Researchably, a young startup incubated at UC Berkeley, is conducting a pilot with Sanofi in China, using AI to sift through thousands of research studies and surface the most relevant ones to pharma stakeholders.

“With Sanofi we’re working with a small medical communications team, and they’re in charge of every day reviewing 100 to 200 articles on Type 2 diabetes to then inform business units who are working on competitive analysis and also external doctors who are essentially prescribing their drugs about what’s going on in their field,” COO Mauro Cozzi told MobiHealthNews. “This team of 10 was spending 30 hours a week reviewing all this information, and they were making a lot of mistakes because they were doing it very quickly, so sometimes they would miss a critical paper or they wouldn’t send it to the right person. And they were actually not reviewing a lot of information that could have been relevant to them because they didn’t have the bandwidth. So they were only looking at 200 papers even though every week there are around 2,000 papers about Type 2 diabetes.”

In a three-month pilot, Sanofi found that using the software cut down the time from an average of 13 minutes per paper to less than one second per paper.

Why it matters

CEO Maciej Szpakowski told MobiHealthNews that 30,000 new research papers are published every week, and this number is only going up. This technology to sift through and surface the most important papers is useful for pharmas, but could also be invaluable to doctors.

“Right now, the way the AI works is it finds the research papers that are relevant to a particular stakeholder within the pharmaceutical company. So it delivers directly the research papers that are most relevant to you with some augmented information,” Szpakowski said. “The way we envision it to work in three years time, once we get more data, is that we’ll be able to also summarize the actual research paper, so you don’t have to even look at the actual research paper content, you can just get the gist of it from the summary which is digestible and much easier to consume.”

The company is working with three other pharma companies and planning to expand its relationship with Sanofi into a global one. After that, the sky’s the limit.

“The implication of that is not only will the pharmaceutical companies will be benefiting directly, but also doctors can look at those papers as well and patients can be informed on the latest research as well, Szpakowski said. “So basically we want to make the research much more accessible for every single human being, and that’s the big vision that really excites us moving forward.”

What’s the trend

Managing the massive data stores inherent in published research has been a dream for AI in medicine for some time — in fact, it was one of the primary early claims of IBM Watson for Oncology. Cozzi says Watson’s product hasn’t impressed the customers he’s been speaking to.

“Watson developed some really amazing technology and they have super-talented researchers on their team,” he said. “But what we’ve heard from our clients at least was that what they failed at was being able to understand the problem itself well enough that they could deliver an actual solution. Their technology was very good but the understanding of the nuances of the domain were not there.”

Right now, most of this work is still done by human teams, either in-house or outsourced.