Voice changes may detect worsening heart failure weeks before other symptoms appear
A smartphone-based artificial intelligence (AI) voice analysis system can detect worsening heart failure symptoms more than three weeks early, significantly outperforming the standard of care. That's according to late-breaking DETECT-HF trial data presented at the American Heart Association Scientific Sessions 2026 and simultaneously published in JACC: Heart Failure.[1]
DETECT-HF evaluated Cordio Medical's HearO system in 563 ambulatory heart failure patients. It found that the technology predicted more than 70% of significant heart failure events an average of 24.6 days in advance. The standard of care, conventional daily weight measurements, have a sensitivity of approximately 20% for predicting heart failure hospitalizations, according to study lead investigator William T. Abraham, MD, a cardiologist and professor of medicine at The Ohio State University Wexner Medical Center.
"Clinicians have known for years that you can hear subtle changes in a patient's voice when they present to the hospital with decompensation, but now that's been translated into a phone-based app that can listen to the patient every day and tell us three or four weeks in advance from home, without ever seeing the patient, when they're on a course of worsening heart failure," Abraham told Cardiovascular Business.
The new technology addresses the longstanding challenge in heart failure management in detecting gradual fluid accumulation in the lungs before patients develop symptoms severe enough to require hospitalization. Heart failure is one of the top expense drivers for Medicare, and hospitals are always looking for ways to prevent readmissions.
HearO uses a smartphone application to collect six standardized spoken sentences from patients each day. Rather than analyzing the words themselves, the system evaluates characteristics of speech, including intonation, breathing patterns and other acoustic features that may change as fluid accumulates in the lungs.
The recordings are uploaded to a cloud-based platform, where AI algorithms develop a patient-specific voice model and identify changes between periods of clinical stability and instability. When the system detects patterns associated with worsening heart failure, it sends an alert to clinicians through a secure web-based portal.
The DETECT-HF trial evaluated heart failure hospitalizations and outpatient heart failure events requiring intravenous or mechanical therapies. Conducted across 46 sites, the blinded study included patients speaking English, Spanish, Hebrew, Russian and Arabic, demonstrating the technology's language-independent approach.
Beyond its predictive performance, the system had strong patient engagement. Participants completed 85% of their scheduled daily voice recordings, while usability ratings ranged from 4.7 to 4.9 out of five. The study also reported an unexplained alert rate of 2.65 per patient per year, meeting its predefined target of fewer than three. No device-related adverse events were reported.
Abraham said the high adherence likely reflects the simplicity of the approach.
"This tends to be among the highest rates of adherence that we've seen with a monitoring device for heart failure, because it's simple," he said. "The beauty is that it is a smartphone-based application. There's nothing to implant, there's nothing to wear, there's nothing to remember to take with you on a day-to-day basis. I think it's a no-brainer. It has high sensitivity, low false positive rate, no adverse events, and high rates of adherence."
The ability to identify deterioration weeks before a hospitalization could give clinicians a substantially longer window to adjust medications or otherwise intervene to prevent patients from progressing to acute decompensation. It also could offer a less invasive and more accessible approach to remote monitoring for a broader heart failure population.
Cordio Medical said DETECT-HF represents the first blinded, prospective validation of AI-assisted voice analysis for heart failure monitoring.
Earlier this year, the HearO technology received the FDA's breakthrough device designation. It has also received CE mark approval in Europe and is approved for use in Israel. The company now plans to use these DETECT-HF data to support an FDA submission. If successful, the company said it hopes to launch the technology in the U.S. in 2027.
Dave Fornell has covered healthcare for more than 17 years, with a focus in cardiology and radiology. Fornell is a 5-time winner of a Jesse H. Neal Award, the most prestigious editorial honors in the field of specialized journalism. The wins included best technical content, best use of social media and best COVID-19 coverage. Fornell was also a three-time Neal finalist for best range of work by a single author. He produces more than 100 editorial videos each year, most of them interviews with key opinion leaders in medicine. He also writes technical articles, covers key trends, conducts video hospital site visits, and is very involved with social media. E-mail: [email protected]