A Deep Learning Algorithm To Eliminate Errors During Fractional Flow Reserve

Fractional Flow Reserve (FFR) is a leading innovative technique for diagnosing narrowings in the coronary arteries, a problem that leads to an impeded oxygen supply to the heart muscle causing severe chest pain, which can lead to heart attack and death. Last year there were 500, 000 FFR procedures. Of these, studies suggest that due to measurement errors, up to 30% were performed incorrectly, leading to unnecessary treatments and increased mortality and morbidity rates. These errors occur due to the subtlety of the readings and even experienced FFR experts can make mistakes or miss key information. We are developing a deep-learning algorithm that will alert doctors in real-time if measurement errors are occurring. It is grounded in our industry expert's years of expertise and research in coronary physiology, and our team’s AI expertise. If successful, this will change the clinical outcomes of up to 150, 000 patients per year. Building such an algorithm is highly innovative and challenging. The learnings and techniques we develop from this algorithm will lay the groundwork for us to attack other similar problems within cardiology and the wider realm of medicine itself, and potentially save millions of lives over the next 20 years.