Zebra Medical has only been around for a few years but is looking to teach computers to read and diagnose medical images. Why is this important? Because, as its CEO Elad Benjamin explains, in the next few years there will be around a billion people globally that will be joining the middle class, but many of them, even in the wealthiest of countries, do not have access to high-end radiology services.
This can lead to a greater risk of poor or even inaccurate diagnoses, increased waiting times and the need for second opinions becoming ubiquitous, according to Benjamin. While lots of companies work on imaging, Zebra has taken the leap to diagnosis via its software. “Our goal is to create a software engine that can help provide a more comprehensive, consistent and accurate diagnoses for millions of people who are looking for answers, but not necessarily getting them.”
What makes Zebra fierce
Benjamin says, “You know, we started out with radiology, so X-rays, CTs, MRIs and so on. So, in a nutshell, what we’re doing is building up an engine which identifies particular findings or diseases in radiology images. It highlights these to doctors in order for them to give their patients an early warning of chronic disease.”
Zebra already has a few algorithms that are commercially available. Up until late November, only hospitals had access to these, and would buy the so-called engine system from Zebra and use it internally to help their docs provide better diagnosis. But just before December, Benjamin says they took things one step further by launching a consumer version of this engine.
“This consumer version allows anyone that has a copy of their CT scan to upload this to our engine. We don’t and as yet can’t give a comprehensive analysis of everything, but for the particular things we know how to analyze right now, you can get an accurate diagnosis of whether you have, say, osteoporosis or fatty liver or emphysema; and there’s more in the pipe that are coming in the future.”
There have of course been issues with consumers using products at home for medical purposes; 23andMe, as well as DNA4Life and Interleukin Genetics, have all been hit with FDA warnings about their tests.
23andMe has been turning things around, getting regulatory blessing to market its DTC genetic carrier test for Bloom syndrome back in March.
FierceMedicalDevices asked Benjamin whether health literacy was a concern now that they had a consumer product, and whether the issues that had hit 23andMe and others had impacted his strategy.
“I think so. Partly due to the perception of whether or not a software algorithm can really tell you what to do with your health, we’re trying to be very respectful and mindful of physicians and tell patients: Look, we’ve analyzed your data and we’ve found something, but we absolutely recommend before taking any action that you need to talk to your doctor," says Benjamin. "Our analysis provides a snapshot; we can’t have the full medical history of that patient, and that’s why they need to see their doctor with what they’ve found.”
So, if they have to go to their doctor anyway, what’s the point in taking the test? Benjamin explains: “We’re trying to give patients more power to be in charge of their health, and get them access to analysis more quickly that they’re probably going to find going through more traditional routes.
“Let me give you a recent example: We had a 65-year-old male with lung disease and he uploaded his scan to our server and we said, hey, we haven’t actually found any signs of emphysema in your lungs. But he was told he had severe lung disease; so he went back to his physician.
“He told them what he had found and was confused, as he was on medication for this condition. His doc was first off angry, asking him, ‘What do these guys know?' But you know what, in the end his doctor was forced to admit that actually, he had lung disease but it wasn’t emphysema, so the drugs and care he was getting were wrong. So yeah, it does given some kind of control to the patient, and in this case, it had the power to change his care for the better.
“I’m not saying this will always change what a patient is doing, but it allows another perspective, akin to a second opinion in some ways, on a patient’s care.”
Given the problems coming out of Theranos and other setbacks for companies like 23andMe, Benjamin says he does feel an extra obligation to be whiter than white, and ensure that his products are as accurate as they can be.
“In the U.S. especially, and in the public eye in general, you have to be really very careful because healthcare is for obvious reasons very sensitive anyway. You’re dealing with news that sometimes people don’t know how to take, and you have to be very careful in how you communicate; and on top of that, you have to be very transparent about just how accurate you are.”
Theranos, Benjamin says, is in a different league of wrong, but he points to what happened at 23andMe as being “very interesting.” He says that they based a lot of their insights on literature that’s out there, but the FDA still came back and said it essentially hadn’t shown the amount of clinical rigor that it would expect, so told it to stop marketing its product.
“Knowing all this, we’re trying to be very, very careful in making sure that we document everything, validate everything, all to the highest standards so that if and when the FDA comes knocking on our door, we have all the information we need.”
Its consumer product has not yet, however, been launched in the U.S., because as Benjamin explains, Zebra felt that with the current “environment” in the FDA, any conversation simply “wouldn’t go anywhere.” At the moment, it’s only available outside the U.S.
Other big companies such as Philips, Siemens and IBM Watson are also working on similar projects and products, with IBM Watson looking, in essence, to build a “virtual doctor” fed with a patient’s medical data, and aiming to one day replace the family doctor, the radiologist and so on. This is some years off, despite the adverts, but how can Zebra compete with this kind of work?
“There’s a big difference between what we do and what others are doing. All the traditional players, like the ones you mention, haven’t really taken a step towards diagnosis, not in the way we have. They all certainly have significant products in the imaging space, but these are all around managing images, helping them get from place to place, helping docs to visualize data, but none of them have taken the leap to actually saying: This is a diagnosis, or this is what’s wrong.
“So even though we’re a young startup, we’re still relatively far ahead of the big companies in that regard, because they haven’t gone there yet.”
Benjamin says IBM has thrown a lot of weight behind Watson but adds that they are taking a different approach with the virtual doc strategy. “It’s a great vision,” Benjamin says, “but it’s different to what we are, and it’s going to be much, much more difficult to achieve.”
What to look for
While there are other small companies that have gone into the imaging space and are aiming to provide some level of diagnosis, Benjamin believes Zebra was the first, the furthest ahead, and the only one with a commercial offering of that kind of tech.
The company has a multiyear partnership with Dell Services, part of which will see them bring algorithm-based image analysis to healthcare providers. Last year, Zebra raised $12 million, with this round led by Salt Lake City’s not-for-profit Intermountain Healthcare. — Ben Adams, @FierceBiotech