Q&A: How AI and deep learning is used for cardiovascular treatment

Ian Bolland caught up with Charles Taylor, founder and chief technology officer of HeartFlow, talks about the recently FDA cleared HeartFlow Planner, and the relationship between medtech and the NHS.  

Tell us a bit about HeartFlow? How has the product been developed? What effect does the technology have when it comes to diagnosis?

I started HeartFlow in 2010, with Christopher Zarins, a cardiovascular surgeon who I met at Stanford University when pursuing my doctorate. I’d been studying computational fluid dynamics for years – using algorithms and data analysis to understand how fluid moves in a specific space. I’d been considering how this could be applied to medicine for some time; specifically, combining computer simulation methods with medical imaging data. The HeartFlow Analysis came from this, with the aim of helping clinicians to understand the severity of coronary heart disease (CHD) in a patient’s arteries.

The technology works by taking the data from a patient’s coronary CT scan and using deep learning, plus input from a team of very skilled analysts, to create a digital, 3D model of the heart and coronary arteries. Complex algorithms then solve millions of equations to accurately understand blood flow in the patient’s coronary arteries and assess how any blockages may affect this flow. Clinicians can look at the information provided by the HeartFlow Analysis, which then helps them to determine the most appropriate course of treatment.

Is your technology used by clinicians? What effect can it have on treatment plans, if any?

Clinicians across the UK, US and Japan currently use the HeartFlow Analysis in their practices. It’s a tool that helps cardiologists make a diagnosis and formulate the best treatment plan for their patients. In many cases, a patient can get a definitive answer as to whether they have CHD in just one visit to the hospital.

It can also help patients avoid invasive tests that they don’t ultimately need. Previously, when a clinician wasn’t sure about the severity of CHD in a patient’s arteries, they would need to send them on for a diagnostic invasive angiogram. This procedure would come with a certain level of risk for the patient, as well as making catheterisation labs busier and contributing to increased waiting times. The HeartFlow Analysis gives clinicians a level of information only otherwise available through invasive procedures. As a result, doctors can formulate their diagnosis for patients without the patient having to schedule and undergo an invasive diagnostic procedure at a catheterisation lab.

Where have you been operating before the UK? And tell us about your work in the NHS?

We started out with a presence in US hospitals before beginning our work with the NHS England. We were initially in a handful of hospitals in the UK, but mostly from a clinical research perspective. Our presence in the UK ramped up commercially when we successfully received funding from the NHS’ Innovation and Technology Payment (ITP) programme in 2018.

The funding meant that we could begin working with individual hospitals and NHS Trusts across England without impacting their annual budgets. We were awarded ITP funding again in 2019 which has allowed us to work with nearly 50 hospitals across England, with plans to be in over 70 by 2020.

The HeartFlow Analysis is helping doctors formulate treatment plans and reduce the number of patients sent for invasive diagnostic testing. It’s also helping to save the NHS money. The National Institute for Health and Care Excellence (NICE) estimates that it could save £9.1 million a year through a reduction in costly invasive procedures and time saved in the catheterisation lab.

What are your thoughts on the relationship between medtech and the NHS? Would you like to see any changes?

The pace of technological development has accelerated rapidly in recent years but there’s still so much opportunity to go further. AI is helping to make the diagnostic process more accurate and efficient, so relationships between those who develop these tools and the NHS is vitally important in improving patient care and delivering much-needed cost savings.

Continued support from the government in its commitment to introduce cutting edge technology to the NHS is key. This has already been the focus of a number of announcements, such as the speech from Baroness Blackwood earlier this year, in which she outlined the government’s ambition for the UK to become the world’s leading health tech ecosystem and pointed to HeartFlow as an example of new technology already benefitting the NHS. These initiatives have been important in helping to raise the profile of innovative products and accelerating adoption across the country.

Where did the idea for the HeartFlow Planner come from? Is this a natural progression for HeartFlow, or is it entering a space that you wouldn’t have previously envisaged?

I’ve always thought that HeartFlow’s technology could be used to help in the planning of procedures, as well as providing immediate medical insights, it was just a case of getting there. The HeartFlow Planner allows clinicians to plan procedures before a patient gets to the operating theatre.

The technology helps physicians to identify blockages, virtually model different treatment scenarios before the procedure, and enable them to understand the potential impact of the modelled treatment in real time. We believe this tool will help to streamline treatment planning for patients in addition to giving doctors the ability to review their plans with colleagues and ensure everyone has a clear picture of the initial treatment plan before entering into the catheterisation lab. The HeartFlow technology is an AI-based system, so the more data we have to train the system, the more improvements we will be able to make.

Planner has just received FDA clearance in the United States and is currently under review for CE-marking.