AI-Powered Analysis of Health Records Uncovers Hidden Hypertension Cases

AI Uncovers Hidden Hypertension in Electronic Health Records

Researchers at Mass General Brigham have made a significant breakthrough in the detection and treatment of hypertension using artificial intelligence (AI) to analyze electronic health records (EHR). The study, published in JAMA Cardiology and presented at the 2025 American College of Cardiology's Annual Scientific Session & Expo, demonstrates how natural language processing can identify patients at risk of hypertension who might otherwise go undiagnosed 1

1

2

2

.

The Silent Killer: Undiagnosed Hypertension

Hypertension, often called the "silent killer," affects nearly half of adults in the United States, with many unaware of their condition. Dr. Jason H. Wasfy, senior author of the study, emphasizes the importance of early detection: "If they're not getting checked for it enough, then high blood pressure can damage the heart and the vessels over time in a way that would have been preventable had the blood pressure been detected early" 1

1

.

AI-Powered Detection Method

The research team developed an AI algorithm using natural language processing to analyze echocardiogram data from EHRs. The algorithm specifically looked for cases of left ventricular hypertrophy, a thickening of the heart muscle often caused by hypertension 2

2

. This innovative approach allowed researchers to identify patients who were not previously known to have heart muscle problems or hypertension.

Clinical Trial Design and Results

The study involved 648 patients with an average age of 59, of whom 38% were women. Half of the patients were randomly selected for the intervention group, where their physicians were notified of the AI-detected findings 1

1

. The results were striking:

• Patients in the intervention group were nearly four times more likely to receive new hypertension diagnoses (15.6% vs. 4.0%)
• Antihypertensive medication prescriptions were also significantly higher in the intervention group (16.3% vs. 5.0%)

Physician Response and Implementation

Importantly, the study was designed to be minimally disruptive to clinical workflows. Dr. Wasfy noted, "Clinicians are often overloaded with alerts that can cause fatigue and burnout, so we intentionally designed our outreach to be delivered by a person" 1

1

. This approach was well-received, with 72% of responding clinicians reacting positively to the intervention 2

2

.

Future Implications and Challenges

While the study demonstrates the potential of AI in improving healthcare delivery, lead author Dr. Adam Berman acknowledges that more work is needed to determine how this approach can be scaled and implemented in various healthcare settings 1

1

. The ultimate goal, according to Dr. Berman, is to "augment traditional care, using the data that already exist" 2

2

.

This groundbreaking research highlights the untapped potential of existing medical data and how AI can be leveraged to improve patient outcomes. As healthcare systems continue to digitize and accumulate vast amounts of data, studies like this pave the way for more proactive and efficient healthcare delivery, potentially saving lives by catching silent but dangerous conditions like hypertension before they cause irreversible damage.