Pharma Innovators Special Feature 2018: Focus on innovation
In seeking innovative players that could change pharma and healthcare, Med Ad News found the developer of an app that helps people determine what illnesses are in their neighborhoods; the creator of a wearable injector that allows patients on biologics to receive these drugs outside the clinic; and a designer of a deep learning network aimed at giving pharma and healthcare companies a handle on their data.
Christopher Bouton, CEO of Vyasa Analytics
From Johns Hopkins to Pfizer to starting up two companies, Vyasa Analytics CEO Christopher Bouton, Ph.D., has made a career of how people think and how to make machines think like people.
Beginning with a Ph.D. degree in molecular neurobiology from Johns Hopkins, with an interest in computational biology that led to working at a couple of biotechs in the Cambridge area, he was recruited to be the head of integrated data mining at the Pfizer Regional Technical Center in Boston.
“I was at Pfizer for eight years, and that really was a wonderful opportunity to understand the breadth of data in use within pharma companies, and how important the processing of data is to research, development, and ultimately healthcare for these types of companies,” he told Med Ad News.
After leaving Pfizer, Dr. Bouton founded a company called Entagen in 2008, which provides Big Data integration and analytics solutions. Entagen’s technologies were named “Innovative Technology of the Year in Big Data” in 2012 by the Massachusetts Technology Leadership Council and Entagen was named a Gartner “Cool Vendor” in the Life Sciences in 2013. Thomson Reuters acquired the company in 2013.
After spending two and a half years at Thomson Reuters, Dr. Bouton says he was “really bitten by the startup bug” and left Thomson Reuters in the beginning of 2016.
“I sort of sat back and tried to understand what I wanted to do next,” he says. “And right around that time is when these deep learning technologies started to become more talked about, and hit their stride in terms of their capabilities. And what was interesting to me about these deep learning algorithms is that in many ways they are based on cortical structure. And so from my neurobiology experience, it was really cool to see an algotrith that was in part modeled after how the brain works. That was sort of an aha moment.”
In that moment, the concept for Vyasa was born.
“What I am really interested in doing is separating the hype from the reality of what these algorithms are really good at, and then once we’ve identified key paradigm-shifting capabilities of these technologies, applying them in a very specialized way, in the life sciences and healthcare space,” he says.
Dr. Bouton says all of the hype around artificial intelligence (AI) is being driven by deep learning algorithms. “AI, the phrase, is one people have thought about ever since we’ve been using computing machines,” he says. “But what’s happening now is that deep learning algorithms are really driving this round of game-changing capabilities for machines and what they could do.”
Deep learning algorithms “simply weren’t at the level of capability that they’re at today when I was working at Pfizer,” he says. “There are plenty of times when I had hoped we could do at Pfizer the kinds of things that we can do now. And that comes back to the perfect storm of capabilities all coming together right now. GPU hardware is at a level of capability that it’s never been at before, and GPUs are very important for these deep learning algorithms; the Big Data phase, learning how to handle large amounts of data, was an important phase for all of us to go through; and then updates to the algorithms themselves are all coming together right now and give us these capabilities that just weren’t ever possible before.
“We’re all – and I mean by all, I mean humanity – just learning what these things are good for and just how far we can apply these types of capabilities,” Dr. Bouton says. “It’s really a revolution, a paradigm shift in understanding what a machine can do. And that means that there’s a wide range of low-hanging fruit to apply these algorithms to, and beyond that, a range of capabilities that are going to generate a significant amount of value for pharmas, healthcare organizations, and beyond.”
The company’s name is inspired from Dr. Bouton’s experiences from living in India as a child. Vyasa – which in Sanskrit means “compiler” – is the ancient Hindu sage who codified the Vedas, the oldest scriptures of Hinduism, and he is the author of the Mahabharata, one of the two major Sanskrit epics of ancient India.
Dr. Bouton and Vyasa are building a platform called Cortex, which merges Big Data, or highly scalable capabilities, with specialized deep learning algorithms for the life sciences and healthcare space.
“One of the things that’s so incredible about these deep learning algorithms is that there are many places where they’re going to become valuable,” he says. “Chatbots are a great example, Alexa is another great example, there is a lot of deep learning in Alexa. But it really extends far beyond that – self-driving cars, the ability for a self-driving car to identify a pattern and then respond to that pattern without a rule set.”
With deep learning algorithms, “you can train a machine on patterns, and how the machine ultimately starts to recognize patterns that are not necessarily what you originally show it. It’s pattern recognition technology and the capability to respond to patterns is very different from what we’ve been doing with machines previously,” Dr. Bouton says.
