Will AI and digital twins make animal testing in drug discovery obsolete?

AI is transforming drug discovery. Digital twins — virtual replicas of biological systems — are beginning to change how scientists test, predict, and validate new treatments. But can these technologies go so far as to eliminate animal testing entirely?

In this episode of Tech Tomorrow, Zühlke’s David Elliman speaks with Professor Julie Frearson, Chief Scientific Officer at Charles River Laboratories, about the promise and limits of this shift, and what the future of preclinical science might really look like.

Professor Julie Frearson is Senior Vice President and Chief Scientific Officer at Charles River Laboratories, one of the world’s leading partners in drug discovery and development. She oversees the company’s strategic venture funds and innovation partnerships and has spent her career at the intersection of pharmaceutical science and technology innovation.

With decades of experience in early-stage drug discovery, she offers a grounded, practical perspective on how AI and digital twins are reshaping the industry — and why animal testing is likely to evolve, not disappear.

AI is already transforming early drug discovery

For small molecules, AI has already become an essential tool. Algorithms now help scientists identify new chemical entities, predict whether they’ll bind to the right biological targets, and model their drug-like properties.

As Professor Frearson explains: “In the computer, you end up doing much bigger experiments and interrogating much more chemical space than you would if this was entirely reliant upon traditional experimental processes.”

The result is faster discovery cycles and more targeted experimentation. But Professor Frearson cautions that cost reduction isn’t the main story; the real value is in quality and scale. In a nutshell, AI allows researchers to explore more possibilities, earlier. Of course, that depends on labs being able to produce high-quality, AI-ready data at scale.

Virtual animals are no longer science fiction

One of the most striking advances is the rise of digital twins (virtual models that replicate biological systems). In the context of drug discovery, this means virtual animals.

Professor Frearson explains that while technology is still not ready to create virtual animals that can show us the difference between an untreated and treated animal, scientists are already using virtual animals to replace control arms in studies.

Instead of using live control animals, which often produce repetitive data, researchers can now use retrospective datasets to simulate those controls. Charles River Laboratories has already shown, across more than 20 studies, that replacing control animals with virtual ones has no impact on experimental conclusions.

This translates into a practical, ethical, and scientific win, reducing animal use while maintaining data integrity.

The complexity challenge: why full virtualisation is still far away

So, can we build a fully virtual organism? Not yet, and maybe not soon. “You would blow everyone’s mind if you really tried to build a virtual human or a virtual animal based on the data we have today", says Professor Frearson.

Instead, the industry is taking a modular approach by building models for specific systems like the liver, heart, or kidneys, where toxicity often causes drug failure in humans. By focusing on these critical sub-models, researchers can identify safety issues much earlier in the pipeline.

Regulators are cautiously opening the door

Regulatory bodies are starting to recognise the potential of AI and digital models. The FDA has already approved several in silico (computer-based) models, and Europe’s EMA has long supported New Approach Methodologies (NAMs) where scientifically appropriate.

But widespread adoption will take time. Professor Frearson notes: “Even 10 years from now, I think regulators will still be looking at hybrid data sets — combining AI-derived predictions with in vivo data.”

Explainability is the new safety net

In high-stakes fields like drug development, black-box AI simply isn’t acceptable.

David Elliman frames it clearly: “If you can’t explain why your AI made a decision, how can you trust it to make important choices? Explainability isn’t a nice-to-have; it’s mandatory.”

Professor Frearson agrees that human oversight remains essential:

That’s exactly why governance-first approaches to AI – where transparency, accountability, and compliance are designed in from day one – are becoming non-negotiable.

The speed of technology vs. the pace of regulation

Technology often moves faster than law — and that tension is especially visible in pharma. Yet Frearson argues that’s not necessarily a bad thing.

“You can derive a huge amount of benefit from that technological clock speed in the earlier parts of drug discovery,” she says. “Where the regulators aren’t yet involved, we can go wild — as long as we don’t go crazy.”

In other words: let innovation thrive early, where the risks are lower, and use regulation to ensure safety and trust downstream.

So, will AI and digital twins make animal testing obsolete?

Digital twins and AI won’t erase animal testing; but they are truly redefining what’s possible in drug discovery. As these models mature, we’ll see smaller, smarter studies, faster development cycles, and a deeper understanding of safety and efficacy.

Estimates suggest that we understand only around 5% of human biology. That leaves a long way to go before simulation can replace experimentation. And while technology can’t replace biology, it’s helping us ask better questions. And that’s how real breakthroughs begin.