A new study conducted by researchers at Mass General Brigham has revealed that vital clues about hypertension may be embedded within electronic health records (EHRs). By utilizing natural language processing, a form of artificial intelligence, scientists identified patients who had undergone heart ultrasounds showing congealing of the heart muscle—an indicator frequently linked to hypertension. When physicians were alerted to these findings, they were nearly four times more likely to diagnose hypertension and prescribe medications for blood pressure management. This study underscores the potential of innovative, automated approaches that leverage existing electronic health data to enhance patient care for cardiovascular conditions. The results were published in JAMA Cardiology and presented at the 2025 American College of Cardiology’s Annual Scientific Session & Expo.
“Hypertension is often called the silent killer because individuals can have dangerously high blood pressure without any symptoms,” said senior author Jason H. Wasfy, MD, MPhil, of the Cardiology Division, Department of Medicine at Massachusetts General Hospital (MGH), a founding member of the Mass General Brigham healthcare system. Wasfy, who is also a physician investigator at the Mongan Institute at MGH, added, “If left undiagnosed and untreated, high blood pressure can progressively damage the heart and blood vessels—harm that could have been prevented with early detection.”
In the United States, nearly half of individuals with hypertension remain undiagnosed or untreated.
“Routine clinical care generates an enormous amount of data, including information from doctor visits and diagnostic tests. Often, subtle indicators of hypertension are buried within these records, making it difficult for clinicians to recognize them all. Our study aimed to validate methods to uncover these hidden signals and improve patient care,” explained lead author Adam Berman, MD, MPH, who conducted the study while at Brigham and Women’s Hospital. At the time, Berman was the David F. Torchiana Fellow in Health Policy and Management at the Massachusetts General Physicians Organization. He is now an assistant instructor in the Department of Medicine, Leon H. Charney Division of Cardiology at NYU Grossman School of Medicine.
The research team developed and implemented a natural language processing algorithm capable of scanning echocardiogram (heart ultrasound) data to detect left ventricular hypertrophy, a thickening of the heart muscle often associated with hypertension. The procedure identified 648 patients at Mass General Brigham who had no prior history of heart muscle abnormalities and were not receiving action for hypertension. The average patient age was 59 years, and 38% were women. Researchers randomized half of the patients to receive an intervention in which a population health coordinator notified their doctors of the findings. These physicians were also provided with resources for further care, such as 24-hour blood pressure monitoring or a cardiology consultation. The control group received standard care, with their clinicians not being notified of the findings.
Patients in the intervention group were nearly four times more likely to receive new hypertension diagnoses (15.6% vs. 4.0%) and to be prescribed antihypertensive medication (16.3% vs. 5.0%) compared to those in the control group. Interestingly, there was no significant difference in the number of follow-up visits with primary care physicians between the two groups. Clinicians largely viewed the intervention favourably—of the 82% who responded to the initial notification, 72% had a positive reaction, as determined by qualitative scoring.
This study highlights the power of AI-driven analysis of EHRs in advancing preventive healthcare. By uncovering hidden indicators within routine medical records, physicians can make earlier diagnoses and implement timely interventions, ultimately reducing the long-term impact of hypertension on public health