One of the best examples of what can be done with deep learning algorithms is AlphaGo, a game playing technology built by Google, in which an artificial technology can actually play the ancient Japanese game. “Go is very nuanced, and has a lot to do with patterns, but it’s very hard to define a rule set for it,” Dr. Bouton says. “And people didn’t think that someone was going to create a machine capable of playing Go at an expert level for a very long time.”
Vyasa’s Cortex program is built on a proprietary AI engine called Layar – “which is a play on the idea of the cortex, because the cortex has many layers,” Dr. Bouton says. Layar was built to be capable of consuming content from many different sources so it is highly scalable.
“We’re merging Big Data capability with AI capability so you can apply these things at scale. And Layar is API driven, so there is a wide range of capabilities built into it.”
Cortex is the first app built on top of Layar. Dr. Bouton says the program gives an end user the ability to interact with all of the core AI features of Layar and apply them to their own data in a secure way behind their firewall. “This is very much a software platform that we’re delivering to end clients so that they can use these capabilities internally within their firewalls in a secure way.”
Vyasa has identified three capabilities for deep learning algorithms in pharma and healthcare: text analytics, image analytics, and small compound analytics. “So what we’ve done is build into Layar specific deep learning algorithms that are specialized for life sciences data in those three areas,” Dr. Bouton says.
One area Vyasa’s deep learning programs can make a difference in is the small compounds space. The company has developed an algorithm called ChemVector, which is capable of doing de novo compound design.
“The AI can actually identify what you want to optimize for a compound in a series, and then generate novel compounds for the chemist,” Dr. Bouton told Med Ad News. “What’s incredible is that as humans, we’ve only ever generated about 108 compounds, but there are about 1080 compounds possible from the combinatorial point of view. And that means we can set the AI on a path to generating far more compounds than is what is sort of conceivable for humans. This becomes a powerful tool.
“When you apply that in conjunction with better toxicity predictions, better admin predictions, you start to build up a portfolio of capabilities driven by AI approaches that are a powerful new way to look at for chemists.”
Another pharma area that deep learning algorithms can be used in is image analytics, to conduct crystal morphology studies for tablet formulation. Dr. Bouton says these AI systems can improve accuracy in terms of identifying various types of crystal structures.
In text analytics, Vyasa has done a lot of work in concept recognition.
“One of the traditional examples of deep learning is that we can train a deep learning engine to identify a dog in an image or a cat in a image just by showing it many examples of pictures of dogs and cats,” Dr. Bouton says. “A big ‘aha” moment for us was, ‘Why can’t we do this with concepts and data?’ And so we’ve built out, in Cortex, a whole proprietary deep learning capability to train an AI on things like the names of diseases or their enzymes or pretty much any of their named identity. And then the system can crawl out into unstructured content and identify novel instances of those types of things for you without rule sets.
“If you’re trying to identify new enzymes in your content, they just train it on a set of enzyme names, and then it starts to learn what an enzyme name looks like, without you having to teach it something like, ‘It has to have -ase at the end of the name’ or whatever. It’s really quite powerful and a more robust way of identifying the naming of things in content.”
Text analytics can also be used in EHR systems or in clinical trials to identify all the ways that somebody has referred to a given symptom or outcome.
To prove its capabilities, Vyasa is starting with very focused clients, identifying key use cases, and then building off of those use cases. “We’re doing that with around 10 clients right now and really want to make sure we enable valuable use cases for those clients,” Dr. Bouton says. Beyond that is when we will really start to grow that client base, and I project that toward the end of 2018 going into 2019.”
Building these case studies is not just valuable for Vyasa, it gives the company the opportunity to demonstrate to clients “that there is value behind the hype for these approaches,” Dr. Bouton says.
In addition to life science and healthcare companies, Vyasa’s technologies have attracted interest from legal firms and marketing companies, particularly in the text analytics space.
“Our eyes have been opened to the values of these technologies beyond the verticals we started out in,” Dr. Bouton says. “The growth of the company, we’re really excited about where we’re going, because we do see the applicability of these technologies beyond these verticals. But we also want to remain focused on specializing in these areas, life sciences and healthcare.”
At present, Vyasa is at 15 employees, but plans to grow, to at least double in size. “And we plan to be profitable this year, which is also exciting,” Dr. Bouton says.
Graham Dodge, CEO of Sickweather
If you have ever woken up with what you believe is a cold or flu and you and your friends on Facebook check in with each other believing there is “something going around,” now there is an app for that called Sickweather, which is a sickness forecasting and mapping program.
