A study conducted by researchers at Mass General Brigham has identified that electronic health records (EHR) may contain significant information regarding hypertension, indicating a possible shift in the early diagnosis and treatment of heart conditions. Researchers employed natural language processing—an advanced form of artificial intelligence—to analyse echocardiogram data and identify patients showing signs of left ventricular hypertrophy, a condition characterised by the thickening of the heart muscle often associated with high blood pressure. 

The research examined individuals who underwent heart ultrasounds indicating abnormal thickening, despite lacking prior diagnoses of heart muscle issues or treatment for hypertension. A research team created an algorithm that analysed the electronic health records of 648 patients at Mass General Brigham. The patients in this study had an average age of 59 years, with women representing 38% of the total group. Fifty percent of the patients were assigned to a group that received an intervention in which a population health coordinator quickly informed their physicians about the ultrasound results. The control group was provided with standard care, which did not include any proactive notification.

Patients in the intervention group exhibited a diagnosis rate of hypertension that was nearly four times higher than that of the control group, with figures of 15.6% compared to 4.0%. Additionally, the intervention group had a higher likelihood of receiving antihypertensive medication prescriptions, at 16.3%, compared to 5.0% in the control group. The frequency of follow-up appointments with primary care physicians was similar between the two groups, despite the differences in diagnosis and treatment. 

Senior author Jason H. Wasfy, MD, MPhil, from the Cardiology Division at Massachusetts General Hospital, highlighted the importance of early detection: “Hypertension is known as the silent killer because individuals may have elevated blood pressure without exhibiting any symptoms.” Insufficient monitoring of high blood pressure can lead to long-term damage to the heart and blood vessels, a condition that could have been avoided with early detection. 

Lead author Adam Berman, MD, MPH, highlighted the unexplored opportunities present in routine clinical data. He stated, “A significant amount of information is produced during standard clinical care, such as during a doctor's visit or when undergoing a test.. Subtle clues within this information may suggest that a patient is experiencing hypertension. However, it is not feasible for clinicians to fully comprehend the entirety of the medical record. The objective of our trial was to investigate the possibility that the data may be readily accessible, and we aimed to assess methods for uncovering it to enhance patient care. 

This study, published in JAMA Cardiology and presented at the 2025 American College of Cardiology’s Annual Scientific Session & Expo, has garnered considerable interest among medical professionals. The intervention garnered a positive response, with 82% of clinicians indicating their approval of the initial notifications and 72% rating the outreach favourably. 

Dr Wasfy stated that the team expressed significant interest in ensuring that this initiative would be valuable to both physicians and patients. Clinicians frequently experience an overwhelming number of alerts, which can lead to fatigue and burnout. Therefore, we purposely designed our outreach for human conduct. 

This approach represents a significant advancement in the utilisation of digital health data aimed at enhancing patient outcomes. While the results show potential, additional research is necessary to investigate scalable approaches for wider implementation that do not burden healthcare providers. The ability to convert standard clinical data into practical insights has the potential to significantly alter the landscape of hypertension management and preventive cardiology.