Currently submitted to: JMIR Medical Informatics
Date Submitted: Mar 14, 2026
Open Peer Review Period: Mar 26, 2026 - May 21, 2026
(currently open for review)
Warning: This is an author submission that is not peer-reviewed or edited. Preprints - unless they show as "accepted" - should not be relied on to guide clinical practice or health-related behavior and should not be reported in news media as established information.
Machine-Learning Approaches to Detecting Implant Loosening and Failure in Hip and Knee Arthroplasty: A Scoping Review
ABSTRACT
Background:
Aseptic loosening remains a leading cause of late failure after total hip and knee arthroplasty. Radiographic signs are often subtle or delayed, and definitive diagnosis frequently relies on intraoperative findings at revision. Machine learning has been proposed as a tool to assist in detecting loosening or implant failure, but its clinical role and methodological robustness remain unclear.
Objective:
To understand if machine learning algorithms can help detect implant loosening and the need for revision in TKA/THA
Methods:
We conducted a scoping review of studies applying machine learning to detect implant loosening or mechanical failure following total hip or knee arthroplasty. Studies using imaging or mechanically derived data with a loosening or failure related endpoint were included. Data were extracted on joint studies, imaging modality, endpoint definition, modelling approach, and validation strategy.
Results:
Eight studies published between 2019 and 2025 met inclusion criteria. Most focused on total hip arthroplasty and relied on plain radiographs. Definitions of loosening varied substantially, including revision confirmed mechanical instability, composite prosthesis failure, expert interpreted radiographic suspicion, and quantified component displacement. Modelling approaches ranged from static image classification to longitudinal prediction and deep learning enabled automation of biomechanical measurement. Internal validation predominated, with external validation uncommon and reporting of sample size considerations, missing data, and reproducibility frequently limited.
Conclusions:
Machine learning can identify patterns associated with implant loosening or failure under controlled conditions, but current applications model heterogeneous constructs that do not align uniformly with mechanical diagnosis. Existing evidence supports a role for machine learning as decision support for triage or risk stratification rather than definitive diagnosis. Broader external validation, clearer alignment between clinical intent and modelled endpoints, and improved reporting will be required before reliable integration into routine arthroplasty practice. Clinical Trial: N/A
Citation
Request queued. Please wait while the file is being generated. It may take some time.
Copyright
© The authors. All rights reserved. This is a privileged document currently under peer-review/community review (or an accepted/rejected manuscript). Authors have provided JMIR Publications with an exclusive license to publish this preprint on it's website for review and ahead-of-print citation purposes only. While the final peer-reviewed paper may be licensed under a cc-by license on publication, at this stage authors and publisher expressively prohibit redistribution of this draft paper other than for review purposes.