First Medicine Developed by Generative AI Enters Human Trials
Biotech firm Insilico Medicine has begun human trials for a generative AI-discovered and developed medicine. The anti-fibrotic INS018_055 that generative AI helped formulate is aimed at treating idiopathic pulmonary fibrosis (IPF), a lung disease that scars lung tissue irreversibly and leads to major long-term health issues. The drug is now in phase II clinical trials in the U.S. and China after passing the safety tests for human patients.
Insilico Medicine describes itself as “generative AI-driven” and first speculated about using AI to design new medicine in scientific journals back in 2016. Since then, the company has created a generative adversarial network (GAN) to build and combine algorithms for coming up with new drug ideas using the commercially available Pharma.AI platform. Researchers used AI to both discover the target and come up with the molecular structure of a medicine that might help combat the disease. The Phase I tests of INS018_055 went well this year, leading to the plans for a randomized, double-blind study that will test the effectiveness and safety of the drug over 12 weeks on 60 subjects with IPF.
“When we started developing generative AI for drug discovery, I never expected to see the clinical and preclinical results we have today,” Insilico Medicine co-CEO and founder Alex Zhavoronkov said. “Initiating Phase II trials with this novel inhibitor for IPF represents a major milestone for deep generative reinforcement learning in drug discovery. We will explore the efficacy for patients of AI-discovered and designed treatments in clinical trials, which is a true validation of our generative AI platform. We are eager to continue to advance this potentially first-in-class therapy forward to help patients in need and show the value of generative AI in drug discovery and development.”
Applying generative AI to biology is still a new idea but is on the cusp of rapid growth. For instance, Drug discovery startup Variational AI used generative AI on its Enki platform to develop and file patents for novel small molecules that could treat COVID-19. The startup’s variational autoencoder (VAE) trained on both public and proprietary datasets to produce the protease inhibitors, which could play a key role in future COVID-19 antiviral treatments.
Nvidia is a key player in the space, counting Insilico among its technology partners for bringing generative AI to pharmaceuticals. Nvidia and pharmaceutical startup Evozyne introduced a new generative AI model that can produce proteins for medicine and other industries earlier this year. The Protein Transformer Variational AutoEncoder (ProT-VAE) is built on Nvidia’s BioNeMo framework and uses generative AI to rapidly create examples of synthetic protein designs that fit given parameters and speed up development. For Insilico, Nvidia is a central component to looking for medical drugs with generative AI. The company engaged Nvidia’s DGX server system to develop its algorithms.
“When we started, there were hundreds of companies claiming to use AI for drug discovery,” Zhavoronkov said. “Insilico was started as a deep learning pureplay at Nvidia GTC, a global conference on AI and the metaverse. We were one of the first to get the DevBox, test the latest hardware early, and now work on Nvidia DGX. We made a bet on generative AI early, tapped Nvidia’s best-in-class technology, worked hard, and now we can tangibly demonstrate substantial R&D performance boosts and real clinical-stage drugs imagined by AI.”