Israeli app sounds alarm before heart failure — just by analyzing your voice - Maccabee Task Force

In congestive heart failure study with recovered patients who used HearO, its technology predicts some 82% of relapses, catching them on average 18 days before they occur.

A new Israeli app can sound the alarm before the onset of congestive heart failure — just by analyzing the sound of a user’s voice.

A study with recovered patients found that the app predicted some 82 percent of relapses before they happened.

HearO “listens” to voice samples that users record on their smartphones and alerts them if they are at imminent risk of congestive heart failure. It works by detecting irregularities in an individual’s speech, comparing them to their healthy “baseline” voice. If anomalies are detected, doctors are immediately informed so they can take preventative action.

The new study was conducted by the company behind the app, Cordio Medical, in partnership with Beilinson Hospital, Barzilai Medical Center and Galilee Medical Center, as well as Clalit Health Services Cardiovascular Centers. It is currently undergoing peer review.

Patients who had experienced heart failure and were therefore deemed to have significant chances of a relapse used the HearO app at home and sent voice samples in Hebrew, Arabic or Russian. The 180 patients recorded several clips a day over two years, meaning there were 460,000 clips reviewed for the study.

Heart failure happens when the heart muscle’s ability to contract has been harmed over time or when it has a mechanical problem that limits its ability to fill with blood. The heart then can’t meet the body’s need for blood, and blood returns to the heart at a faster rate than it is pumped out. The heart becomes congested; hence the term congestive heart failure.

Cordio Medical CEO Tamir Tal said that around one-third of the patients experienced heart failure during the two-year study, as anticipated. When their medical records were cross referenced to the app’s analysis of their voice, it emerged that the app had predicted the failure in 82% of cases. It detected the warning signs roughly 18 days before an incident on average.

Continue Reading: