The study of a new AI model that examines 30 years of routine electronic health record (EHR) data may improve screening for primary hyperaldosteronism, according to research presented Saturday at the Endocrine Society’s annual meeting ENDO 2026 in Chicago, Illinois. Primary hyperaldosteronism is a major cause of hypertension, often unrecognized, but increases patients’ risk of cardiovascular complications.
Primary hyperaldosteronism occurs when the adrenal glands (small glands at the top of each kidney) produce too much of the hormone aldosterone. This builds up aldosterone in the body, which helps balance sodium and potassium levels. People with primary hyperaldosteronism have a higher risk of cardiovascular disease than people with primary hypertension.
The true prevalence of primary hyperaldosteronism is unknown, but it is estimated that up to 20 percent of people with high blood pressure have primary hyperaldosteronism, said study principal investigator Frank Lee, M.D., of the Mayo Clinic in Rochester, Minnesota. Effective treatments exist for primary hyperaldosteronism, so early diagnosis can prevent future complications and reduce healthcare costs, Lee explained.
The Endocrine Society’s “Primary Hyperaldosteronism: Endocrine Society Clinical Practice Guidelines,” released in 2025, calls for more widespread screening for primary hyperaldosteronism. This cause of high blood pressure increases patients’ risk of cardiovascular complications such as stroke, coronary artery disease, atrial fibrillation, heart failure, and kidney disease.
Using de-identified data from more than 22,000 patients collected on the Mayo Clinic platform (a federated, privacy-preserving infrastructure with multimodal clinical data) between 1986 and 2025, researchers developed an AI screening model that analyzed variables such as age, gender, hypertension and hypokalemia-related ICD diagnoses, systolic blood pressure measurements, potassium blood levels, and prescribed antihypertensive or potassium supplement medications. They then tested data on 225,887 adults with hypertension. The XGBoost architecture, a machine learning library, predicted patients at risk for primary hyperaldosteronism 12 months before diagnosis.
Lee said this model showed that an AI-based approach to screening for primary hyperaldosteronism may be feasible. When researchers set thresholds to identify people at low risk, the model correctly flagged more than 90% of cases of primary hyperaldosteronism, but missed fewer than 10%. In this setting, approximately two-thirds of study participants were identified as candidates for screening.
During a study of hypertensive patients who had not previously been tested for primary hyperaldosteronism, our model identified approximately 2 out of 3 patients as candidates for further workup. Clinicians are challenged to effectively screen for primary hyperaldosteronism. The tools our team has developed have the potential to provide solutions based on routine information from patients’ medical records. ”
Dr. Frank Lee of the Mayo Clinic in Rochester, Minnesota.
