Insilico Medicine Becomes the Face of AI Drug Discovery
In many industries, there’s usually one or two companies that become nearly synonymous with their particular market. One of the most obvious examples is Coca Cola (KO). Think about it: We still refer to a soda as a coke, even though the soft drink manufacturer
allegedly abandoned using cocaine in its formulation long ago. In emerging technologies like artificial intelligence, it’s more difficult to pinpoint a clear leader in many cases. Nvidia (NVDA) is still the obvious choice for AI chips. After that, there are very few household names that represent anything close to a pure play in AI technology. However, one name has emerged in the last few years that has become the face of AI drug discovery: Insilico Medicine.
Regular readers will certainly recognize the startup as a key player in the longevity industry. We’ve profiled the company on several lists here and here related to drug discovery, as part of our ongoing coverage of life extension science. September was a particularly good month for the Rockville, Maryland startup. At the beginning of the month, it published a paper in Nature Biotechnology that detailed its efforts to design and validate a drug candidate for treating fibrosis and other diseases in just 46 days, shaving off months if not years from the discovery process, not to mention saving millions of dollars. A week later, the five-year-old startup completed a $37 million Series B, bringing its total disclosed funding to $51.3 million.
We recently caught up with Insilico Medicine co-founder and CEO Alex Zhavoronkov, who is on a crusade to prove that automation will be a big part of the future of healthcare. We talked about the company’s recent achievement in accelerating the drug discovery process; its open challenge to big pharma; Insilico’s numerous partnerships and joint ventures; the difficulties in getting funding for longevity therapy research and development; and the AI hype surrounding healthcare in general.
Putting AI in the Spotlight
We won’t spend too much time dissecting Insilico’s latest achievement, as that’s already gotten plenty of press. You can find a good deep dive into the paper and its significance by Margaretta Colangelo, a managing partner at Deep Knowledge Ventures, which made some of the first investments in the company about five years ago. It wasn’t just the fact that Insilico Medicine’s new AI platform, called Generative Tensorial Reinforcement Learning (GENTRL), accomplished the feat of designing and validating a drug in such a short timeframe. It was also the first time anyone had combined two AI techniques known as generative adversarial networks (GANs) and generative reinforcement learning for drug discovery. Both Colangelo and Zhavoronkov refer to the achievement as pharma’s AlphaGo moment, referring to Google DeepMind’s defeat of a professional Go player.
Zhavoronkov first presented the paper at a conference in Basal, Switzerland, hometown to two of big pharma’s biggest players – Roche and Novartis. The choice of venue was obviously intentional, as Insilico has actively sought the spotlight since it was founded in 2014 in order to prove the value of AI to improve human health and quality of life. In 2015, for example, a group of Zhavoronkov’s students and colleagues founded, Youth Laboratories, listing him as an adviser. It was a machine vision company that focused on aging and skin health which made headlines in 2016 for hosting an online beauty contest, Beauty.AI, that was judged solely by machines – and attracted a bit of controversy for picking mostly fair-skinned people. That led to yet another project, Diversity.AI, an effort to use machine learning to make sure we all have an equal opportunity to be
spied on marketed to. Last year, the company was named to CB Insights prestigious AI 100 list.
A Moonshot Challenge
But back to Insilico’s primary focus: defeating age-related disease. Zhavoronkov believes his company’s AI platform has advanced to the point where it could automate drug pipeline development from end to end and produce a marketable drug within 24 months or less – with the right kind of backing. So while its efforts have focused on pre-clinical activities like identifying drug targets and developing therapeutic molecules around those specific disease targets, Insilico’s AI also boasts predictive powers.
“We also work with some of our pharma partners on predicting clinical trial outcomes; we also analyze clinical trials data,” said Zhavoronkov, who is looking to partner with a major pharmaceutical company on an XPRIZE-type challenge where Insilico races against the clock to develop the first drug fully developed using artificial intelligence – and one that targets a rare disease. If no contenders emerge, Zhavoronkov said he still believes it will be possible to develop a viable drug using AI within four or five years.
