Posted on by Dr. Francis Collins
So much has been reported over the past six months about testing for coronavirus disease 2019 (COVID-19) that keeping up with the issue can be a real challenge. To discuss the latest progress on new technologies for SARS-CoV-2 diagnostic testing in the United States, I spoke recently with NIH’s Dr. Bruce Tromberg, director of the National Institute of Biomedical Imaging and Bioengineering (NIBIB). Not only does Bruce run a busy NIH institute, he is helping to coordinate the national response for expanded testing during the COVID-19 pandemic.
Bruce also has a leading role in one of NIH’s most-exciting new initiatives. It’s called the Rapid Acceleration of Diagnostics (RADx) initiative, and it is on the fast track to bolster the country’s diagnostic testing capacity within months. Here’s a condensed transcript of our chat, which took place via videoconference, with Bruce linking in from Bethesda, MD and me from my home in Chevy Chase, MD:
Collins: Let’s start with how many COVID-19 tests are being done right now per day in the United States. By that, I’m referring to testing for the presence of the novel coronavirus, not antibodies as a sign of a previous infection.
Tromberg: The numbers fluctuate—anywhere from around 400,000 to 900,000 tests per day. So, the national capacity, with all these complex laboratory tests and emerging point-of-care assays, is getting close to 1 million a day. That’s substantially higher than in mid-April, when the nation was doing about 150,000 tests per day. But most testing is still being done in laboratories or complex facilities, and it can take a while for those tests to be run and for people to get answers. What we’d like to have are more convenient tests. We’d like to have tests that people can have at the point of care, where you get an answer on the spot and very quickly, or tests that can be performed easily in their homes.
Collins: Yes, we’d all love to have point-of-care tests for COVID-19. And there are some out there already. Every time I go to the White House, they have this gadget, called Abbott ID Now, that gives a result in about 15 minutes. That sounds pretty good. Do we just need to make more of those machines to solve the problem?
Tromberg: Abbott ID Now is one of the first point-of-care technologies. It’s not complicated, so a specialized laboratory isn’t required to run them. That’s what makes Abbott ID Now very appealing, but its performance could be better. There’s a bit of a risk when it’s used in individuals for which you really need to know, with absolute certainty, if they have the virus or not. Those performance issues have created opportunities to build platforms that are better, faster, and possible for people to do on their own.
Collins: Congress provided a big infusion of resources last April to assist in the development of new diagnostic technologies for COVID-19. A lot of that infusion came to NIH, and, Bruce, you were asked to step in and make something amazing happen on a timetable that’s pretty breathtaking. It’s called the RADx Initiative. Tell us a little about that.
Tromberg: RADx is short for Rapid Acceleration of Diagnostics. The goal of the initiative is to make it possible for everyone to have access to diagnostic testing for COVID-19 as easily and quickly as possible. As we pivot to doing surveillance in large populations, we will need greater testing capacity to help optimize the management of each individual. So, that’s really the aim of RADx, or RADx-tech, which is a special flavor of RADx.
Collins: Right, the goal of RADx-tech, which you are overseeing, is to identify some of these exciting new technologies and help scale them up quickly to the point where they can help people across the nation. Could you give us some examples?
Tromberg: Sure. One general class of technologies is called a lateral flow assay. These tests are small enough to fit in your hand and come in a convenient container. Basically, you can use a swab from your oral cavity and place it on one of the pads, and then you add a little bit of solution. The actual assay itself has a membrane inside of a little plastic container. The fluid flows across the membrane, and there’s chemistry that goes on inside the container to detect, for example, genetic material from the coronavirus. So, it can tell you if there is a presence of virus inside the swab. It’s very quick and straightforward. A line will “light up” if virus is present.
Another type of lateral flow assay, also small enough to hold in your hand, looks for proteins on the surface of the virus. You don’t have to break up the virus particle itself, but in this specific example, what captures the virus in this membrane is what’s called an aptamer. An aptamer is similar to an antibody, except it’s made from nucleic acid. It’s designed to bind very tightly with any molecule of interest. If you put a saliva sample into this assay, it moves up the membrane and some chemistry takes place. And then, you’ll see a line appear if there’s presence of a virus.
Collins: You just said saliva. I think a lot of people would much prefer, if they had to provide a sample, to use saliva instead of having a swab stuck in their nose, especially if it has to go all the way to the back of the nose. Does saliva work?
Tromberg: We hope so. Right now, RADx-tech has at least nine companies that are in what we call phase one, which is a significant step towards commercialization. Of those companies, more than half are looking at saliva or other kinds of sampling that’s not sticking swabs way up into the nasal cavity.
