An average cost of developing a new medication is currently at about $2.6 billion, according to the Pharmaceutical Research and Manufacturers of America (PhRMA). Medical research requires huge spending, but nearly half of all funds invested by the companies in R&D, are spent on medicines which will never make it to the market. Now, new innovative technologies come to help market players (pharmaceutical companies and funds) by lowering investment risks. That may be accomplished through automated analysis of big data pertaining to the pharmaceuticals industry, such as patents, clinical tests descriptions, scientific reports and materials, information on the medicines in the process of development, and even media releases and professional conferences posters.
Best set goal
That is exactly the service offered to the pharmaceutical industry by a Russian startup company, Semantic Hub, which operates in the market since 2015. By employing the service, one can now assess a medication development prospects, thus decreasing investment risks and choosing the right ‘investment candidates’. Research developments’ valuation will require much less time and, most importantly, reduce a chance of making a wrong decision through simply missing essential information, down to 30%.
“It is about dealing with a text, understanding its essence, and using huge volumes of unstructured texts to extract all sorts of valuable information to then form a database”, Irina Efimenko, CEO at Semantic Hub, says.
The company now offers in the market two intellectual products. One of them is more about marketing tasks which within a traditional approach to the problem are dealt with by medical directors. It is identifying and analyzing problems and needs of doctors and patients. The service, for example, can help set limits and drivers for prescribing medications, understand why a patient follows the prescribed therapy or, alternatively, abandons it. For that, analysis of user content (internet forums, social media, etc.) is performed.
Semantic Hub’s other product directly relates to R&D. It is a search of promising pharmaceutical developments (be that an innovative molecule or a product which is well known in the market but has got new properties). A task may also be set in a different manner, as the instrument allows to assess the potential investment goals already identified by a customer.
As Irina Efimenko notes, “A company may be reviewing some specific development and considering if it is worth being financed”.
In the pharmaceutical industry such a situation is most common. At the earlier stages developments are performed by technology startups or university labs, whereas the clinical tests process can only be undertaken by major players, primarily due to high costs of such tests.
It is certainly about analysis of the already existing data. But to some degree the company also moderates the activities of technology scouts who are employed by all large pharmaceutical companies and, among other things, attend conferences attempting to locate interesting and potentially breakthrough developments.
“What we do is about the same thing”, Irina Efimenko comments, “But our monitoring is focused on the web”.
Supporting toils
“Our business is about decision-making support”, Semantic Hub’s Efimenko continues. “We relieve an expert of that grunt work, thus allowing to concentrate on more intellectually intensive elements. True, respective professionals have been managing their businesses without our involvement – just like without any other automated services. Still, it is the pharmaceutical industry where the issue of big data analysis is the most acute and where a cost of every mistake is exorbitantly high. Therefore, in this sector automation is most useful”.
In fact, it is about evaluating medications’ potential and identifying promising developments through the Artificial Intelligence and Big Data Analysis technologies. One of the system elements is a web robot, or crawler, which plugs to authorized data, patents, articles, and clinical tests results bases, to collect respective information. It can also logon to forums and portals on medical matters, depending on the tasks set. Even though the information collection is broadest possible, it only takes few hours to process it.
Once the collected data array is refined (by deleting banners and links, for instance), it is then analyzed by a linguistic processor. “It is our major asset and technology core”, Irina Efimenko explains. “It is a system capable to understand the meaning of a text”.
The mechanism of its functioning is somewhat reminiscent of associative human thinking. If, for instance, you see in a text a word starting with a capital letter and followed by Inc. (incorporated), you will most likely understand that it is a name of a company, as a certain learnt pattern will help. Semantic Hub’s linguistic processor is functioning in approximately the same manner. From a scientific article on a new formation it will extract the name of the molecule, the problems it can cure (say, rheumatoid arthritis), the models it has been tested at, etc. For each specific area like oncology, antibiotics or others there may be extracted hundreds of items, factors, characteristics.
Since information from various sources may be contradictory, the service also allows evaluating reliability of a source and testing information significance.
Overseas’ requests
Along with Irina Efimenko, other founders of the project are Vitaly Nedelsky, Russian Association of Robotics’ President and Semantic Hub’ BD, and Vladimir Khoroshevsky, Chief Scientist, a prominent researcher of Artificial Intelligence. The Semantic Hub project’s investors include Internet Initiatives Development Fund (where the startup underwent an acceleration program) and entrepreneurs Valentin Doronichev, cofounder of Invitro private medical clinics chain and of Medme investment fund, and Vladimir Preobrazhensky, coinvestor of Multikubik startup and former CFO at Wimm Bill Dann food products manufacturer and SUEK coal producer. Last summer, they all invested in the project RUR 24 mio ($430 K). Prior to that, RUR 2.1 mio ($36K) were invested by Internet Initiatives Development Fund in exchange for a 7% stake in the company.
SemanticHub has by now got orders from various customers including Russian pharmaceuticals market players. The latter are mainly interested in the ready-made marketing products, while foreign clients pay more attention to R&D.
“Pharmaceutical industry in Russia is regretfully insufficiently developed, especially as far as R&D are concerned, but we are already getting some world class contracts”, Irina Efimenko notes.
Spendings on an R&D project average some €100K and a bit less on a marketing project.
Semantic Hub’s strategic partner is Bayer, a pharmaceuticals giant which has supported the startup in a number of projects. Semantic Hub’s solutions were highly valued by the participants of the Global Pharma R&D Informatics Congress, which took place in Portugal in late 2017.
Aiming to expand the spectrum of its foreign customers, the company intends to actively advance to the global markets, including the Asian one. The platform now handles English and Russian and is experimenting with Chinese (the resource itself can be tuned to various languages).
Come on
At the beginning Semantic Hub had projects intending for various sectors, including orders from oil, gas, and mining companies. It started getting first projects in the pharmaceutical industry back in 2016. Jointly with Health.Mail.ru service the startup performed analysis of communications of doctors and patients. The ultimate choice of the industry was made in 2017, and it appears the choice was right.
“Pharmaceuticals manufacturers start using AI technology at various stages of developing new medications, ranging from creating new molecules to clinical tests. The main benefit of its use is cutting development costs”, Sergei Sorokin, CEO at Interlogic, says.
“The industry has accumulated huge volumes of data, hence both analytics and predictive modeling based on such data may be useful to all market players”, Irina Milekhina, IT practices and telecommunications senior consultant at Odgers Berndston, explains.
The time has come for their smart use, she adds, including, among other things, for improving and accelerating medicines development, optimizing production processes and cutting the costs involved.
“For corporations, that is a matter of billions of dollars which they can save or additionally make. Besides, that is about billions of human lives as well, about people who can get timely and qualified medical assistance”, Irina Milekhina believes.
“In the pharmaceutical industry the data always had to be properly processed and interpreted. Still, as years go by, advancing technologies increase processing efficiency and the data volumes that can be processed”, Alexei Sidorin, business solutions architect at CROC business integrator, notes. In his view, the main scenarios for processing big data in this sphere will be clinical tests results collection and analysis (pharma datalake), omex technologies, analytical systems combining publication and scientific research in the pharmaceutical industry, as well as the supply and production chains intellectual monopolization.
“Currently, all of these scenarios are in the initial development stage, but do have a great potential”, – Alexei Sidorin says. He also believes that in some while use of the big data in the pharmaceutical industry and in testing medicines specifically, will become obligatory. Evidently, technological requirements to pharmaceuticals manufacturers will only get stricter. The companies establish technology units focused on developing products for internal use. They also introduce new positions of big data business partners. Such a trend is certainly to the advantage of innovative startups.
By Olga Blinova