A St. Petersburg entrepreneur has combined his experience in the financial technology industry with the power of artificial intelligence to increase the efficiency and success of biomedical research.
Speaking at Wednesday’s 1 Million Cups event, hosted by Thrive DTSP, the founder of local startup Pico Portal told a panel of area business leaders how his company is changing how the world conducts biomedical research.
Eitan Agai said he came up with the idea for Pico Portal in 2020, during the height of the pandemic.
“How can I help researchers around the world to do better research?” he wondered. “Being that I come from financial technology and work with the mortgage industry, I realized that many of the technologies we develop to eliminate pain points that the financial world is facing can be transported – and can be useful for researchers around the world.”
According to the Pico Portal website, the platform’s mission is to transform research findings into decision-ready evidence. The service utilizes machine learning and natural language processing algorithms, enabling researchers to systematically and easily review scientific literature.
Because he is not a doctor – despite his mother’s wishes – Agai said he put together an advisory board that features researchers from John Hopkins University, the University of Minnesota and the University of Colorado. He then launched Pico Portal in September 2020. “Really to have experts in the field of research to help me build the tool,” he said.
While Agai hopes to grow the platform’s community of users, he said prestigious clients such as the National Academy of Sciences, the Veteran’s Administration (VA) and the aforementioned universities utilize Pico Portal’s services.
In less than two years, Agai said researchers uploaded over a million scientific studies to the platform, which also offers a free version of its subscription model.
“From researchers that represent over 70 countries and about 180 different institutions,” he added. “So, it took off – we definitely hit some sort of a nerve.”
Agai explained the research process for doctors and scientists is similar to the average person looking to buy a car. Just as the potential car-buyer turns to Google and enters keywords such as the make, color, year and price of a vehicle, researchers comb through biomedical databases, like HealthMed or Google Scholars.
With databases consisting of 30-40 million articles, Agai said that searching for keywords still returns anywhere from hundreds to tens of thousands of published studies that one must sift through and read to find what is relevant.
According to the website, Pico Portal offers the fastest and most accurate deduplication technology, outperforming existing tools by 10-30%. Its machine learning capabilities adjust results according to a user’s previous behavior, allowing them to review the most relevant articles first.
“It learns what the scientist actually is looking for,” said Agai. “Then it starts to push all the relevant articles to the beginning of the process.”
Agai said the platform’s interface – supported by browsers such as Chrome, Safari and Firefox and available on mobile devices – was designed to make results easy to read and provide a quick view of important metrics. That allows researchers to quickly ascertain a study’s relevancy.
“They don’t need to understand the article,” he added. “Why did I design it like this? I’m personally a little bit dyslexic.”
The founder explained he used his “inefficiency” to help him design the user interface that is simple for anyone to read. He said if it is easy for him to read and understand, other people will find it effortless.
Pico Portal also offers an innovative and collaborative aspect due to its crowd engine, which can automatically assign an article to reviewers. He said five people could work together simultaneously while tracking each other’s progress.
Agai called Pico Portal a work of love and said he bootstrapped the startup. He told the panel that he invested “in the hundreds of thousands of dollars” of his own money to launch the service.
“I did it because during the pandemic, really the whole market was down,” he said. “People were not buying houses … and I’m in the mortgage industry on the technology side.”
Maintaining the product is also costly, and Agai noted hosting the platform on the cloud cost around $1,000 per month. He said about 50 people are working on Pico Portal, and “some of them even wanted to work as volunteers.”
He said the people involved with the platform believe in the concept and enjoy helping him build it, but everything is on a shoestring budget.
“Let’s say the market price for an AI person is $100 per hour – they get maybe 20%,” said Agai. “They love the idea.”
Agai said that Pico Portal generates income from a subscription model. Using Johns Hopkins University as an example, he said consumers can pay annually or monthly on a tiered basis according to how many users need access to the platform.
Pico Portal charges a flat rate for every 10 users. As opposed to other companies that Agai said “nickel and dime you,” his platform is “all you can eat.”
Agai said he wanted to create a lighthouse to guide a market segment and do it so well that doctors, researchers and Ph.D. candidates have no choice but to use his product if they want to save time.
Agai asked the panel to help spread awareness for Pico Portal. He established a user base in higher education but said those subscriptions typically only last for a year before coming before committees and undergoing budgetary considerations.
He realized that researchers working for pharmaceutical companies have a quick decision-making process and deeper pockets.
“They tried Pico Portal, and they said, ‘ok, it’s $5,000, not a problem,’” relayed Agai. “That is my focus right now – primarily to be able to make this product sustainable.”