• January 2019

    The fourth industrial revolution: How AI is shaping the future of cardiology

Today’s healthcare AI landscape

The impact of the first three industrial revolutions — the steam engine, the age of science and mass production, and the rise of digital technology — fundamentally changed the world and transformed society. The fourth industrial revolution is upon us powered by the cloud, social, mobile, the Internet of things (IoT) and artificial intelligence (AI), along with increasing computing power. We’ve gone from neural networking to machine learning and now deep learning, or black-box learning.

Today’s powerful cloud-based systems can transmit and integrate data to sort it all out and make decisions rapidly. AI encompasses many aspects of machine learning, including but not limited to augmented reality, big data, body sensors, robots and genomics. McKinsey Global Institute reports that AI could potentially deliver economic output of around $13 trillion by 2030, boosting global GDP by about 1.2 percent a year.

Healthcare is now beginning to realize the benefits of AI in big ways. Virtual nursing assistants offer patients immediate communication that could prevent hospital readmissions and unnecessary visits. Advancements to robot-assisted surgery have been made possible through AI by using data from previous operations to inform current and future surgical techniques.

This year, new software received clearance from the U.S. Food and Drug Administration (FDA) for fully automated AI echocardiogram analysis. It combines deep learning with cardiovascular imaging to help diagnose and manage heart disease.

The opportunity for AI in diagnosing cardiovascular disease

John Stevens, MD, the CEO of HeartFlow delivered this year’s keynote at the Dallas Leipzig conference and shared how AI has opened the door to a promising advancement in cardiology.

Cardiovascular disease (CVD) is the leading cause of death globally with approximately 400 million known cases. As physicians know all too well, despite its prevalence, there is no way to evaluate who does or does not have CVD with 100 percent accuracy. By many estimates, that nets at roughly $30 billion worth of non-invasive testing alone to figure out if there is need for intervention.

While we have made progress in reducing the proportion of people in the world who develop and die from CVD, the changing demographics of our aging population are putting us on an exponential curve in terms of potential cases. The need for more accurate diagnoses has never been greater.

HeartFlow: Harnessing the power of deep learning

Accurate, precise and meaningful AI results begin with ground truthing, i.e., our ability to gather disparate pieces of data in a holistic way — histology, pathology, MRI, radiology, echo, vectoring, trajectories, outcomes, OCT, IRIS, QCA, and post-op imaging — allowing the AI system to do its job. Current diagnostic tests lack the ability to quickly and accurately generate these types of data models. In his keynote, Dr. Stevens used his company’s product, HeartFlow, as an example to demonstrate AI’s ability to fill this void.

HeartFlow takes data from a standard cardiac CT scan then analyzes it with a unique set of algorithms. These algorithms allow for the development and transmission of a quantitatively precise anatomic model of the coronary arteries while also understanding the hemodynamics, the vessel wall, and the forces on plaques — all from deep learning. This is a prime example of what makes artificial intelligence so unique. AI can enable high performance computing of enormous amounts of data while analyzing and correcting the inputs in a virtual cycle.

Dr. Stevens reported that in the last six months, HeartFlow has returned diagnostic information in less than five hours for interactive viewing to a physician’s mobile device for about 15,000 patients. These return times have been falling by about 30-50 percent per year for the last five years. The company expects that trend to continue for the next five years with a goal of getting to about 5-10 minutes. All of this is possible because of improvements in AI.

As entities like the FDA embrace new technology, pathways to incorporate AI systems into healthcare could take shape. The future for AI is bright and its evolving role in healthcare is without a doubt one to watch.

This content was medically reviewed by:

David Brown, MD
President of medical staff affairs
Baylor Scott & White The Heart Hospital – Plano

Paul Grayburn, MD
Medical director of non-invasive cardiology
Baylor Scott & White Heart and Vascular Hospital  – Dallas

Robert Smith, MD
Vice chairman of cardiovascular surgical services
Baylor Scott & White The Heart Hospital – Plano