New Zealand-based medtech startup HeartLab has raised $2.45 million in seed funding that it says will help the company expand its AI-powered heart scanning and reporting platform to cardiologists in the United States by early next year.
HeartLab provides an end-to-end solution for echocardiograms, the ultrasound tests that doctors use to examine a patient's heart structure and function. Not only does the software help sort and analyze ultrasound images to help doctors diagnose cardiovascular disease, but it also streamlines the workflow by generating patient reports for doctors that can then be added to a patient's health record.
Will Hewitt, 21, started HeartLab when he was 18 years old studying applied mathematics and statistics at the University of Auckland and working as a researcher at the Auckland Bioengineering Institute. The idea for the startup came to him as he listened to cardiologist, and now co-founder, Patrick Gladding explain how time-consuming and potentially inaccurate it is for doctors to have to review multiple scans manually everyday.
"You've got a really repetitive manual task done by a highly trained professional," Hewitt told TechCrunch. "To start with, we just decided to train the AI to do one really small part of the doctor's job, which was to look at these scans and generate a couple of different measurements that normally the doctor would have to do themselves," said Hewitt.
In order to replicate the tedious process that doctors were doing, HeartLab built its own in-house labeling tool with sonographers that includes step-by-step guides and prompts to collect data on a range of different measurements. Hewitt said this initiative was one of the most valuable efforts of engineering the company has invested in to date because it has lead to cross validation, which is used to test the ability of the machine learning model to predict new data, as well as flag problems like selection bias and overfitting.
Once HeartLab was able to successfully replicate the scanning process, the company worked to expand its services in a way that would relieve doctors of further admin minutiae so they could spend more time actually treating their patients. Usually, doctors use a software tool that analyzes the images, another that visualizes patterns and another that actually writes up the report, says Hewitt. HeartLab's platform, called Pulse, can now condense those processes into one software.
Cardiologists and sonographers at four different sites in New Zealand are trialing HeartLab's tech now, which is also awaiting regulatory approval from the U.S.'s Food and Drug Administration. HeartLab anticipates FDA approval of Pulse by the first quarter of 2022, which is when the startup can begin selling the SaaS product.
"To begin with we want to talk to small and medium clinics over in the U.S.," said Hewitt. "We've actually found that our products are most popular at those clinics because it replaces more software than at a larger clinic. At a larger clinic some of these bits of software they've already had to purchase, versus a smaller clinic, it's stuff that they couldn't access anyway. So when we get to the states, we want to start shipping mostly to those sorts of users while we work out how to best pitch our value proposition to the larger clinics."
Hewitt says the funds from this round will also help the startup hire 10 more staff members to join the existing 13-member team based in Auckland. Having more tech talent on board will help HeartLab advance its product offering. At the moment, Pulse is at the point where it sees so many scans and takes so many measurements that it can get through the process quicker than a doctor could on their own and actually pick out patterns that a doctor wouldn't see, according to Hewitt. The next step, which a good chunk of the seed funding is going toward, is how to be diagnostic about disease rather than just being able to indicate it.
"How do we actually provide something that normally doctors would have to order another scan for?" said Hewitt. "One of the key ideas with AI is you can create mappings from low-resolution images like ultrasounds. How can we try to learn a pattern from an ultrasound that's similar to what you might see from an MRI, for example?"
If HeartLab can figure out how to glean advanced information from an echocardiogram instead of an MRI, it would be able to save hospitals, clinics and patients a lot of money. Each cardiac MRI can cost about $1,000 to $5,000, which is about five times the price of an echocardiogram.
"I'd say the biggest challenge for us is, how can we transform from a company that at the moment can deliver products to a few local clinics successfully to actually building a product that scales and delivers a really good experience to lots of users and different hospitals?" said Hewitt.
Advancements in early diagnostics and imaging tech like HeartLabs' is causing an increased demand for such tools. As a result, the global AI-enabled medical imaging solutions market is expected to reach $4.7 billion by 2027. By extending its reach to the U.S., where heart disease is the leading cause of death, HeartLab is poised to take a big piece of that pie.
In total, HeartLab has publicly raised about $3.2 million in funding, which includes a pre-seed last October of about $800,000 led by Icehouse Ventures with support from Founders Fund, the San Francisco-based VC firm that led the round announced on Thursday. Icehouse Ventures also contributed to the oversubscribed seed round, along with another New Zealand firm Outset Ventures and private investor and CEO of design platform Figma, Dylan Field.
“The use of AI in medicine is reducing pressures on health systems and ultimately saving lives," said Founders Fund partner Scott Nolan, who has led investment rounds for three other New Zealand startups, in a statement. "The HeartLab team has built a really compelling AI-powered platform that doctors love to use.”