Comparative predictive accuracy of machine learning versus traditional statistical methods for diabetes‐related complications: A systematic review and analysis
Authors' conclusion
Machine learning models, particularly ensemble methods, offer modest discrimination improvements over traditional statistics for predicting diabetes-related complications. However, widespread methodological limitations, specifically the lack of external validation, inconsistent calibration reporting and high bias, substantially limit our confidence and clinical readiness. Rigorous external validation and transparent reporting are needed before routine implementation.
Study context
- Population
- Studies developing or validating prediction models for long-term complications in people with diabetes; MEDLINE, PubMed, Cochrane and Scopus (2014–2025)36 studies · 280 comparisons
- Comparison
- Machine learning models vs traditional statistical models
- Random forestn = 68
- Extreme gradient boostingn = 24
- Support vector machinesn = 55
- Outcome
- Predictive accuracy (primarily C-statistics) of ML vs traditional statistical models
Methodological detailscase-control studies excluded, risk of bias assessed with PROBAST
- Case-control studies excluded
- Risk of bias assessed with PROBAST
Pooled result
As reported in the abstract
Comparisons with higher discrimination vs logistic regression
Further pooled results
Methodological quality
- External validation reported
- 8% (3/36) of studies
- Calibration reported
- 13% (5/36)
- High risk of bias
- 59% of studies
Original publication
Comparative predictive accuracy of machine learning versus traditional statistical methods for diabetes‐related complications: A systematic review and analysis. Diabet Med. Published 18 June 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →