Remote Voice Analysis for Monitoring Ambulatory Heart Failure Patients: Results of the DETECT-HF Study
Year: 2026
Type: article
Abstract: Background The Cordio HearO System is a software medical device that analyzes daily voice recordings to identify changes associated with worsening HF. Objective The objective of the DETECT-HF study was to evaluate the ability of the HearO System to detect impending HF events (HFEs) using a voice-based monitoring algorithm. Methods DETECT-HF was a prospective, international, multicenter, single-arm, blinded, noninterventional study in HF patients who performed daily voice recordings on their smartphones. Co-primary endpoints were sensitivity for detecting adjudicated first HFEs (HF hospitalization or outpatient HFE requiring intravenous therapy) and the unexplained priority notification rate (UPNR). Performance was assessed against predefined criteria of a lower 95% confidence interval (CI) ≥60% for sensitivity and <3 notifications/person-year for UPNR. Analyses were also performed using all HFEs (first and recurrent events). Secondary endpoints included safety and usability assessments, including patient adherence to daily voice recordings. Results Between March 2021 and July 2025, 563 patients were enrolled. Mean age was 67.6±11.5 years, and 70.9% were male. The HearO System detected 52 of 76 first HFEs, yielding a sensitivity of 68.4% (95% CI: 58.0%-78.9%; p=0.0573), which was below the pre-specified performance target. For all HFEs (n=112), sensitivity was 70.5% (95% CI: 60.8%-78.7%). On average, HFEs were detected 24.1 days before clinical presentation. UPNR was 2.65 notifications/patient-year (95% CI: 2.53-2.76; P<0.001) for first HFEs, meeting the pre-specified performance criterion, and 2.59 notifications/patient-year (95% CI: 2.48-2.70) for all HFEs. Sensitivity and UPNR estimates were consistent across prespecified subgroups. A questionnaire demonstrated excellent usability (mean scores ranging from 4.7 to 4.9 on a 5-point scale across usability questions). Mean adherence to daily recordings was 85%, and no device-related adverse events occurred. Conclusions Voice-based monitoring with the HearO System detected impending HFEs weeks before clinical presentation with reasonable sensitivity, low alert burden, high adherence, and an excellent safety profile.
Source: JACC Heart Failure
Authors: William T. Abraham, Offer Amir, Antoni Bayes-Genis, Michal Laufer-Perl, Andrew P. Ambrosy, Liviu Klein, Anuradha Lala, JoAnn Lindenfeld, Steven V. Manoukian, Daniel Bensimhon, Pedro Martinez-Clark, Michael Koren, Elazer R. Edelman, Chaim Lotan, Paul D. Feigin, Sam Galloway, George DeMuth, Sandra Chaparro, Debbie Rinde-Hoffman, W.H.Wilson Tang, Vinay Thohan, Stefan D. Anker, Daniel Burkhoff, Sean Pinney
Institutions: The Ohio State University, Hadassah Medical Center, Hospital Universitari Germans Trias i Pujol, Tel Aviv University, Tel Aviv Sourasky Medical Center, Kaiser Permanente San Francisco Medical Center, Kaiser Permanente Northern California, University of California, San Francisco, Vanderbilt University Medical Center, HCA Healthcare, Moses H Cone Memorial Hospital, Jacksonville Center for Clinical Research, Massachusetts Institute of Technology, Technion – Israel Institute of Technology, Baptist Health South Florida, Tampa General Hospital, Cleveland Clinic, Mission Health, German Centre for Cardiovascular Research, Charité – Universitätsmedizin Berlin, Cardiovascular Research Foundation
Language: English
Subfield: Cardiology and Cardiovascular Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
Estimated APC paid: $5,000
Funder: BIRD Foundation
OpenAlex ID: W7225644355