Predicting total knee replacement in knee osteoarthritis using a machine learning-guided approach in patients of the Osteoarthritis Initiative (OAI)
Authors' conclusion
We present a pragmatic, non-imaging, ML-informed model that predicts TKR with clinically acceptable discrimination and calibration using routinely collected data.
Study context
- Population
- Osteoarthritis Initiative participants with knee OA
- Participants
- 3,790 participants
- Outcome
- Prediction of total knee replacement (TKR) (discrimination: AUC; calibration: Brier score)
Methodological detailsMachine learning (ML)-guided feature selection, multivariable logistic regression, prespecified analysis by Kellgren-Lawrence (KL) grade
- ML-guided feature selection
- Multivariable logistic regression
- Prespecified analysis by KL grade
Performance
As reported in the abstract
TKR prediction, final logistic model (full cohort)
AUC
0.765
95% CI 0.736 to 0.793
Brier score (acceptable calibration)
0.097
Further findings
Classifier AUCs on test set (Elastic Net and XGBoost best)
- AUC range
- 0.716-0.748
Discrimination by baseline KL grade (prespecified)
- KL 0-1: AUC
- 0.827
- KL ≥2: AUC
- 0.720
- Decile plots indicated broadly aligned observed vs predicted risks
Incremental value of Polygenic Risk Score (PRS)
- PRS added modest, statistically significant gains in specificity at several fixed sensitivities without materially changing AUC
- Bonferroni-adjusted McNemar tests
Original publication
Predicting total knee replacement in knee osteoarthritis using a machine learning-guided approach in patients of the Osteoarthritis Initiative (OAI). RMD Open. Published 10 February 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →