How This App Can Predict Heart Failure Just by Listening to Your Voice
Earlier this year in March, the COVID-19 disease was declared a global pandemic by the World Health Organisation. Although cases are falling in many places around the world, this will most likely not be the last pandemic the world faces. In fact, current hospitalisation statistics indicate that, for many industrialised countries, another might be just around the corner. In fact, in the UK, it’s already the nation’s leading cause of death.
Heart failure affects over 26 million people around the world, and is particularly focused in developed countries with high life expectancies and ageing populations. It occurs when the heart becomes weak or stiff, and is therefore unable to adequately pump blood around the body. This is often fatal, although risk of death can be offset with treatment and lifestyle changes. Nevertheless, heart failure is a serious condition which places a significant burden on both the patient and their healthcare provider, especially when the patient is hospitalised.
Normally, healthcare professionals would regularly monitor heart failure patients in an attempt to reduce hospitalisation rates. However, in times when this is difficult to do consistently, remote monitoring can be a more practical solution. These difficulties motivated Israeli AI startup Cordio Medical to develop Cordio HearO™, an app which uses artificial intelligence to detect symptoms of heart failure in a patient’s voice.
The link between cardiac and vocal function generally isn’t immediately obvious, but can be explained through the effect of heart failure on the lungs. When the heart is unable to pump effectively, blood can become congested in the veins supposed to carry it to the lungs. Increasing pressure in these veins forces fluid into the alveoli - small air sacs throughout the lung tissue which act as sites of gas exchange. This condition, known as pulmonary edema, reduces normal oxygen movement through the lungs, and causes shortness of breath.
This shortness of breath can be subtle and difficult to detect in the early stages, but the AI-powered voice analysis provided by the Cordio HearO™ app has been shown in a Cordio Medical-funded study to be successful at distinguishing between congested patients pre-treatment and uncongested patients post-discharge. This research was submitted to the European Society of Cardiology last week.
The app works by taking five 2-5 second recordings in an uncongested state to form a personalised healthy model. Then, patients are asked to submit 30 second recordings each subsequent day in any language, which the app compares to previous data. Any small deviations indicating fluid accumulation generate an alert, which is picked up by physicians using a designated web portal. Through machine learning, the analysis algorithm is able to adapt and become more accurate over time as more data is collected.
Speech is personal and as such, very small changes (related to the same person) can be detected – for example, the ability of parents to notice health issues by listening to their kids. Today we report results of the first easy to use, non-invasive, personalised heart failure monitoring device. It requires a simple 30-second recording each day, in any language.
This technology is another example of the use of telemedicine platforms becoming increasingly commonplace during the COVID-19 pandemic.
(Thumbnail Source: Business Insider)