Related Experiment Video
Updated: Apr 28, 2026
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Predicting primary aldosteronism: risk stratification-based and machine learning-based models
Jean-Baptiste de Freminville1,2,3, Laurence Amar1,4,5, Jean Feydy2,6
1Hypertension Unit, Vascular Medicine Department, Université Paris-Cité, AP-HP, Hôpital Européen Georges Pompidou, Trousseau Universitary Hospital, Paris, Chambray-lès-Tours 37170, France.
New algorithms improve screening for primary aldosteronism (PA) in hypertensive patients. A risk stratification tool, PAstrat, showed good sensitivity and interpretability, outperforming machine learning models in identifying secondary hypertension.
Area of Science:
- Endocrinology
- Cardiovascular Medicine
- Medical Informatics
Background:
- Screening for primary aldosteronism (PA) in hypertensive patients is challenging.
- Improved diagnostic algorithms are needed for efficient PA detection.
Purpose of the Study:
- To develop and compare novel algorithms for primary aldosteronism screening.
- To evaluate risk stratification and machine learning models for PA detection.
Main Methods:
- Developed PAstrat, a risk stratification algorithm.
- Created logistic regression and XGBoost machine learning models.
- Validated algorithms on derivation and external cohorts (15,507 and 768 patients).
Main Results:
- PAstrat, logistic regression, and XGBoost showed AUCs of 0.80-0.83.
- PAstrat demonstrated high negative predictive value (0.96).
- Machine learning models had lower negative predictive values in the validation cohort.
Conclusions:
- PAstrat offers a sensitive and interpretable tool for PA screening.
- Machine learning models lack explainability and showed lower performance in validation.
- PAstrat is a promising, user-friendly alternative for PA screening.
