AI Tool Screens for Primary Aldosteronism Using Routine EHR Data

Underdiagnosis can delay targeted treatment, increasing downstream hospitalizations and costs. To support earlier detection, researchers have developed anA I model that uses electronic health record data to screen large hypertensive populations and identify patients at risk.

Developed at Mayo Clinic, the AI screening model was presented at ENDO 2026, the Endocrine Society’s annual meeting in Chicago. The tool examines longitudinal electronic health record data to estimate risk for primary aldosteronism among adults with hypertension. It was built using the Mayo Clinic Platform, a federated, privacy‑preserving infrastructure that houses multimodal clinical data.

The system evaluated routinely captured variables including age, sex, International Classification of Diseases diagnoses for hypertension and hypokalemia, systolic blood pressure, serum potassium, and prescriptions for antihypertensives or potassium supplements. Researchers used an Extreme Gradient Boosting (XGBoost) architecture to generate risk predictions up to 12 months before diagnosis. The approach relies on information already present in the medical record.

Investigators developed the model on de-identified records from more than 22,000 patients gathered between 1986 and 2025. They then tested performance in a separate dataset of 225,887 adults with hypertension. The analysis assessed feasibility of large-scale screening using three decades of routine care data.

When tuned to exclude low-risk individuals, the model correctly flagged more than 90% of primary aldosteronism cases while missing fewer than 10%. In that configuration, approximately two-thirds of patients were identified as candidates for further screening. The Endocrine Society’s 2025 clinical practice guideline calls for more widespread screening for this hormone-driven cause of high blood pressure.

“During testing on patients with high blood pressure who had never been screened previously for primary aldosteronism, our model identified approximately two out of every three patients for further work-up,” said Frank Lee, M.D., lead researcher, Mayo Clinic in Rochester, Minnesota. “Clinicians have been challenged to screen primary aldosteronism effectively. The tool developed by our team could offer a solution based on routine information available in a patient's medical records.” 

Related Links
Mayo Clinic
Endocrine Society