Another type of test is a lateral flow assay that fits directly into a mobile device like a tablet. It has a separate lateral flow apparatus, which looks like an elongated zip drive, and it slides right into the tablet itself. It’s something that’s not complicated. It would be easy to do at home. But rather than watching for the presence of a reaction, you look for a light inside the tablet to say the result is ready. And then, there is another color of light that comes directly from the lateral flow strip, that’s an indicator that the virus is present.
One last example is a nucleic acid test. This rectangular, hand-held device (see photo), reminiscent of a computer disc, looks inside the virus to amplify small traces of its nucleic acid to detectable levels. It is completely self-contained. To find that technology today, you generally must go to complex laboratories where the test is done on big machines, operated in multiple steps. Efforts are being made to reduce the size and the complexity of these devices so they can move out to point of care, without sacrificing the performance that we expect from a laboratory-based device.
Collins: That’s totally cool. Is the nucleic-acid test device that you just mentioned made for one-time use, and then you throw it away?
Tromberg: That’s their business model right now. I should probably mention something about cost. For example, you can imagine scaling up lateral flow assays very quickly to make tens of millions of tests. The components are inexpensive, and the tests may cost just a few dollars to make.
If you’re throwing away a nucleic acid test with its more-expensive components, obviously, the cost will be higher. Right now, if you go to a laboratory for a nucleic acid test, the cost may be on the order of $40 or so. With these one-time-use nucleic acid tests, the aim is to scale up the manufacturing to produce larger volumes that will bring the cost down. The estimates are maybe $60 per test.
Collins: That needs to come down more, obviously. In the months ahead, we’re talking about testing millions of people, maybe even fairly often to make sure that they haven’t been infected by SARS-CoV-2, the novel coronavirus that causes COVID-19. Is frequent testing the kind of thing that you’d like to be able to do by next fall?
Tromberg: Yes, and I think that that really speaks to the diversity of the types of tests that we need. I think there is a market, or the capacity, for some of the more expensive tests, if they’re extremely accurate and convenient. So, the nucleic acid test may cost more, but it will give you an answer very quickly and with very high sensitivity. It’s also very convenient. But the performance of that test may be very different from a standard lateral flow assay. Those tests will be far more accessible and very, very inexpensive, but they may have a higher false negative rate. We envision that every test that comes out of our innovation funnel will have documentation about its best-use case.
Collins: You mentioned your innovation funnel, sometimes called a “shark tank.” Say a little more about the RADx-tech shark tank. Who gets into it, and what happens when they get there?
Tromberg: At NIH, we’re into processes, and NIBIB created a very effective one 13 years ago with the Point of Care Technology Research Network (POCTRN). We’ve now leveraged this network to focus almost exclusively on COVID testing. POCTRN has five sites in the US. All have core resources, personnel, and expertise that are contributing to RADx-tech. Those include the ability to validate tests independently, the ability to do clinical studies in real-world samples and patients, and the ability to analyze manufacturing and scale-up needs while creating a roadmap for every project team to follow.
We have more than 200 people around the country working day and night on this process. If anyone has an idea about a COVID-19 test, you can and apply for funding on the POCTRN website. Your application will be reviewed by a panel of 30 experts within a day and, if approved, will move to the next stage, which is the shark tank.
In the shark tank [also called phase zero], a team of experts will spend about 150 to 200 person-hours with you evaluating the strengths and weaknesses of your test technically, clinically, and commercially. From this careful analysis, a detailed proposal will be presented to a steering committee, then sent to NIH. If we think it’s a great idea, the project will enter what we call phase one, with considerable financial support and the expectation that the company will hit its validation milestones within a month.
Collins: How far have things progressed, given that you just started RADx on April 29?
Tromberg: We have almost 60 projects that have entered or emerged from this shark-tank stage. I’m expecting that we’ll have around 15 projects in the phase one stage this month, and it’s very exciting to see them move there. If they can reach their validation milestones in that first month, they will be eligible to move to phase two. It involves a much larger chunk of money, so companies can move into manufacturing and scale up for distribution. We’re hoping to have between five and 10 companies emerge over time from this innovation funnel. But, by the end of the summer, we’d like to see at least two come out with products that will make a difference.
Collins: Wow, that’s just a few months away. How will you can get there so fast?
Tromberg: Sure. Some companies are further along than others. I can think of one that is quite far along with an established platform concept. This company has lots of expertise and has raised lots of money. We may be able to give them the surge that they need, plus the additional support with regulatory issues, commercialization, and manufacturing, in that short period of time to go to market.
Complementing that work is another of our initiatives called Advanced Technology Platforms (RADx-ATP). It’s designed to scale up existing technologies. For example, I mentioned a one-time-use nucleic acid test. It still needs validation, emergency use authorization, a little bit of manufacturing optimization. But we have other platforms out there that are much closer to commercialization, and RADx-ATP could be very impactful in getting some of those technologies out earlier.
Collins: You mentioned RADx-ATP, and we’ve been talking about RADx-tech, which is your shark tank approach. But there are a couple of other RADx components. Say something about those, please.
