Over the last few decades, data generation has really exploded. However, the “Big Data paradigm” is not so much concerned with the volume of that data, but with how companies and, indeed, industries can derive meaningful insights from what has become an information overload.
With the currently popular approach to Artificial Intelligence (AI) centered around the Big Data paradigm, pharmaphorum also spoke with Adityo Prakash, CEO of Verseon, about the whys and wherefores, digging deeper into the processes to manage the current mountain of data and how it can be generated, as well as the purposes for which it can be approached constructively and efficiently.
Data generation and chess
“The underlying assumption is that there is a huge amount of data available to teach an AI program how to handle the problem at hand,” Prakash began. However, she explained, “the number of examples known to train AI is at least many thousands of times greater than the number of variables or characteristics to track.”
“These training examples can be real-world data or data generated synthetically by computer software,” he continued. “AI chatbots use real-world examples: the text of hundreds of millions of web pages […] Of course, for every real-world problem that has large data sets available, there are dozens of others that have only limited training data.
This mention of the “real world” is, of course, a central concern at the moment within the pharmaceutical industry, ongoing conversations seeking to ensure that the data collected is useful because it is so, i.e., real and not ignorant of the variables of the patient’s daily life.
Comparing AlphaZero, the chess program developed by Google’s DeepMind, to the training examples mentioned, Prakash said that, in contrast, AlphaZero “is based on self-generated synthetic data. The system plays millions of games against another instance of itself, generating sample data from each game.”
A slow journey of drug discovery
When asked about the momentum behind Verseon’s work in the industry, Prakash explained that he founded the company “to change the way the world finds new drugs.” Quite an ambitious goal.
“Today’s pharmaceutical drug discovery process is essentially trial and error,” he explained. “To find potential drug candidates, disease-associated protein targets are tested against a small pool of fewer than ten million distinct chemotypes synthesized at great effort and expense over the past 100 years. This slow and erratic process fundamentally limits our ability to explore large numbers of new drug-like molecules in search of the new treatments we need,” she said.
It might seem like the perfect entry point for AI, but there are caveats to the essential purpose of jumping on the bandwagon.
“Industry now hopes that providing the same limited experiential data to AI could help modify existing drug-like compounds slightly faster than before. Unfortunately, using AI in this way doesn’t help find new drugs for all the diseases we can’t cure today.”
Overcoming the bottleneck, at the molecular level
How, then, can such a limiting problem be circumvented? For Prakash, the key is physics-based molecular modeling.
“At Verseon, we have developed significant advances in multiple distinct areas of science over the past two decades to overcome [….] bottlenecks in current drug discovery and development,” he said. “On the computer, we design billions of new drug-like molecules that have never been produced before. Using physics-based molecular modeling, we determine whether new molecules will bind to a target protein without having to create them first in the laboratory.”
The process doesn’t end there, though.
“We then select the most promising candidates, synthesize them in our chemistry lab, advance them through extensive biochemical testing, and optimize them using artificial intelligence. Through this process, we systematically generate more clinical candidates with unique therapeutic profiles,” Prakash explained. “As more candidates advance in the clinic, our adaptive AI for clinical trials helps segment patient populations and personalize therapies.” .
From Big Data to small ones
Back in 2020 – this is a fast-moving industry – computer scientist and tech entrepreneur, as well as co-founder and head of Google Brain, Andrew Ng was talking to Forbes about big data vs small data issues, stating that “The number of teams that can build good AI systems from big data […] it’s in the hundreds, but a much smaller number can build good AI with little data.”
So it was that, having Verseon acquired Edammo for the purpose of solving the “small data problem”, pharmaphorum asked Prakash to provide further details on what his company hoped to achieve with such a move.
“Most AI systems require huge amounts of dense data to make accurate predictions,” he said. “But in fields like life sciences, especially drug development and clinical trials, the amount of data is small and sparsely distributed, compared to the number of variables or characteristics an AI model needs to track.”
This is where problems arise.
“In these contexts, traditional big data AI systems struggle to produce accurate results, if they can even build a predictive model from the limited data available,” Prakash continued. “Verseon has continued to develop its own specialized AI tools internally to handle these situations [but] has also kept a close eye on external developments and has found that Edammo’s Extreme AutoML technology performs particularly well in a variety of life science tasks.”
In short, what Edammo brings to Verseon’s goals is efficiency:
“As we gather data on our new compounds through laboratory tests and clinical trials, Edammo’s technology will help generate insights from the data more efficiently. These insights will help us bring to market different treatment alternatives for every disease we face and offer options to patients who currently lack any,” Prakash explained. disease treatments up to a certain point [yet] possible today”.
In contrast to the trend of composed bookcases and to the future
When asked if such technology had been used in the company’s product pipeline, Prakash’s response was overwhelmingly yes.
“To date, all candidates in Verseon’s pipeline have been developed using ours […] platform. Unlike other current companies promoting so-called ‘AI-first’ or ‘AI-only’ approaches,” Prakash said, “we are not dependent on university labs or industry partners to find candidates for us. The drug candidates we find are very different from the compound libraries used by the rest of our industry.”
And as for future AI drug discovery, it looks like the medical horizon will become more and more intertwined with this technology, at least for Verseon.
“[Using the platform,] We expect our pipeline growth to accelerate in the coming years as we address an ever-increasing number of health conditions that impact our quality of life. We’re not only changing how efficiently drugs can be discovered, we’re also changing what people can expect from 21st century medicine.”
About the interviewee
Adityo Prakash is the Chief Executive Officer of Verseon. He enjoys building fundamental scientific solutions to major business problems that impact society. Prakash led the development of Verseon’s drug discovery platform, new drug pipeline, and overall corporate strategy. Previously, he was the CEO of Pulsent Corporation. Prakash received his bachelor’s degree in physics and mathematics from Caltech.