What if your heart could tell you something was wrong before symptoms ever appear? At Florida International University, researchers are harnessing the power of AI and sound to revolutionize how we detect cardiovascular disease
In this video, discover how digital stethoscopes, machine learning, and one scientist’s personal journey are combining to catch heart disease in its earliest, most treatable stages. From vocal fold inspiration to at-home monitoring, this research could reshape the future of diagnostics and save countless lives
Transcript___________________________________________
Cardiovascular disease moves in stealth, and it often takes years for symptoms to develop. And by the time they do develop, the heart is often irreparably damaged. But what if we could listen to the body’s hidden signals so we could find heart disease before it becomes too late to treat?
My lab focuses on cardiovascular disease, and a lot of my research focuses on how to better treat patients that have cardiovascular disease, covering gaps to reverse the effects of disease. But a big problem there is we often find patients way too late. And that’s really bothered me for a long time. That really got me interested in this research into how could we do this better? How could we find patients better?
The aha moment for me, at least, my wife is a classically trained opera singer, and she was studying vocal fold pathologies and how changes in the structure of vocal folds change how your voice sounds. And it really got me thinking that we may be underutilizing what we can do with heart sounds. Digital stethoscopes have been around for at least a decade now, but they have not been widely adopted, and part of the reason is that once that sound is recorded, oftentimes doctors don’t really know what to do with it.
What we’re enabling is those stethoscopes to then take the signal onto a computer equipped with an algorithm, and you can do analyses that go beyond just trying to listen for a sound but take advantage of things like machine learning, artificial intelligence, to find patterns that a human may or may not be able to see just with their eyes or ears. What we do is we feed thousands and thousands and thousands of sounds into a computer. We have that computer compare those thousands of sounds and say what little blip in those sounds, what little features in those sounds are different — something that might be inaudible to the human ear, but the computer can say this little subtle change here that is well associated with a certain disease. And then we can use that information to better diagnose patients early.
AI diagnostics are going to revolutionize the way that we find and treat heart disease. They are able to take a lot of data — personal characteristics, genetic data, sounds — and integrate them together in a way that we can better understand an individual’s unique risk for developing heart disease, where they currently exist in deficiencies in cardiovascular function, and how we might be able to treat that individual to restore them to normal cardiac function
Some major conditions that this technology can perhaps aid in early detection include heart valve disease. The heart valves are what allow the blood to come out of the heart but not go back in, and those tissues move a lot, so we can pick those up very nicely in the sound. But also heart failure. So if you get changes in the heart muscle that affect its ability to pump, and that affects a lot of people, we can also detect that in the sound just based upon the delays in the blood movement or how quickly the blood moves. Those are major conditions that this technology could benefit almost immediately.
One place we see this going is to enable at-home monitoring of cardiovascular disease, especially for those who are most at risk. A lot of people now have blood pressure monitors in their home where they track their blood pressure daily. We could do the same thing here, where a person could take daily recordings of their heart sounds, have that sent to the physician, and that way they wouldn’t have to necessarily go into the office every month or two to get a checkup. The doctor could then remotely monitor, say, the effectiveness of a medication or an intervention to understand how a patient is progressing or regressing.
And any organ that makes sound could perhaps benefit, all the way from the bladder to the kidneys to the joints. Perhaps there are ways to use sound technology to identify abnormalities that could enable cheaper, more reliable detection early on
