Home›Diabetes & Endocrinology› AI prescreening tool identifies primary aldosteronism with AUROC 0.709 in hypertension
AI prescreening tool identifies primary aldosteronism with AUROC 0.709 in hypertensionAI tool helps identify patients with high blood pressure condition
medRxivPublished October 9, 2026Study authors: Lee, F. G.; Choudhary, A.; Li-Han, L. Y.; Varghese, C.; Larson, E. L.; Perry, W. R. G.; Bancos, I.; …DOI ↗Editorial oversight: Dr. Amelia Tan, PhD · Internal Medicine & Chronic Disease
AI-generated summary of the cited source, checked by automated accuracy review.
How we work
Share
Key Takeaway
Consider AI prescreening as a triage aid for PA, but confirm diagnosis with standard testing.
This retrospective cohort study evaluated an AI-based prescreening tool for primary aldosteronism (PA) in adults with hypertension, an ICD diagnosis of PA, or who had plasma renin/aldosterone testing. The tool used an ensemble architecture (eXtreme Gradient Boosting, random forest, and extremely randomized trees) trained on electronic health record data from a single institution spanning community clinics and tertiary/quaternary centers. The training/evaluation set included 22,264 patients, and a separate test cohort included 1,279,455 patients.
The primary outcome was prediction of PA diagnosis (disease) or negative workup (control). The model achieved an AUROC of 0.709 (95% CI 0.688-0.730), with an estimated calibration error (ECE) of 0.037. Specificity was 71.6%, and false positivity was 28.4%. The tool identified 5,295 patients for first-pass screening, representing 0.4% of the cohort.
Safety and tolerability outcomes were not reported. The study did not report follow-up duration, comparator, funding, or conflicts of interest. Limitations were not reported. As a retrospective cohort study, the findings are associational and do not establish causality. The model is a prediction tool, not a clinical intervention, and its performance in prospective clinical settings remains uncertain.
Practice relevance: The model could provide a pre-screening PA pathway using routine EHR data to identify and prioritize patients for PA screening, potentially facilitating adoption of current guidelines. However, given the modest AUROC and false positivity rate, clinicians should interpret the output cautiously and not rely on it as a standalone diagnostic tool.
How this fits prior evidence
This retrospective cohort study extends prior coverage of hypertension and aldosterone-related conditions by focusing on pre-screening for primary aldosteronism using an AI tool. Prior items covered aquatic exercise therapy for blood pressure reduction in elderly patients, mineralocorticoid receptor antagonists for cardioprotection via coronary microvascular function, and selective aldosterone synthase inhibitors reducing systolic blood pressure in CKD. The current study does not test a treatment but rather a prediction model, addressing a gap in early identification of PA. The AUROC of 0.709 and specificity of 71.6% suggest moderate discrimination, which contrasts with the more definitive blood pressure reductions reported for pharmacological interventions in prior coverage.
Managing high blood pressure is a constant challenge for doctors. Sometimes, a specific underlying condition called primary aldosteronism makes it even harder to treat. Because this condition can be tricky to spot, a new AI tool was tested to see if it could help doctors identify which patients need more intensive testing.
The tool looked at data from over 1.2 million patients. It used an ensemble architecture, which is a way of combining different computer models to make a more accurate prediction. The system was able to identify 5,295 patients who could be prioritized for initial screening based on their existing medical records.
While the tool shows promise in helping doctors prioritize their workload, it is important to remember that this is a prediction model, not a replacement for a doctor's judgment. It works by finding patterns in data to flag patients who might need a closer look, helping clinics follow standard guidelines more effectively.
What this means for you:
An AI tool can help doctors identify patients with high blood pressure who may need specific hormone testing.
Common questions
How does the AI tool work to find this condition?
The tool uses an ensemble architecture, which combines three different types of machine learning models. It looks at existing electronic health records to predict which patients with high blood pressure might have primary aldosteronism, helping doctors decide who needs more testing.
How accurate was the tool at identifying the condition?
The tool showed a 71.6% specificity rate, meaning it was fairly accurate at identifying the condition. It also had a 28.4% false positivity rate, and it successfully identified 5,295 patients for first-pass screening.
Is this tool a replacement for a doctor's diagnosis?
No, this is a prediction tool, not a clinical intervention. It is designed to help doctors prioritize which patients should be sent for screening based on their records, but it does not replace the need for a doctor's evaluation.
Context Current guidelines for primary aldosteronism (PA) recommend screening all patients with hypertension. Given low screening rates, an automated artificial intelligence (AI)-based risk-stratified approach may facilitate guideline implementation. Objective To develop an AI-based prescreening tool using electronic health record (EHR) data to identify at-risk patients. Design Model development retrospective cohort study. Ensemble architecture using eXtreme Gradient Boosting, random forest, and extremely randomized trees was trained on EHR data (1986-2025) and evaluated on a test set of patients with hypertension. Setting Single institution system spanning community clinics and tertiary/quaternary centers reported via Mayo Clinic Platform. Patients Adults with hypertension, International Classification of Disease (ICD) diagnosis of PA, or had plasma renin/aldosterone testing. Main Outcome Measures Predicting PA diagnosis (disease) or negative workup (control) using age, sex, ICD diagnoses, vitals, labs, and medications by area under receiver operating characteristic (AUROC), estimated calibration error (ECE), and specificity. Results Of 22,264 patients (1,833 disease and 20,431 controls), median time from initial encounter to screening was 5.0 years (IQR 0.5-12.6). The model demonstrated AUROC 0.709 (95% CI 0.688-0.730) with ECE 0.037. At a 0.5 cut-off threshold, the model had high specificity (71.6%) and low false positivity (28.4%). In a test cohort of 1,279,455 adults with hypertension, the model identified 5,295 (0.4%) for first-pass screening. Conclusions A high-specificity model can provide a pre-screening PA pathway using routine EHR data to identify and prioritize patients for PA screening. This AI-based approach can risk-stratify patients, guide sequencing of testing/referrals, and pragmatically facilitate adoption of current guidelines into practice.