Tromberg: Our centerpiece component for doing demonstration projects is called RADx-UP. This is an effort across NIH to provide cutting-edge testing technologies in underserved populations. If I’m allowed to be the interviewer and turn the tables, I might bounce the question back to you. This is where your thinking directly influenced the whole RADx portfolio. So, maybe you can tell us more.
Collins: I can try. It’s very clear that COVID-19 has hit certain populations particularly hard, especially African American and Hispanic communities. And yet, those communities often have the least access to testing, which is sort of upside-down. We want to help identify people who are infected quickly, do the quarantining, and prevent the infection from spreading. That has simply not worked very well in a lot of underserved communities.
With resources from Congress, we made it a very high priority to set up demonstration projects of these advanced technologies in communities that would benefit significantly from them. We’re trying to bring together two really important NIH priorities: technology development and addressing health disparities. I’ve got to say, at this particular moment, when we’re all really focused on the fact that our nation is still riddled with health disparities, health inequities, and even racism, this is a moment where we should be doing everything we can to try to take our scientific capabilities and apply them to finding solutions.
So, we’re all pretty excited about RADx-UP. But there’s one other RADx, and I’ll throw this one back to you. It’s called RADx-rad. What the heck is that, Bruce?
Tromberg: Well, RADx-rad is the home for the technologies that are really far forward and futuristic. These are the technologies that won’t quite be ready for the time pressure of the innovation funnel. But they’re fantastic ideas. They’re projects that may be non-traditional in terms of the application of technology. They have been generated largely by other NIH institutes and centers. They’re important ideas and projects that just need to be supported with a longer time-window of return. We don’t want to lose out on the energy and the ideas and the creativity of those concepts.
Collins: Right now, the focus is on COVID-19 and the need for testing, especially within this calendar year. We hope, by the end of 2020 or the early part of 2021, to have vaccines for COVID-19 ready to go. But, moving forward, there will be other events that will probably make us wish that we had point-of-care diagnostics. So, in the process of doing what you’re doing with all of these components, hopefully we’re also preparing for future challenges.
Bruce, you’re an optimistic guy. At the same time, we’ve got to be realistic. Around September, when schools and colleges are contemplating whether it’s safe to open up, what would we hope that RADx could contribute to make that a better outcome?
Tromberg: That’s a tough question to answer, but I have a lot of confidence in our process. I’m confident that we’re engaging the innovation and entrepreneurial community in such a way that a lot of these ideas will move out and give us better performing tests and more of them. A rough number that I like to think about is the capacity to test roughly 2 percent of the population, around 6 million people per day. I think we’ll hit that target by the end of the year.
I’d like to see testing technologies move away from being based predominantly in laboratories. I’d like to see them more accessible to people as technologies that they can use in their homes. We’re now doing so many things from home. We’re working from home, we’re talking from home, we get our entertainment from home. Home-based testing is really the direction a lot of healthcare is going. We need to have these technologies. I think the level of sophistication and performance that we’re hoping for is possible, and the innovation and entrepreneurial community is working extremely hard to make it happen. No one has really asked us to do anything of this scale before, and I like to compare it to our Super Bowl.
Collins: Well, this is one exciting Super Bowl, that’s for sure! You’ve applied the venture capitalist strategy to RADx of trying to discover what’s out there, while not being afraid to invest in risky endeavors. You’re figuring out how to help promising technologies take their best shot and fail early, if they’re going to fail. And for technologies that are further along, you give them the needed resources to advance to commercialization.
We have great hopes and expectations that RADx will make a real difference. What we’re doing here is not just about cool science, it’s also about saving lives. I want to thank you for your incredible dedication, and your intellectual and engineering contributions to this initiative, which make it one of the most exciting things that NIH is doing right now.
Tromberg: Thank you, Francis.
Coronavirus (COVID-19) (NIH)
Rapid Acceleration of Diagnostics (RADx)
“Social engineering and bioengineering together can thwart the COVID-19 pandemic,” Director’s Corner, National Institute of Biomedical Imaging and Bioengineering/NIH)
Video: RADx Tech and POCTRN: Diagnosing Disease-Delivering Health (NIBIB/NIH)
Posted on by Dr. Francis Collins
My last post highlighted the use of artificial intelligence (AI) to create an algorithm capable of detecting 10 different kinds of irregular heart rhythms. But that’s just one of the many potential medical uses of AI. In this post, I’ll tell you how NIH researchers are pairing AI analysis with smartphone cameras to help more women avoid cervical cancer.
In work described in the Journal of the National Cancer Institute , researchers used a high-performance computer to analyze thousands of cervical photographs, obtained more than 20 years ago from volunteers in a cancer screening study. The computer learned to recognize specific patterns associated with pre-cancerous and cancerous changes of the cervix, and that information was used to develop an algorithm for reliably detecting such changes in the collection of images. In fact, the AI-generated algorithm outperformed human expert reviewers and all standard screening tests in detecting pre-cancerous changes.
