Researchml predictionxgboostjoint arthroplastyexternal validation
Machine Learning Evaluates VTE Risk After Arthroplasty
7.2
A systematic review in JMIR Medical Informatics (2026) assessed 34 machine learning prediction models from nine studies through searches to December 15, 2024. The authors found high heterogeneity and pervasive risk of bias; several models reported AUCs >0.9 but lacked external validation and likely overfit. They call for prospective designs, robust data handling, and external validation to improve clinical applicability.
Key Points
- 1Identifies 34 ML prediction models across nine studies, commonly XGB and logistic regression
- 2Finds high heterogeneity and pervasive high risk of bias undermining reported model performance
- 3Recommends prospective designs, robust data handling, and external validation for clinical applicability
Scoring Rationale
Systematic synthesis with PROSPERO registration and peer-reviewed source, but scope limited to nine heterogeneous studies
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
