Roman Chernyshev: “We are waiting for a warning medicine instead of a responsive medicine”

We continue a series of interviews with leaders of industrial practices DataArt. Our editors Daniel Lurie and Vadim Mazin talked about medical technology with Roman Chernyshev, the leader of medical practice. We discussed medical wearables, smart prostheses, the conservativeness of the industry, Big Data in medicine and much more.

- Who is the main one on the medical tech market - startups or corporations?

- There is no simple answer to this question. Medicine, due to the inevitable conservatism, very slowly adapts to new technologies. Here is a huge potential, many new ideas and technologies that promise to turn everything upside down and make medicine completely different, but this update will take a very long time. All innovative ideas go through several long stages - from the birth in start-ups and academic institutions, where they run in, but do not have a big impact on the industry, to the realization of the potential by large players who already have the resources to change the landscape and transfer innovations to the mainstream.

I will give an example. A couple of years ago, there were a lot of startups that created telemedicine solutions - all sorts of platforms with which people could get medical advice remotely. This was an incredibly popular destination, but widespread adoption never happened. Now the situation is gradually changing - the industry begins to accept the rules of the game. Just a month ago, DataArt was contacted by one of the largest insurance companies in the world, so that we would develop just such a project - a solution for telemedicine. Here they, with their resources, with their penetration everywhere and with serious influence on the industry, have a chance to radically change all prevailing rules. And just a couple of years ago it was absolutely uninteresting to them.

I think in five years there will be big companies where startups are now based. But there will be new niches that will create and occupy new startups. It is this process that moves the industry forward.

- The accumulation of large amounts of information about patients in different parts of the globe - in particular, genetic or about the effects of drugs on the body - should theoretically allow systems that work with big data to find previously unknown dependencies and correlations. For example, the FoxP2 gene, which is responsible for language, was discovered in this way. Are there any other such stories? What to expect in this area in the near future? What role do wearable devices play here?

- Potentially, this is the most breakthrough field in medicine. Talking about personalized medicine is largely about that.

Everything about wearables is now the territory of startups, although big players are already looking there. Not so long ago we talked with a large prosthetics company. They are beginning to understand what potential wearable devices have in their business. The introduction of sensors in their products allows you to collect detailed statistics on how they are used, where they are used, and whether the patient follows the doctor's recommendations. This is irreplaceable information.

Wearables are capable of collecting a great deal of data. And, for example, correlating the data of a particular person with the historical data array, it will become possible to predict many deviations even before they appear. When it works to its full potential, a predictive medicine awaits us instead of the usual reactive medicine to an already obvious problem. Medicine has a chance to become literally a healthcare - as opposed to a cure, which describes the current state of the industry much more accurately.

Alas, all of these wonderful perspectives are being hindered by the conservativeness of the industry. To create solutions in the field of big data, you need to be able to collect these data, and to compare and analyze them, you need to have access to them. This is a very problematic space - everything regarding personal data is now very tightly regulated. To get a truly three-dimensional picture, you need to cross data from so many sources. Now, one company’s access to data that another has is difficult. Especially if one of the services is in America, the second in Europe, and the third, for example, in Australia. In this case, it is necessary to satisfy the requirements of the three regulators, which is now almost impossible to do.

- And what are the main risks? Privacy? Too much dependence of decisions on the data that systems receive in automatic mode - or even make decisions in automatic mode? External interference in such systems?

- The roots of industry conservatism are in the Medina ethics. The main rule of all medicine is “do no harm”, and medical technology is no exception. Each new technology carries a big risk, because it interacts with a living person. Until there is full confidence that the technology is not harmful in any of the possible scenarios, it will not be implemented.

The introduction of new technologies in medicine requires the same approaches as the introduction of new drugs. To enter the market, you need to go a long way: to make sure that the technology is not harmful, to conduct clinical research according to all the rules (there are many), to make sure that a particular technology in a particular case brings concrete benefits. It takes a lot of time and money. That is why startups do not steer the medical tech market, they do not have enough resources.

- And which markets are more open now? Perhaps such technologies will first appear somewhere in Japan or South Korea, and then the rest will catch up? Any concrete examples?

- We work mainly in the American and European markets - I have second-hand information about others. By indirect signs, you can see that emerging markets are the platform for many new solutions. Just because they are less regulated and more relaxed. Very often, for the introduction of innovative, breakthrough solutions, large companies choose, say, India or Brazil.

In fact, in developed markets, we observed a similar situation about ten years ago with cloud services. Then everyone began to insist that cloud solutions are the future, that services need to be transferred to the clouds. But security inevitably arose: it was difficult to determine who was responsible for the safety of the data. Nevertheless, over the past five years, the market has evolved significantly, the legal subtleties have settled down, the infrastructure has settled down, and this has ceased to be a problem.

Here the same thing will happen. The market must mature and realize everything, legal mechanisms should appear that will regulate technological processes.

- Back to wearable devices. How much medical wearables - not toy fitness bracelets, but real devices like cardiographs-holters, mobile sensors of epileptic seizures, glucometers - have already come to life? What to expect on this front?

- Glucometers, thermometers and other similar devices are also, by and large, toys. If you look closely at what is happening at industrial meetings (who brings what, what they are talking about), you will notice that this area is in its infancy. Most startups are now focusing on the production of devices whose potential is not used - like the same fitness trackers. Many can produce such devices, and only a few can build a serious infrastructure for their use. Almost no one, to be honest. It is here that the main potential for development is now. More serious possibilities of using wearable technologies appear - below I will talk about prostheses.