Nearly all cervical cancers are caused by the human papillomavirus (HPV). Cervical cancer screening—first with Pap smears and now also using HPV testing—have greatly reduced deaths from cervical cancer. But this cancer still claims the lives of more than 4,000 U.S. women each year, with higher frequency among women who are black or older . Around the world, more than a quarter-million women die of this preventable disease, mostly in poor and remote areas .
These troubling numbers have kept researchers on the lookout for low cost, but easy-to-use, tools that could be highly effective at detecting HPV infections most likely to advance to cervical cancer. Such tools would also need to work well in areas with limited resources for sample preparation and lab analysis. That’s what led to this collaboration involving researchers from NIH’s National Cancer Institute (NCI) and Global Good, Bellevue, WA, which is an Intellectual Ventures collaboration with Bill Gates to invent life-changing technologies for the developing world.
Global Good researchers contacted NCI experts hoping to apply AI to a large dataset of cervical images. The NCI experts suggested an 18-year cervical cancer screening study in Costa Rica. The NCI-supported project, completed in the 1990s, generated nearly 60,000 cervical images, later digitized by NIH’s National Library of Medicine and stored away safely.
The researchers agreed that all these images, obtained in a highly standardized way, would serve as perfect training material for a computer to develop a detection algorithm for cervical cancer. This type of AI, called machine learning, involves feeding tens of thousands of images into a computer equipped with one or more high-powered graphics processing units (GPUs), similar to something you’d find in an Xbox or PlayStation. The GPUs allow the computer to crunch large sets of visual data in the images and devise a set of rules, or algorithms, that allow it to learn to “see” physical features.
Here’s how they did it. First, the researchers got the computer to create a convolutional neural network. That’s a fancy way of saying that they trained it to read images, filter out the millions of non-essential bytes, and retain the few hundred bytes in the photo that make it uniquely identifiable. They fed 1.28 million color images covering hundreds of common objects into the computer to create layers of processing ability that, like the human visual system, can distinguish objects and their qualities.
Once the convolutional neural network was formed, the researchers took the next big step: training the system to see the physical properties of a healthy cervix, a cervix with worrisome cellular changes, or a cervix with pre-cancer. That’s where the thousands of cervical images from the Costa Rican screening trial literally entered the picture.
When all these layers of processing ability were formed, the researchers had created the “automated visual evaluation” algorithm. It went on to identify with remarkable accuracy the images associated with the Costa Rican study’s 241 known precancers and 38 known cancers. The algorithm’s few minor hiccups came mainly from suboptimal images with faded colors or slightly blurred focus.
These minor glitches have the researchers now working hard to optimize the process, including determining how health workers can capture good quality photos of the cervix with a smartphone during a routine pelvic exam and how to outfit smartphones with the necessary software to analyze cervical photos quickly in real-world settings. The goal is to enable health workers to use a smartphone or similar device to provide women with cervical screening and treatment during a single visit.
In fact, the researchers are already field testing their AI-inspired approach on smartphones in the United States and abroad. If all goes well, this low-cost, mobile approach could provide a valuable new tool to help reduce the burden of cervical cancer among underserved populations.
The day that cervical cancer no longer steals the lives of hundreds of thousands of women a year worldwide will be a joyful moment for cancer researchers, as well as a major victory for women’s health.
 An observational study of Deep Learning and automated evaluation of cervical images for cancer screening. Hu L, Bell D, Antani S, Xue Z, Yu K, Horning MP, Gachuhi N, Wilson B, Jaiswal MS, Befano B, Long LR, Herrero R, Einstein MH, Burk RD, Demarco M, Gage JC, Rodriguez AC, Wentzensen N, Schiffman M. J Natl Cancer Inst. 2019 Jan 10. [Epub ahead of print]
 “Study: Death Rate from Cervical Cancer Higher Than Thought,” American Cancer Society, Jan. 25, 2017.
 “World Cancer Day,” World Health Organization, Feb. 2, 2017.
Posted on by Dr. Francis Collins
As long as she can remember, Ashley Matthew wanted to be a medical doctor. She took every opportunity to pursue her dream, including shadowing physicians to learn more about what a career in health care is really like. But, as Matthew explains in today’s LabTV video, she also became attracted to the idea of doing research because of her affinity for solving problems and “figuring things out.”
So, Matthew decided to give biomedical research a try, landing a spot in an undergraduate summer program sponsored by the University of Massachusetts. Ten weeks later, she was convinced that her future in medicine just had to include a research component. That’s why Matthew is now well on her way as an M.D./Ph.D. student at the University of Massachusetts Medical School, Worcester, where she works in the lab of Celia Schiffer.