acePromises and limitations of deep learning for predicting knee osteoarthritis progression from medical imaging: A systematic review
To systematically evaluate the performance, methodological quality, and translational barriers of deep learning (DL) models for predicting knee osteoarthritis (KOA) progression from medical imaging. Following PRISMA guidelines, we searched PubMed, Scopus, and Web of Science (inception to June 2026) for peer-reviewed studies applying DL to predict KOA progression from medical imaging. Two reviewers independently screened studies, extracted data, and assessed risk of bias using PROBAST-AI. The primary outcome was…
Deep learning models demonstrated proof-of-concept ability to predict knee osteoarthritis progression from medical imaging, with internal median AUCs up to 0.87 for surgical endpoints.
Deep learning models for knee osteoarthritis progression showed limited generalizability, with performance degradation on external validation and heavy reliance on a single training dataset without rigorous multi-site validation.
Review found substantial heterogeneity preventing meta-analysis, heavy reliance on single dataset, limited external validation, and high risk of bias in analysis domain, indicating models not ready for clinical deployment.
Evidence
- Peer-reviewedKnee Surgery, Sports Traumatology, Arthroscopy2026-08-03
How should this claim be treated?
Truvace Impact Record TRV-2026-0638, v1: “Promises and limitations of deep learning for predicting knee osteoarthritis progression from medical imaging: A systematic review.” Truvace, 2026-08-04. /record/TRV-2026-0638 (accessed at citation time). sha256 04bb6362d63e8e94…
Calibration history
Every change to this record since certification, in the open. None yet — the reading has held since it entered the record.
Certified into the record
How to verify without trusting this page
Fetch the canonical text of any version from /api/record/TRV-2026-0638 and hash it yourself — for example shasum -a 256 on the saved canonical field. The result must equal content_hash, and each version’s text ends with prev:followed by the prior version’s hash (version 1 chains to 64 zeros). If a single character of any version had been altered since certification, the chain would not reproduce.