional 'black-box' methods
Study context
- Population
- Patients with knee osteoarthritis (OA) in the OA Initiative: FNIH OA Biomarker Consortium (training and hold-out validation) and Pivotal Osteoarthritis Initiative MRI Analyses study (external validation)
- Outcome
- Rapid progression of knee OA (multiclass: pain and/or radiographic; and binary)
- Follow-up
- 2 years
Methodological detailsautomated machine learning, hold-out validation, external validation, subgroup analyses by age, sex and ethnicity, Web-based applications developed to visualise personalised predictions
- Automated machine learning
- Hold-out validation
- External validation
- Subgroup analyses by age, sex and ethnicity
- Web-based applications developed to visualise personalised predictions
Performance
As reported in the abstract
External validation performance of clinical-only models (AUC-PRC)
Multiclass predictions, AUC-PRC
0.727
95% CI 0.726 to 0.728
Binary predictions, AUC-PRC
0.764
95% CI 0.762 to 0.766
Further findings
Multiclass models in early-stage OA (subgroup), AUC-PRC
- AUC-PRC range
- 0.724-0.806
Binary models in patients younger than 60 (subgroup), AUC-PRC
- AUC-PRC range
- 0.617-0.693
Key predictors of progression
- Patient-reported outcomes and MRI features emerged as key predictors; subgroup differences noted
Original publication
Predicting rapid progression in knee osteoarthritis: a novel and interpretable automated machine learning approach, with specific focus on young patients and early disease. Ann Rheum Dis. Published 2 January 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →