AI Model Identifies Hidden Cause of High Blood Pressure, Years Before Diagnosis
A new artificial intelligence tool could transform how doctors screen for primary aldosteronism (PA), a common but frequently overlooked cause of high blood pressure. The research was presented at ENDO 2026, the Endocrine Society's annual meeting in Chicago.
The Problem: A Missed Diagnosis
- Primary aldosteronism occurs when the adrenal glands produce too much aldosterone, leading to hypertension.
- It is estimated to affect up to 20% of hypertensive patients, yet the majority of cases remain undiagnosed.
- Left untreated, PA significantly increases the risk of heart attack, stroke, and other cardiovascular complications.
The AI Solution: Predicting Risk from Routine Data
The screening model was developed using de-identified data from over 22,000 patients (spanning 1986 to 2025) from the Mayo Clinic Platform. It was then validated on a separate group of 225,887 adults with hypertension.
Key variables used by the model include:
- Age, sex, and existing diagnoses of hypertension and hypokalemia (low potassium)
- Blood pressure readings and potassium levels
- Current medication history
A Powerful Performance Metric
The XGBoost-based model demonstrated remarkable predictive capability: it could identify a patient's risk of PA up to 12 months before clinical diagnosis.
When configured to identify low-risk patients, the model flagged over 90% of PA cases while missing fewer than 10%.
This performance is a significant step forward, as it would allow clinicians to use the tool as a routine screening step, simply by analyzing existing electronic medical records.
Practical Impact for Doctors and Patients
By designating roughly two-thirds of hypertensive participants as candidates for screening, the model efficiently narrows the population that requires further testing.
"This tool could help clinicians screen effectively using routine medical records," said lead researcher Frank Lee, M.D., of Mayo Clinic.
Why It Matters
- Earlier diagnosis of PA can lead to targeted treatments, reducing the risk of cardiovascular complications.
- Lower healthcare costs are anticipated, as late-stage management of PA-related heart and kidney disease is expensive.
- The AI approach is non-invasive, using only data already collected in standard medical visits.