He noted that investment banks are also interested in Insilico’s AI predictive analytics around clinical outcomes. The reason for the attention is obvious: If investors can leverage a tool that can better predict the chance of success or failure of a particular drug, that would surely influence how much money – if any – they are willing to put into a particular project or company. We’ve noted previously that venture capitalists are increasingly turning toward AI to guide their investments into the hottest startups.
While Insilico awaits to see what big pharma company will emerge to take up its challenge, it has plenty of other partnerships and ventures to juggle. In fact, the company just announced today a new collaboration worth up to $200 million with one of China’s biggest pharmaceutical companies, Jiangsu Chia Tai Fenghai Pharmaceutical Co. The goal of the collaboration is to accelerate drug discovery for triple-negative breast cancer using artificial intelligence.
“We actually started making those kind of advanced partnerships where we would plan to take products into [clinical trials] only recently,” Zhavoronkov said, for about the last year or so.
He noted that his company has about 16 ongoing collaborations, including several with one of the key players in the longevity industry – Juvenescence. One of the more advanced joint ventures with Juvenescence is with a company called Generait Pharmaceuticals that is targeting senescent cells, which are cells that have stopped dividing but continue to secrete inflammatory molecules that damage nearby cells and tissues, leading to disease and an earlier death. Generait has already identified several disease targets but Zhavoronkov could not go into details at this time.
Another joint venture between Juvenescence and Insilico, which also includes the nonprofit Buck Institute for Research on Aging, is Napa Therapeutics. Insilico has already “done its part” and Napa is now working on several potential drug molecules that could help boost the levels of an oxidized compound called nicotinamide adenine dinucleotide (NAD+), which is involved in getting the power plant of the cell called the mitochondria revved up again. “So far, the molecules look very good,” Zhavoronkov said.
“Many [partnerships] are with smaller companies, you know, where we do get less money up front, but we do get the data,” he noted. “We need their pre-clinical data so we can train [our algorithms].”
Raising Money in the Age of AI Hype
While Zhavoronkov is obviously pleased to have just pocketed $37 million from investors like Chinese AI tech giant Baidu and a pharmaceutical firm like Eli Lilly and Company, he feels that respect for the longevity industry is still lacking. For example, few of Insilico’s major investors backed the company because they are specifically interested in life extension science, according to Zhavoronkov.
“They invested not because of longevity. I actually need to divest of some of my longevity-focused programs,” he explained. “They invested because of the generative chemistry. It works; they know it works. That’s a major disruption. So that’s why they invested. They didn’t consider longevity; longevity is not being perceived as credible in financial circles yet.”
Legitimacy is also difficult to build when there is so much hype around artificial intelligence.
“There are lots and lots of scientific charlatans very often re-positioning very old technology as AI,” he said. “Suddenly, every statistician within big pharma became an AI scientist.”
In terms of direct competitors, Zhavoronkov noted that he deeply respects the work being done by San Francisco-based Atomwise, but the technology from others is mostly “smoke and mirrors.” That’s why Insilico is focused on publishing its work in peer-reviewed journals in order to back up its claims. The company has published about 60 papers in the last five years.
That’s because Zhavoronkov believes “that there is nothing more important than longevity,” whether you’re talking about the big-picture economics of healthcare or living more productive, healthier years in the twilight of life.
“The goal is to create the longevity economy,” he said. “So, in the next couple years, as we ingest a little bit more data and develop more advanced algorithms, we will be able to go after more complex problems.”
If the longevity industry does live up to its lofty goals of becoming the biggest industry in human history, Insilico Medicine will likely play a major role in that success. Even if we don’t all end up living as old as Moses, the potential to cure some of humanity’s most debilitating diseases is too good to ignore. And, right now, no one is ignoring Insilico Medicine.