Alex Zhavoronkov, PhD, knows a thing or two about AI. That’s because Zhavoronkov is the chairman and CEO of Insilico Medicine, a biotech company that aims to transform drug discovery and development using generative artificial intelligence.
Using its proprietary platform, Insilico discovers novel disease targets, generates new molecules, and then predicts the success of clinical trials. The company targets diseases including idiopathic pulmonary fibrosis, cancer, COVID-19 and more. Six drug candidates are in clinical stages of development, according to the company’s website, and more than 40 pharmaceutical companies use Insilico's technology to advance their programs.
Zhavoronkov spoke at Cure about the many ways AI can revolutionize healthcare. This conversation has been edited for length and clarity.
AI and Aging
Many people have characterized AI as the greatest threat to humanity. “There is valid concern about AI running amuck,” said Zhavoronkov. But he also sees AI as one of our greatest hopes. He and his team at Insilico are working to unleash the potential of generative AI to tackle some of the world’s most challenging health problems, including aging.
Diseases related to aging are killing millions of Americans each year. Alzheimer’s kills more people than breast cancer and prostate cancer combined, according to the Alzheimer’s Association.
Plus, aging research is the gateway to understanding an enormous array of human diseases, Zhavornkov said. “To unleash the maximum potential of generative AI, you have to focus on aging. It’s one single feature that is very predictive of when you die and also when you are going to get certain diseases.”
Over time, environmental exposures as well as physiological and psychological stressors can accumulate to create dramatic systemic effects. “Only with AI can you really understand the fundamental biology,” said Zhavornkov. He said that it is possible to train deep neural networks to predict how some biological data types, like gene expression, change with time.
Deep neural networks are also very good at recognizing patterns once they are trained to predict a single feature. The power of AI can be unleashed when it can learn on one problem, like aging, and then transfer the learning onto additional problems, like diseases. “If the drug works on aging, it should work in some disease,” he said. The conundrum is figuring out where and what type of drug will be most effective.
Using AI in Virtual Research
One of the biggest challenges for clinical researchers is finding patient volunteers to participate in drug and vaccine trials. This is another area where generative AI has the potential to help, according to Zhavornkov. AI is able to create billions of patient profiles using massive data sets, allowing researchers to simulate some of the biological processes instead of using patients.
“You can take those deep neural networks that are trained on age and [create] a template,” said Zhavornkov. “Make a billion of me in the future with me as a starting condition, but with different parameters and age as a generation condition. This is a huge idea. You basically can create a virtual population.”
AI may also play a role in determining which disease targets are “druggable” – like receptors, enzymes or ion channels. These targets could be disabled, potentiated or regulated by therapeutic intervention in a commercially viable fashion.
The Impact of AI Beyond Healthcare
There are numerous applications of AI beyond healthcare, explained Zhavornkov. In the realm of sustainable energy, for example, AI has promise for developing different types of materials for carbon capture. “There’s the idea to make a material that would absorb carbon better than trees – that would be huge,” said Zhavornkov. This could open up much more cost effective solutions for reducing carbon emissions, which could be deployed widely.
New fuels and energy sources could be studied, too. “New gas combinations might allow you to do things that were previously impossible,” he said. And further down the road, we could see the “generation of new types of energy like fusion and improvements in the ability to capture solar energy or geothermal.”
These AI applications will probably take more time “but the impact is going to be much greater,” he said.