According to a preliminary survey published at the American Heart Association's Hypertension Science Session 2025, AI voice agents helped elderly people with hypertension accurately report blood pressure readings and improve blood pressure management. Genetics.
Although blood pressure control remains the basis for improving patients' cardiovascular outcomes, capturing timely and compliant blood pressure measurements remains a challenge, especially for patients with limited access to care. Our study was able to improve the accuracy of blood pressure measurements and patient outcomes. ”
Tina Anne Kerr Thompson, Maryland, Chief Research Author, Senior Vice President of Primary Care Services Line, Executive Director of Population Health Cooperation at Emory Healthcare, Atlanta
AI Voice Agent is a conversation system equipped with a large language model that allows you to understand and generate natural speeches in real time when interacting with humans. The study included a majority of 2,000 adults aged 65 and over, and was designed to self-report accurate blood pressure measurements, instead of calling with a healthcare professional about blood pressure measurements, the effectiveness and scalability of voice-enabled AI agents that make patients rely on patients. The AI agent also identified patients who needed follow-up medical care based on blood pressure measurements.
AI voice agent calls to patients were created using commercially available AI in multiple languages, including English and Spanish. Reading blood pressure outside the threshold range of different measurements based on the presence of other conditions such as diabetes was called to a licensed nurse or medical assistant. The presence of symptoms such as dizziness, blurred vision and chest pain also encouraged the escalation of the call. Escalations to additional care were immediate in emergencies or within 24 hours on non-urgent issues.
Patients were contacted by voice agents and either provided recent blood pressure readings or performed live measurements during the call. After the call, measurements were entered into the patient's electronic health records and reviewed by the clinician. Routing and referral for care management was spurred in patients with hypertension who were difficult to control. This process reduced manual workload by clinicians and resulted in an 88.7% lower cost per read. This amount was calculated by comparing the cost of commercially available AI voice agents with the use of human nurses to perform similar tasks to ensure that patients' self-reported blood pressure measurements were successfully obtained.
This study found that integration of AI into clinical workflows improved patient care management. During the survey period:
- 85% of patients were successfully reached by voice-based AI agents.
- Of these patients, 67% completed the call and 60% received a compliant blood pressure measurement during the call. Of these patients, 68% met CBP (control of blood pressure).
- Overall, the 1,939 cbp gap was closed, improving performance (17% improvement) with measurements from 1-4 star. Medicare Advantage (MA) and Healthcare Efficacy Data and Information Set (HEDIS) CBP Measurements increased from previously reported 1-star ratings to 4-star ratings.
- At the end of each completed call, patients underwent a two-question survey to assess satisfaction on a scale of 1-10, with 10 being 100% satisfied. Among completed calls, the average patient-reported satisfaction rate was over 9 out of 10, reflecting excellent overall experience with voice-based AI agents.
“After interacting with an artificial intelligence-based voice agent, I was surprised at the high patient satisfaction,” Thompson said. “We are excited that patient involvement and satisfaction are so important to medical outcomes that it means what it means in the future.”
“This could be a game-changing study,” said Eugene Yang of FACC, MS, MS, MD, MS and MS, volunteer experts at the American Heart Association. “A precise blood pressure measurement is essential to improve control, and new approaches can enable that. Such breakthrough AI techniques can change the way blood pressure is managed by addressing important barriers such as patient reaching patients wherever they are and restricting access to care and GAP in patient support.” Yang, who was not involved in this study, is a professor in the department of delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene Benke's delegated Karl and Rene B
This study has several limitations. This study was observational and there was no control group. Sequential AI calls were not compared to human calls. Instead, an AI voice call was deployed. This is because only the right number of people could not be called. Furthermore, this study was retrospective and implied that it reviewed existing data, and the assessment was completed after clinically identified calls had already been made.
Details, background, design:
- Participants included 2,000 adults. The majority were over 65 years old (average age 72, 61% of women) receiving hypertension care.
- Electronic health records reviews identified patients who had missing blood pressure data or who had not within the normal range of the latest BP reading below 120/80 mm Hg. Patients with these gaps in the data were tagged and received calls from AI voice agents.
- This study was conducted with patients from Emory Healthcare in Atlanta over a period of 10 weeks. The patient received at least one call during the study. Patients received multiple calls if they did not answer the phone.
- Patients with open gaps in blood pressure management were identified through electronic medical records (EMR) and payer analysis. Patient lists were reviewed to ensure that the information in the records was correct and validated for outreach by the clinical surgical team to ensure real-time accuracy of gaps prior to outreaching to patients.
- Using AI texts, calls from traditional care teams, recent clinical visits where blood pressure measurement documents can be found and generator AI voice agents, patients were contacted to provide recent blood pressure measurements and to receive blood pressure measurements during the phone call. These included recent clinical visits in which documented BP documents were found.
- Post-call validation steps were integrated into the workflow, measurements were entered into the EHR, reviewed by the clinician, and submitted as supplemental data to close the star quality gap. For patients with uncontrolled hypertension, a clinical escalation referral was given to the care management team.
- The Centers for Medicare and Medicaid Services (CMS) has developed a star rating system known as MA Stars, planning Medicare Advantage (MA) (Part C) and Prescription Drug (Part D) plans on a 5-star scale, with 1 being the lowest score and 5 being the highest score. The MA plan is a plan for a private insurance company approved by Medicare and was not issued by Medicare itself. Hospitals, care centers and clinicians are eligible for an increased bonus payment if they achieve at least a four-star rating.
Self-measured blood pressure is the focus area of BP:BP, an initiative by the American Heart Association that helps healthcare organizations improve blood pressure control speeds through evidence-based programs. As stated in the association's new 2025 guidelines on hypertension released last month, home blood pressure monitoring is recommended for all adults with all levels of hypertension.
sauce:
American Heart Association
