There is that old adage of “the future belongs to the youth” – and technically that is true – but the statement also carries a risk of underestimating where we come from.
Vladimir Khoroshevskiy – one of the three founders of Semantic Hub – is a Professor in AI (computer sciences). He was 69 when he started the company. Vladimir is also the founding member of the Russian Association for Artificial Intelligence, and he’s been writing about artificial intelligence since the 1970’s.
To put that into a pop culture context, Blade Runner – a film about intelligent androids hiding amid the human population - was released in 1982.
Irina Efimenko, co-founder and CEO, has 20 years experience working with computational linguistics, which is the science of processing natural language with computers. “We were working with AI before it became sexy!” she quips.
Their current AI – the core product of Semantic Hub – is a powerful multi-lingual natural language processor that scours the public web, trying to find people who are looking for help with severe and unexplained medical conditions. Its ultimate goal – to make sure the people get diagnosed and hopefully treated.
It’s a pretty wild idea – just like most 80’s sci-fi movies.
Semantic Hub joined the Salto Growth Camp accelerator to fine tune their business model – having already 50+ contracts from top 20 pharmaceutical companies in their pocket, they’ve set their sight on scaling up on a global level, recently opening a new HQ in Switzerland.
“We were working with AI before it became sexy!”
Vladimir and Irina met while working at a computational linguistics company in Moscow. Vladimir is from Odessa, while Irina was born in Obninsk. It’s an otherwise unremarkable town not far from Moscow, if not for the fact that in 1954 it housed the world’s first grid-connected nuclear reactor.
It was also in Moscow where the two started working on their dream of using their vast knowledge on AI on something more practical – and something that was their own.
Irina and Vladimir were working for a company providing semantic analysis before starting Semantic Hub. But back in 2000, it was mostly used for media monitoring – effectively gauging the perception of public figures. It was pretty simplistic, and boring.
Irina and Vladimir’s ambitions were greater. Not content with oversimplified applications, they sought for an idea that had a solid business application, but also made better use of the science. In 2015, they met their third cofounder, Vitaly Nedelskiy, and started Semantic Hub together.
Initially, they used their natural language processor as a tool for spotting emerging trends and technologies for tech companies. The idea, simply put, was to have an AI that could analyse thousands of documents and highlight “the next big thing” before humans caught wind of it.
But that application was far too broad and they soon ran into a problem that is, sadly rather common – everybody wants Big Data, but no one knows what to do with it.
Irina says they quickly realised that in order to have success with their AI, they would first need to find a very niche focus area and secondly, they would need to be proactive about offering very concrete value, not just raw data. They looked toward the pharmaceutical industry.
“I always wanted to know how the brain works,” says Vladimir when I ask him where his interest in the medical field stems from.
The pharmaceutical industry is a huge market, but it’s also one that is facing a rather unique challenge when it comes to the research and development of treatments for rare diseases.
Such treatments are called “orphan drugs”. It’s a grim term that reflects a grim reality – some diseases are simply too rare to get affordable treatment without government sponsorship. Pharmaceutical companies need to know where these patients are, but since it takes on average 7 years to get a diagnosis, most of these people aren’t on anybody’s radar.
But these people are out there – 475 million of them, in fact.
Even without a diagnosis, these people are sharing stories about their mysterious conditions online. Vladimir explains that people with strange symptoms want to find others like them - and that’s where the AI comes in.
Semantic Hub translates information from “ordinary human language” into clinical language
“Just imagine a real situation, where a mother has a child who is experiencing some really strange symptoms.” Irina says. She pauses, looking for a word to emphasise what’s at stake: “really ugly symptoms.”
Irina runs through a common story of people with rare conditions – they go from one doctor to another, one diagnosis to the other. These tend to be ordinary, because even doctors aren’t normally looking for freak illnesses, the one in a million kind. Sooner or later, people turn to the internet in their despair.
Normally, lengthy posts about strange symptoms on some family messaging board would fly under the radar of medical professionals - but to the never-sleeping eye of an A.I., they stand out.
“Sometimes, something which is described by a person with no knowledge of medicine can be a sign of a very specific clinical symptom,” Irina explains.
Semantic Hub translates that information from “ordinary human language” into clinical language, and then packages that information so that pharmaceutical researchers can see how many potential patients exist in a country, and which medical centres they could reach out to. Semantic Hub also helps patients to find doctors with knowledge on rare diseases.
The data is not just useful for finding specific markets for orphan drugs. It also helps collect patient stories.
“Until recently, patients were treated as objects of care – they had no voice. There was a perception that they understood nothing. Nowadays, they are decision-makers – the voice of the patient is becoming more important.”
People with severe diseases know more about their illness than doctors do. “People become experts, ” Irina says. “They know their disease better than an ordinary doctor, you could say.”
“What is the success criteria for treatment in a patient who will die?”
Rich narratives, contained in online forums, hold a lot of information – not just about symptoms, but extra-medical details, for example about quality of life.
“What is the success criteria for treatment in a patient who will die?” Irina asks me. It’s a tough question, and one that knocks the wind out of you a bit.
There are still, unfortunately, diseases out there that win out in the end whatever drugs you throw at it. Horrific as it may sound – sometimes you have to know what is the definition of the best possible life for someone who doesn’t have a lot of it left.
The answer to that is not always evident - and patient stories are a unique source of information for such knowledge.
Aside from building a successful startup, both Vladimir and Irina are still active academics.
“In science, you can sometimes be alone – but in business, you are always part of a community,”
I ask them which is easier – getting a PhD, or building a startup? The latter, for Irina, has a lot more responsibility – for the team but also the clients. “In science, you can sometimes be alone – but in business, you are always part of a community,” Vladimir says.
But fundamentally, science and business are quite similar, Irina says. “You go in some unknown direction and you never know what’s going to happen.” She adds that in both cases, you have to create some value in the end. “Otherwise it’s bad science – or bad business.”
There’s another thing science and startups share: “you should be really dedicated. And you should be a little bit crazy - otherwise you better not start!”
I think we can all wholeheartedly agree with that.