Predicting primary aldosteronism: risk stratification-based and machine learning-based models
Authors' conclusion
Our algorithm demonstrated good sensitivity and negative predictive value for the screening of secondary hypertension, with the added value of great interpretability. Machine learning models lacked explainability and demonstrated a lower negative value in the validation cohort which makes them less easily usable.
Study context
- Population
- Patients aged 18–65 years with confirmed hypertension and full workup for primary aldosteronism (PA) screeningvalidation cohort n = 768
- Participants
- 15,507 patients
- Comparison
- PAstrat risk stratification algorithm vs logistic regression vs XGBoost
- Outcome
- AUC of ROC curves for PA prediction
Methodological detailsexternal validation cohort
- External validation cohort
Performance
As reported in the abstract
| Test or model | AUC, derivation cohort | AUC, validation cohort | NPV |
|---|---|---|---|
| PAstrat | 0.80 | 0.79 | 0.96 |
| Logistic regression | 0.82 | 0.82 | 0.87 |
| XGBoost | 0.83 | 0.82 | 0.88 |
AUC, derivation cohort
PAstrat0.80
Logistic regression0.82
XGBoost0.83
AUC, validation cohort
PAstrat0.79
Logistic regression0.82
XGBoost0.82
NPV
PAstrat0.96
Logistic regression0.87
XGBoost0.88
Further findings
False negatives, validation cohort
Share of patients with PA missed PAstrat 4.2% (6 of 142)
- False negatives, validation cohort: PAstrat: 6 of 768
PA among patients with screening indication (age, kalaemia or drug resistance) but not per algorithm
- Patients with PA among 128
- 3 (0.23%)
Original publication
Predicting primary aldosteronism: risk stratification-based and machine learning-based models. Eur J Prev Cardiol. Published 27 April 2026. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →