Related Experiment Video
Updated: Nov 12, 2025
07:15
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
7.1K
Machine learning prediction models in orthopedic surgery: A systematic review in transparent reporting
Olivier Q Groot1, Paul T Ogink2, Amanda Lans1
1Orthopedic Oncology Service, Department of Orthopedic Surgery, Massachusetts General Hospital, Harvard Medical School, Boston, Massachusetts, USA.
Summary
Machine learning (ML) studies in orthopedics often lack transparent reporting and have a high risk of bias. Over half of studies incompletely report methods or performance, hindering clinical trust and implementation.
Area of Science:
- Orthopedic Surgery
- Medical Informatics
- Biostatistics
Background:
- Machine learning (ML) studies are rapidly increasing in orthopedic surgery.
- Critical appraisal of adherence to peer-reviewed guidelines for these studies is lacking.
- This gap hinders the reliable implementation of ML in clinical practice.
Purpose of the Study:
- To evaluate the quality and transparent reporting of ML prediction models in orthopedic surgery.
- To assess the risk of bias in these ML models using established tools.
Main Methods:
- Systematic review of ML prediction studies in orthopedic surgery published up to June 18th, 2020.
- Screening of 7138 studies, with 59 included for analysis.
- Data extraction and risk of bias assessment performed by multiple reviewers using TRIPOD and Prediction model Risk Of Bias ASsessment Tool (PROBAST) criteria.
Main Results:
- Median completeness for the Transparent Reporting Of a multivariable model by Individual Prognosis Or Diagnosis (TRIPOD) checklist was 53%.
- 41% of studies exhibited a high risk of bias, driven by incomplete reporting, inadequate handling of missing data, and small datasets.
- Over 50% of studies incompletely reported methods or performance measures.
Conclusions:
- A significant proportion of ML prediction models in orthopedics suffer from high risk of bias and incomplete reporting.
- These quality issues create a gap between ML model development and their trustworthy implementation in orthopedic practice.
- Addressing reporting standards and bias is crucial for clinical adoption.