“Just as Doppler radar scans the skies for indicators of bad weather, Sickweather scans social networks for indicators of illness, allowing you to check for the chance of sickness as easily as you can check for the chance of rain,” the company’s website says. “Use Sickweather to track illnesses, compare symptoms, and see which viruses are going around in your area.”
CEO Graham Dodge’s career initially was not focused on the healthcare space. “My background was actually in mapping crime data, so I had some experience scraping crime data, sourcing crime data, working with federal employees to gain access to data,” he told Med Ad News. “The biggest challenge of that was how to visualize it, how to show it on a map. This was back in the late ‘90s.”
One day, however, Dodge was sick with a stomach bug, “and I just wanted to know if something was going around or if it was food poisoning.
“I had a newborn baby in the house, and I didn’t want them sick and I wasn’t really quite sure as a nonmedical professional what to do. So I turned to Google and I turned to CDC, and my local public health department.”
After failing to find anything helpful, relevant, or accessible to him on any of these sites, Dodge happened to go on Facebook and saw a friend reporting the same exact symptoms.
“That’s when it occurred to me that social media worked like a real-time census for reporting these types of things,” Dodge says. “And I thought there could be a way of taking the data that people were volunteering out there, distilling it, and giving it back to the people in a format that has more value to it than before. I felt that disease surveillance and health trends really lend themselves to developing a tool, similar to my crime data days, to make them accessible to people and put them in a format that was easy to understand, given the fact that these were data that people were already volunteering and putting out there. So rather than us doing it behind closed doors and only selling it to companies, our first, primary goal was to make it accessible to consumers and see if it has value to other parents like me.”
Dodge describes Sickweather as a “Waze for illness.” Just like the popular traffic mapping program has pop-up ads for sponsoring restaurants in the area of travel, one way for companies in the pharma and the healthcare space to get involved with Sickweather is as a map sponsor.
For example, when Walgreen’s was a sponsor, the locations of its stores were overlaid on Sickweather’s maps during flu and cold season. “We had all 8,000 Walgreens locations in the map and it was helping to drive people to those destinations” for OTC remedies, Dodge says.
The technology even has been used to recruit participants for clinical trials. “We did a pilot with Radiant Research, which is now Synexus, for a common cold nasal remedy that needed people who were symptomatic within the first 48 hours,” Dodge says. “That was a real challenge for them, so they used Sickweather’s platform to serve messages to people, in real time, when they were reporting illnesses to Sickweather and whether they were in a certain radius of a testing location.”
From this campaign, Radiant saw a conversion rate of more than 10 percent. “Considering that digital advertising usually has a conversion rate of 1 percent or less, in this case we’ve seen 10 percent and higher conversation rates of people who wanted to participate because it was extremely relevant to them, they were experiencing the symptoms and there was the convenience of the centers being nearby,” Dodge says.
According to Dodge, Sickweather knows where sick people are and where they are going to be, and that value stretches between patient recruitment for clinical trials to driving sales for over-the-counter and prescription medications, and enabling those companies to more effectively target their advertising.
“We can tell Theraflu exactly when and where to be triggering ads for their product, same with Flonase and any number of over-the-counter medications,” he says. “We do know that in working with these companies, they see a two-to-three times higher clickthrough rate, so higher engagement rates on those ads, and sales lifts of at least 4 percent. That is 4 percent greater than when they are not using Sickweather for the data to trigger those ads.”
Sickweather already powers about $320 million a year in ad spending, Dodge says. “We provide the data that triggers that spending,” he told Med Ad News. “Our own revenues aren’t quite there yet but that’s the kind of influence we have so far. We have several programmatic platforms using our data. Several brands are using it, not only OTC but CPG ones, including Clorox and Honeywell air filters.”
Between the crowdsourcing done through Sickweather’s app, and the social listening that the company conducts on Facebook and elsewhere, the company has a pretty well-rounded data set that shows what that consumer intent or need is at any given time in the population.
“We’re really a population health monitoring tool, in that sense, because we can not only assess what the consumer intent is going to be but what the risk is going to be,” Dodge says.
For example, when hospitals are discharging patients, and there are increased reports of pertussis or pneumonia, Sickweather can help these organizations determine if their patients are vulnerable to the complications may be facing from these diseases. “That may impact certain treatments and therapies, especially if people come back into the system,” Dodge says. “So I think that the pharmaceutical companies would want to know that information because it helps them understand where their supply chain is needed, why is it going to be needed, in advance.”
And Sickweather’s clients in the pharma space already use the data to predict sales of their products, “so they can use that not only for guidance to their leadership, but to make strategic decisions around marketing campaigns and advocacy campaigns well in advance,” Dodge says.
