Latest News

Building Diagnostics With Data In Mind

December 18, 2019 | Ativa’s diagnostic workstation claims to be the only point-of-care system using a big-data approach to blood and urine testing to provide results on the major test panels used today to diagnose patients, and advanced diagnostic tests that gives deeper, more concise, disease information. One drop of blood—finger stick or venous—is applied to an Ativa test card. The card is read with a small analyzer and a digital blood sample is created that is interpreted by either conventional or machine-learning-based algorithms. The results are returned in three to five minutes while the patient is with the doctor.

If that sounds to you too good to be true, you aren’t alone. But David Deetz, co-founder and CEO, is quick to point the company’s secrets to success: a team approach with solid partnerships, AI-driven analysis, and a seismic shift in the way data is collected for diagnostic testing.

On behalf of Diagnostics World News, Kaitlin Searfoss Kelleher recently spoke to Deetz about the future of point-of-care testing, and Ativa’s approach.

Editor’s Note: Kaitlin Searfoss Kelleher, Senior Conference Director at Cambridge Healthtech Institute, is planning a conference dedicated to Point-of-Care Technologies next year at the Molecular Medicine TriConference, March 1-4 in San Francisco. Deetz will be speaking on the program; their conversation has been edited for length and clarity.

Diagnostics World News: So David, at Ativa, you and your team have developed a breakthrough point-of-care test that utilizes AI and machine learning to test all of the major blood, metabolic, and urine panels used in primary care in one system. How did you develop this system?

David Deetz: Good question. It was our belief that no one company had the know-how needed to create a system as complex as the Diagnostic Workstation. We knew it was going to take a team of companies to do this. We looked at all the pieces that were needed and started out with Honeywell as our partner. We had our own labs within Honeywell, and we worked side by side with their scientists to get the initial development done. Honeywell is best-in-class in microfluidics and micro-control technology.

Next, we worked with 3M personnel to solve the disposable cost issue. We felt the only way to meet the cost target so that it could be adopted worldwide was to utilize three-dimensional tape for the disposable card.  Of course, 3M is best-in-class in utilization of tapes and they were also our neighbors because the headquarters is right by us. We worked closely with them and with the tape processors in town. We again had some of our people based within the tape processing facility and worked closely with their experts. It is not just the basic knowledge, it's the working knowledge and that is gained by close proximity.

Today many companies are bolting AI on to their existing systems. However, in order to get full utilization of AI and machine-learning capabilities, you need to design it in on day one. We created the architecture early on and the data integration strategy. In order to execute on our strategy, we brought in Bosch as a partner. Similar to Honeywell and the tape processing, Bosch had people full time at Ativa to intertwine all the digital health aspects of this. LabCorp was also guiding and supporting us. We did an awful lot of work ourselves on this, but we needed some specialists in some of these other areas. I am immensely proud of what our internal team accomplished.

A quick Google search will show a lot of articles comparing your technology to Theranos. How are you different?

Well, first of all, we are similar in our beliefs that blood testing has to change and be available to all people. Theranos helped to highlight the urgent un-met needs and that the market demand is huge. However, their technical approach was different, Theranos was basically taking conventional lab systems and putting all the big mechanical pieces into a box, a single box, and it was more of an engineering challenge which they tried to do all internally. They focused on, “Can we make this compact enough to fit into one single box?” That box would have cost $120,000 and because it’s conventional equipment, only a certified lab technician could run it.

We created some new science here at Ativa. We have over 200 patents and shrunk all the fluidic aspects of these big machines onto something the thickness of five pieces of tape. Some of these tapes are the thickness of only a couple hairs. We really miniaturized the entire process—micro-miniaturized it in fact. This low-cost consumable enables affordable and access to much-needed testing for people in all regions of the world.

Our technical breakthroughs took away all moving parts, the things that break, the things that cost a lot or require an expert in the field to run it. We started with human factors as first principle and designed it so that almost anybody could run this test and get lab comparable results. What we have now nurses or nursing assistants can run the tests.

Our system, unlike Theranos’ system, was also developed with AI in mind which is important for many reasons; this was because of our AI background which is not common in this field. At Diametrics Medical, a POC company, which we took public on NASDAQ, we utilized machine-learning to overcome the major hurdle in that field. At MultiLogic we developed JP Morgan’s Wealth Engine and Citigroup’s Restricted Stock Engine applying some of the same principles used in Ativa’s Diagnostic Workstation.