There is another problem. Wearable devices will remain toys until doctors can accept the data received from them as objective. Now this is basically impossible. You can create any cool wearable cardiograph that collects data for a long time. But, having come to the doctor, your user will hear something like this: "This is all wonderful, but now go and make a normal cardiogram." Just because in the present conditions he does not have the right to accept such data.

- Now there are devices that read brain signals responsible for basic actions. In the entertainment field, this is already being used somehow - you can control the ball with the power of thought and all that. Is something similar used in medicine?

- I recall, perhaps, a variety of devices for the disabled, helping paralyzed people to use computers. This works by analyzing brain signals or eye tracking. Such a thing helps Stephen Hawking to write books and speak to the public. In general, it is clear that this direction is developing, but this is not a particularly large market, so there is no base here, like consumer devices.

- And what real problems are now being solved in the medical tech? What problems do clients come to DataArt with?

- They come mainly for technologies for which there is already a place in the market. This, for example, electronic medical records or medical business services: practice management systems, systems for laboratories - we worked on this with Charles River Laboratories .

We are quite active in the field of Life Sciences. These are pharmaceutical and clinical studies and similar interesting things. Next year we set this area for ourselves as one of the priorities. There is a huge space for the introduction of technologies, and many of our experiments in this area are quite in demand. This is more likely not about global changes in the industry or the invention of new drugs, although we also have such projects. The main work is related to the automation of existing processes that were previously carried out without the use of new technologies.

- It looks like building a foundation for further development.

- To a large extent, it is. We are preparing the foundation, automating processes that previously happened manually. This is a necessary requirement in order to further bring here Big Data, data streams with wearables and much more.

- By the way, about the foundation. How do you think ResearchKit is promising in the new iOS, as an attempt to create a large platform for collecting medical data about a person? Did they manage to do something significant?

- This is not a new idea, actually. At different times (and quite a long time ago) many companies were engaged in this. There were Microsoft with their HealthVault, which still exists. This is a centralized base where data on health-related indicators should have flocked. There was Google with their now closed Google Health. They didn’t invent anything fundamentally new; they simply connected it to their mobile platform.

How good is it? Definitely good. Sooner or later, such platforms will be in demand. To analyze data and make predictions, data must be collected. It will be difficult to say whether this will be with Apple or someone else. But experiments in this area will not stop for a very long time. Exactly until a stable design is found.

- Now Google is closely engaged in medicine with their Life Sciences (now Verily ). But they seem to be interested not so much in the consumer market as in serious fundamental research.

- Google is good at delivering revolutionary technologies, and it’s not at all necessary that they commercialize them. Things are coming out of Google’s labs that are changing the direction of the industry in one way or another. Now, let's say, they have released their deep learning algorithms, laid out their code for free access.

With Google Health, they tried to do a somewhat atypical thing for them - a widespread mass product. There was no fundamental know-how, no academic science at the core. All these platforms appear because the one who will ultimately control this data will earn a lot of money. Population analysis will be possible thanks to such platforms, and this is the main condition for the transition to the medicine of predictions, which we have already talked about. Attempts will continue until a solution is found.

Now Google has gone the other way. They are trying to do what they do well - revolutionary breakthrough technologies.

- I recently heard about an interesting project at the intersection of pharmacology and text analysis. The system turns to PubMed and analyzes in detail the latest publications, collecting information on where everything is heading. At the junction of medicine and text analysis, have you heard about any interesting projects?

- Yes, there were interesting projects. There was a project where this was used for a deep analysis of academic publications over many years and comparing them with the studied data in order to pre-screen non-viable options.

Speech analysis has great potential wherever, one way or another, there is technology, especially in medicine. Most of the time, the doctor’s hands and head should be busy with business, not computer work: a computer is simply an inevitable evil. Throughout the civilized world, people are trying to get as far away from this practice as possible. There are even services that make a lot of money by decrypting doctors' voice notes, but so far this has not been done automatically. To remove the intermediate link in the person of a decryptor is a very noble task and very commercially promising, since the time of a doctor is very much appreciated. Here, speech analysis will certainly find its application.

Voice control is the interface of the future. All these beautiful screens with buttons and even touch interfaces will die. At that moment, when a person can enter into a speech dialogue with the machine, the user interface in its traditional sense will come to an end.

- And of the things that have already become reality, which projects have most struck you in the last couple of years?

- One mentioned - it's dentures with sensors. This was the first truly understandable and completely useful application of wearable devices in medicine, and not in the fitness / wellness area, where most players graze now. For me, this was the moment when I realized that wearable devices have real applications outside the entertainment market.

The second case is directly related to DataArt. We are helping a large company shorten the testing cycle for new chemical compounds by replacing tests in a wet laboratory (the one where the tubes are) with digital tests. By running a large amount of data about compounds through the system and comparing them with the large amount of previously obtained data, we can predict which compounds will be known to be inoperative. The system eliminates a huge number of initially non-viable options that otherwise would have to be tested live. And only the screened compounds are sent to the laboratory, where they undergo further tests. This greatly reduces the cost of the production cycle of substances.

- What awaits us in the medical laboratory in 2016?

- It seems to me that next year telemedicine will really begin to progress. Small companies already have some success in this area, and big players are starting to come here. A couple of years ago, it would be difficult to predict that insurance companies will be the main players in this market. But now it seems already obvious, since it drastically reduces their costs. I don’t know if 2016 will be a turning point, because such transformations are far from one year. But the fact that this will be a very big trend next year, and possibly further, in my opinion, is absolutely certain.