Insilico Medicine Showcases Generative AI’s Role in Accelerating Drug Discovery and Aging Research.

Speaking at the recent Bio-IT World Conference & Expo, Alex Zhavoronkov, founder and CEO of Insilico Medicine, emphasized the company’s mission of extending human healthspan by harnessing generative AI. Celebrating its 11th year, Insilico is advancing AI-powered drug discovery with a particular focus on aging-related diseases.

Zhavoronkov acknowledged the complexity of aging, noting that while it’s easy to observe, effective interventions remain scarce. He explained Insilico’s core belief: if a drug can influence the biological mechanisms of aging, it may also provide therapeutic benefits across a spectrum of diseases.

To support this approach, Insilico has built a robust AI platform covering the full spectrum of drug development—from target identification and molecule design to clinical trial prediction. Unlike companies that rely solely on partnerships to test AI-generated outputs, Insilico invests heavily in validating its own models. “We actually bet a lot of money to actually see if the molecule works or not,” Zhavoronkov remarked, highlighting their unique model of internal testing.

The company’s strategy is centered on creating and licensing new drug candidates. One of their notable successes includes an $80 million upfront deal with Exelixis for a USP1 inhibitor. Meanwhile, their lead therapeutic program targeting idiopathic pulmonary fibrosis (IPF) has successfully completed Phase 2A trials.

Zhavoronkov used a culinary metaphor to describe their approach: “We precook the cookies and sell them to the cookie makers, while testing our own cookie-making software—and eventually sharing that software with others.”

Tackling the Validation Bottleneck:

One of the biggest hurdles in AI-driven drug discovery, Zhavoronkov explained, is experimental validation. Unlike outputs in text or image generation that can be quickly assessed by humans, biological predictions must be confirmed through laboratory research. Nonetheless, Insilico has shortened the development timeline dramatically—reducing the process from concept to preclinical candidate to roughly 13 months, including efficacy studies in animal models and preliminary toxicology testing.

To support this rapid pace, the company has established a global footprint, with a significant operational base in China. They work with more than 40 contract research organizations (CROs), often executing experiments in parallel or redundantly to ensure dependable outcomes.

“These are realistic timelines,” said Zhavoronkov. “At this point, we've optimized drug discovery to its practical limits—further speedups would require regulatory evolution.”

AI in Action:

Zhavoronkov also shared case studies that demonstrate the potential of Insilico’s technology. In one 2019 experiment published in Nature Biotechnology (DOI: 10.1038/s41587-019-0224-x), the team used its GENTRL model to generate six drug candidates targeting DDR kinase in just 21 days—four of which succeeded in binding studies, and one advanced to animal testing within 46 days.

In another study, Insilico tackled so-called “dark targets”—proteins with unknown structures—by combining early versions of AlphaFold with molecular dynamics simulations. They succeeded in designing potent molecules (180 nanomolar range) in 50 days, with findings published in Chemical Science (DOI: 10.1039/D2SC05709C).

Although Insilico does not conduct drug repurposing internally, it has collaborated with external groups like the ALS Consortium. One joint project led to the identification of promising targets and drug candidates, which were later tested in fly models. One such candidate progressed to a Phase 2 investigator-initiated trial by 4B Therapeutics, all within two and a half years.

Modeling Aging with AI:

A key pillar of Insilico’s work is their development of "life models"—AI systems designed to interpret biological processes across the human lifespan. Inspired by the concept of "world models" in AI, these tools aim to understand biology from birth to old age using data like methylation, transcriptomics, and proteomics.

“Age is the one variable everyone shares,” Zhavoronkov said. Their flagship life model, dubbed Precious, can perform multiple tasks such as predicting biological age, annotating omics data, and generating synthetic datasets across generations. The latest version, Precious3, is a multimodal transformer model trained on a mix of biological and textual data—and it’s now open-source.

To validate AI predictions, Insilico established a highly automated lab in Suzhou during the pandemic. The facility runs a complete reinforcement learning loop, where AI models are constantly refined based on experimental feedback. The team is even training humanoid robots to carry out lab tasks typically not suited to automation.

With AI and automation at its core, Insilico Medicine is pushing the boundaries of what’s possible in drug discovery and the science of aging.

Source: https://www.bio-itworld.com/news/2025/04/17/insilico-s-alex-zhavoronkov-highlights-generative-ai's-impact-on-drug-discovery-and-aging-research

This is non-financial/medical advice and made using AI so could be wrong.