We have designed our Workstation for the doctor's office at first, but this technology will evolve, it will get smaller, it will get more capable. It will eventually lead to a home health appliance, with symptom-based cards, A.I. and telemedicine capabilities, probably within a decade.

You're also bringing in a lot of AI and machine learning to make this device possible. Could you tell us a little bit about the work you've done here, maybe setting the stage for a change in how blood-based diagnostics will be developed?

We think that this field is going to change drastically. Instead of taking an approach like Abbott or Siemens where they develop the whole complex lab system that gives you the basic numbers, I think it's going toward more of a big data approach. Previously it's been “big chemistry”: you do all this chemical processing, the separation, this specific reaction, and then you measure that. It's very chemistry intensive and expensive. Well, it's going to become more data intensive, which is much cheaper. The diagnostics companies will become data generators, and then they will partner with companies such as the Googles of the world, the Microsofts of the world, the Boschs of the world, that will help interrogate the big data to give it meaning. This process will allow you to unlock more of the information that is locked in the bloodstream.

Right now diagnostics can only derive very limited things from the bloodstream, yet the blood has so much information in it. Almost every process in your body leaves a trail in the blood. There are fingerprints of what's going on. They're very complex, but they're there; machine-learning can find them.

Now researchers look for a biomarker, a single molecule that tries to tell us about a complex disease. Well, almost every process in your bloodstream involves twenty to over a hundred variables at once. It's very unlikely that the disease is going to characterize itself in just one marker. So if you really want to look for complex diseases, it's more likely going to be a disruption of the healthy pattern, it's going to have a fingerprint that is unique to that disease. There are over 5,000 diseases yet today there are only a hundred or so definitive biomarkers. So patterns are going to become the new biomarkers, and the ones who are going to be really gifted at extracting that are going to be the big-data companies, the Microsofts, the Googles.

Ativa will create a huge amount of cell-level blood and urine data, and it can be interrogated for twenty, thirty, forty different diseases at once. Ativa is going to work with many clinical specialists to feed in definitive samples that can teach our A.I. to identify the fingerprints of the highest priority diseases globally. There's no new chemistry needed, it's all data interpretation. So that's going to allow us to unlock so many of the secrets of the bloodstream that we haven't been able to touch before.

This is a major breakthrough, and it's going to have a huge impact on our ability to diagnose patients rapidly and at very low cost at the point of care. The blood is your major database and being able to unlock that is going to change now. Ativa will unlock unprecedented menu growth with doctors creating “Disease-Apps” rather than biomarkers. A new biomarker usually takes 6-10 years and $50-100 million to develop, Ativa’s big-data approach could cut this to months. New tests will be added with software updates. Our initial results are already very promising and we are moving quickly now to prove this disruptive approach.

This absolutely is a major breakthrough, and it's really exciting to hear you talk about it. But like any new scientific approach, there have to be roadblocks. Could you tell us a little bit about the challenges that you faced and what challenges others who may want to venture down this road can expect?

I think the biggest barrier will be the resistance to change in thinking about measuring disease. It is such a different way of thinking about the problem, especially for the conservative diagnostics field. And we're all human, our minds, as they become more educated, become a little bit more rigid. So, it's going to be a human problem, that will be the biggest barrier. So Ativa overcomes this barrier by initially providing the most commonly ordered tests today, at the point-of-care, which is an accepted but limited menu today. As Ativa builds evidence and acceptance of machine learning apps we will simply provide these as software menu options to provide clinicians advanced disease screening as an incremental change.

Although many people recognize the benefits of machine-learning in healthcare there are many substantial barriers to implementation. To access patient data for mining there are many roadblocks, you need deep partnerships with the healthcare providers and also then EMR companies. Ativa avoids these barriers by creating our own de-identified data set. Healthcare providers have told us they need our technology now.

What is your advice for companies interested in embarking on more machine learning and data aggregation and analysis to reach your vision of more handheld and in-home point-of-care diagnostics in the next 10 years?

Start thinking about your data strategy first, your data generation platform, and be creative in ways that you can generate more data before you start designing your product. Think about large unstructured datasets. This is hard for people in diagnostics today who are used to focusing on specific numbers, ranges or cut-offs. Put in the digital backbone that other people can utilize. Then form a partnership, like we've done with Honeywell and Bosch; they really worked with us to create this data generation capability. We had to go outside the diagnostic industry to get this kind of capability.

Load more comments
comment-avatar