Pharma companies would also find Sickweather helpful in programmatic ad buying, Dodge says, as the company deals in forecasts 15 weeks in advance and can do that down to the city level. “We’ve been calling them regional forecasts but really they go all the way down to the metro area,” he says. “And it will allow retailers and pharmaceutical companies to understand where their products are needed. For the retailers, it could be pretty meaningful in advance to know where to send, where to stock certain medications so they’re not expiring over time.”
Besides the common cold and pertussis, Sickweather also tracks asthma and the vectors of illness, such as mosquitoes and ticks. “Those are areas we know that as we gain more data, we would be able to start creating the forecasting models to predict those things,” Dodge says. “There is already evidence that it can show in real time where these things are happening down to the street level..All of our data is tagged to the latitude and longitude. We can reverse geocode in several different geographies and we can compute the data as well, not only predicting it throughout time, but also geographically where it will spread. Predicting where it’s going to be and when it’s going to be a problem.”
Besides reports on social media and through the Sickweather app, the company also gets data from its partnership with the Weather Channel. Sickweather has replaced the CDC as the flu map provider for the Weather Channel, as well as providing common cold data. “The Weather Channel app users report to Sickweather through the Weather Channel app and Website, and it all goes to Sickweather’s servers,” Dodge says.
Sickweather also validates against other clinical data received by third parties for sales data. “We’re always looking for other data sets to correlate to,” Dodge says. “Although I think by this date, we’ve correlated enough time now to a variety of data sets that we’re very confident that our process is accurate and will reflect what the clinical trends will be showing, 10 days, two weeks, two months later.”
As Sickweather looks to grow, Dodge says it is less about creating new types of products and more about concentrating on different types of surveillance, different therapeutic areas, and different countries. “Our efforts now are for localization in Japanese and Chinese,” he says. “We are now translated into 10 different languages. Those have not been integrated into the consumer app yet, but soon will be.”
Another key to Sickweather’s future is the recent hiring of Laurel Edelman as chief revenue officer.
“She really personifies the direction that we’re heading into as a company, by virtue of her experience in this space, having worked with IMS Health previously and SDI before that and AstraZeneca before that,” Dodge says. “The modeling that we do, the forecasting that we do, is a different kind of forecasting than somebody else may be doing. While they’re looking at what the overall seasonal trends are going to be, we’re actually looking at the week-to-week trends by city. We’re adding a level of granularity that doesn’t exist.”
Mike Hooven, CEO of Enable Injections
Patients on certain complex biologics face a saddening reality: spending a few hours at least once a month, or perhaps every week, being connected to an I.V. at an outpatient hospital center getting the drug they need to live their lives.
Mike Hooven, CEO and founder of Enable Injections, wants to change all that with a glowing device that looks like half a yo-yo or a miniature Iron Man power heart. The device, activated and placed on a patient’s abdomen, slowly administers complex biologics subcutaneously without the need to go to an infusion center.
Hooven’s medical device career began 37 years ago when he started working at Cordis Corp. in neurosurgical valves, His mentor, Norman Weldon, is “a legend” in medical devices, Hooven says. “He has started or has helped start about 20 different companies … and has been my mentor since I started working. Norm has helped me start all of my companies, and has been on the board of all of my companies, along with a number of other folks I have worked with over the years.”
After five years at Cordis, Hooven went to Siemens Pacesetter, this time in pacemaker development. Then Hooven left Siemens to come to Cincinnati, taking a job to run all internal product development for a company that started out as Ethicon Mechanicals, and eventually became Ethicon Endo-Surgery.
“And the person that hired me in there, by the name of Rick D’Augustine, is another one of my best friends and has been on all the boards of my companies as well,” Hooven says. “When I interviewed for the job, I was asked why I wanted to work there, and I said I wanted to work at a Johnson & Johnson company, get the experience, make the contacts, and start my own business in about five years. And he hired me anyway. And it took me six.”
In 1994, Hooven left J&J to start his first company, Enable Medical Corporation, and built that up with almost no dilution. “When I quit J&J, I cashed out all my money in the bank, all my stock options – I had $25,000, I took $22,000 and put that into the business, with $3,000 from Norm and his partner – and we did really well,” Hooven says. “It took off, and built that business to over 50 people, we were selling, developing, manufacturing, marketing surgical instruments and really cool technology.”
Then Hooven spun out another business, AtriCure, in 2000. “AtriCure is currently over a billion dollar company and trades on Nasdaq, and it’s the leading company in the world in the surgical treatment of atrial fibrillation,” Hooven told Med Ad News.
Hooven started Enable Injections in 2010, and worked part time for that company. He was also working with CincyTech, a state-supported venture capital fund for high-potential technology companies in Southwest Ohio. “They hooked me up with people at Children’s Hospital, a doctor [there] had developed a concept, a device for painless children’s vaccinations,” he says. “Certainly, you could see the need for that. And I said I really like Children’s, I really like this doctor, I think there’s something here. Maybe not in this specific device, but something about body-worn devices and making it simple and comfortable for people to inject stuff.”
Unfortunately, the technology Enable Injections had developed for painless vaccinations did not work. “So we were kind of stuck, we figured we had better develop another type of technology,” Hooven says. “And we had gained a tremendous amount of knowledge and experience on injection and injection pain, so we said let’s really look at this market, let’s see what kind of technologies we can develop to address the market of injectable products.”
Hooven had talked with a number of different pharma executives and gone to a number of conferences. He also worked with his VP of R&D, Matt Huddleston, and a lead engineer looking at a variety of different areas, and decided that viscous biologics seemed to have a lot of promise from a market standpoint.
“If we could figure out how to easily and comfortably have people administering high-volume biologic drugs at home instead of having to go into the hospital or a clinic and get an I.V. or have a port installed – all of these tremendous inconveniences and costs that are incurred – if we could eliminate those, there’d be a big market,” according to Hooven.
“So I talked with some pharma executives and I got that input from pharma folks and conferences and we decided to develop something that addresses this market.”
The device finally developed by Enable is the size of half of a yo-yo and can come prefilled, or with a transfer station that allows the patient to use a syringe to put the drug into the device, or insert a vial of the drug. “You push the vial in and the transfer system automatically delivers all the drug in the vial to the injector,” Hooven says. “And it takes perhaps 30 seconds to a minute, but what it does while it’s doing that, it also warms the drug up. Almost all of these drugs have to be refrigerated. That’s a big advantage because a cold drug is very uncomfortable to inject.”
Hooven says one of the main things Enable Injections heard from pharma companies about the system being developed was “don’t make us change the container that we’re putting the drug in,” Hooven says.
The easiest way to address that concern was that pharma companies would continue to fill the drug into their containers on the manufacturing line and then when the drug is delivered to the patient, the patient transfers the drug into the injector.
The transfer is “a really simple, foolproof, quick way to do it,” and patients can’t overdose or underdose themselves, or introduce any bubbles.
The first company to sign up with Enable Injections was CSL Behring, the top manufacturer in the world of immunoglobulin therapy.
Additionally, Enable has commercial contracts with two other companies, and four commercial agreements where the company will be commercializing products. “And we are in the process of executing another three agreements, I would say we’re in the process of negotiating multiple agreements,” Hooven told Med Ad News.
The company had not only met with pharma leaders when developing its product, but with patient groups as well.
“It’s very moving when they see the product and you show them that they don’t have to go into the hospital every Saturday morning for the rest of their lives, and they don’t have to sit there tethered to a syringe pump once a week for a couple of hours,” Hooven says. “They can just put this product on and go about their normal, daily activities. It can get very emotional, people have cried, people have said, ‘I get my life back.’
“I think it’s going to have a revolutionary effect on the way that chronic, high-volume biologics therapies are delivered. If you were a family member and you had the choice of going to a hospital or a clinic to get an I.V. infusion for a few hours once a week, or just taking a vial out of your refrigerator, putting it into a transfer device, putting the injector on your abdomen and pressing a button, what would you rather do?”
Enable Injections’ product has a compelling story from both the patient benefit perspective and the pharma partners’ perspective, Hooven says.
“Many companies are targeting the same target receptor or molecule, so what’s happening is they’re looking for ways to differentiate themselves, and what better way than to say to a patient, ‘Hey, my competitor is going to make you go into the hospital, I can offer you something where you don’t have to do that, you can stay at home,’” Hooven told Med Ad News. “And the pharma companies aren’t charging any more for the products, so that’s really a competitive differentiation.”
From a healthcare cost savings standpoint, the injector can remove “massive costs by significantly reducing the need for these intravenous infusions,” Hooven says.
Enable Injections stands at more than 70 employees now and will be growing very rapidly in 2018, according to Hooven.
“We are on the verge of signing multiple commercial agreements with multiple major pharma partners,” according to Hooven. “And we’re going to be in two different clinical studies this year in humans with three products, and a fourth product, possibly a fifth product, in a third and fourth clinical study next year.”
The company expects U.S. approval for an Enable Injections-based drug and device product to occur in the 2021 timeframe, with marketing approvals outside of the United States expected during 2020.