Predicting cardiovascular death in overweight/obese people with prediabetes using machine learning—a proof-of-concept study
Authors' conclusion
We have developed a prediabetes-specific proof-of-concept model that predicts whether a person is at high risk of cardiovascular death. External validation of the model is crucial before adoption to a real-world setting to clarify whether the model generalizes beyond the studied population.
Study context
- Population
- Participants ≥45 years with prediabetes (HbA1c 39–47 mmol/mol (5.7–6.4%)), established CVD and overweight/obesity182 (3.2%) cardiovascular (CV) deaths
- Participants
- 5,636 participants
- Comparison
- Prediabetes-specific model vs SCORE2 and SCORE2-Diabetes benchmark models
- Outcome
- Prediction of CV death (ROC AUC, precision and recall)
Methodological detailsproof-of-concept, binary logistic regression, stratified threefold cross-validation, calibrated risk estimates, Mean time-to-event 2.0 years; mean trial duration 3.3 years
- Proof-of-concept
- Binary logistic regression
- Stratified threefold cross-validation
- Calibrated risk estimates
- Mean time-to-event 2.0 years; mean trial duration 3.3 years
Performance
As reported in the abstract
CV death discrimination (ROC AUC)
| Test or model | Value | ROC AUC |
|---|---|---|
| Our model | 0.730 (95% CI 0.659–0.801) | 0.730 |
| SCORE2 | 0.630 | 0.630 |
| SCORE2-Diabetes | 0.643 | 0.643 |
Value
Our model0.730 (95% CI 0.659–0.801)
SCORE20.630
SCORE2-Diabetes0.643
ROC AUC
Our model0.730
SCORE20.630
SCORE2-Diabetes0.643
Further findings
Major adverse cardiovascular events (MACE) prediction by established models in this cohort (ROC AUC)
- Lower performance in our cohort (ROC AUC) vs original populations (ROC AUC 0.739 and 0.66-0.73)
- 0.596 and 0.603
Original publication
Predicting cardiovascular death in overweight/obese people with prediabetes using machine learning—a proof-of-concept study. Cardiovasc Diabetol. Published 17 May 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →