Related Experiment Videos
Incomplete Reporting Persists in Orthopedic Machine Learning Models: A Systematic Review of TRIPOD and TRIPOD+AI
R Harmen Kuijten1, Tom M de Groot2, Maarten A van Weezenbeek3
1Division of Imaging and Oncology, University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands.
Journal of Clinical Epidemiology
|July 31, 2026
Summary
Reporting for machine learning (ML) prognostic models in orthopedic surgery shows significant gaps. Adherence to TRIPOD+AI standards is low, highlighting a need for improved transparency in ML model development and reporting.
Area of Science:
- Orthopedic Surgery
- Medical Artificial Intelligence
- Biostatistics
Background:
- Machine learning (ML) models for predicting orthopedic surgery outcomes are rapidly increasing.
- Transparency and reporting standards for these models are crucial but not well-established.
- The TRIPOD+AI (2024) guidelines aim to standardize reporting for AI in clinical prediction models.
Purpose of the Study:
- To assess the characteristics and trends of preoperative ML prognostic models in orthopedic surgery.
- To evaluate the reporting completeness of these models according to TRIPOD+AI (2024) standards.
- To compare TRIPOD 2015 reporting completeness with previous systematic reviews.
Main Methods:
- Systematic review of studies developing or evaluating preoperative ML prognostic models in orthopedic surgery published up to December 31, 2024.
- Extraction of study characteristics and trends from 433 eligible studies.
- Assessment of TRIPOD+AI reporting completeness in 93 studies from top orthopedic journals, deriving TRIPOD 2015 completeness from these items.
Main Results:
- Nearly half of the studies were published in 2023-2024, indicating rapid growth.
- Significant concerns regarding sample size (27%) and limited model accessibility (13%) were noted.
- Median TRIPOD+AI completeness was 45%, with similar TRIPOD 2015 completeness in development studies but improvement in evaluation studies compared to prior reviews.
Conclusions:
- Substantial gaps exist in the reporting completeness of ML prognostic models in orthopedic surgery.
- Improvements are needed in sample size justification and performance reporting.
- Prioritizing TRIPOD+AI implementation, journal enforcement, and re-evaluation is essential for future research.