Can AI End Animal Testing?

From animal lovers to laboratory technicians, no one enjoys subjecting animals to scientific testing.

Instead, it is done to help ensure that drugs and other substances are safe for human consumption.

Researchers have long looked for non-animal alternatives. Artificial intelligence (AI) systems are now accelerating this task.

One application of AI in this field is simple but said to be proving effective – using it to sift through all existing and available global animal test results to prevent the need for unnecessary new tests.

That's useful because it can be difficult for scientists to find and analyze decades of data for what they are, says Joseph Manopello, a senior research analyst at the Physicians Committee for Responsible Medicine, a U.S. nonprofit.

“I'm very excited about the application of AI models like ChatGPT to extract and synthesize all this available data,” he says.

Thomas Harting is a professor of toxicology at Johns Hopkins University in the US, and director of the Center for Alternatives to Animal Testing. He says: “AI is as good as or better than humans at extracting information from scientific papers.”

When it comes to current animal testing, Professor Harting says the need to test new chemicals is one of the main reasons. And with more than 1,000 new such compounds entering the market each year, there is much more to be tested.

Professor Harting says trained AI systems are beginning to be able to determine the toxicity of new chemicals.

“Having tools available where we can push a button, and we get an early diagnosis, that's giving us flags of 'there's a problem' … will be very helpful.”

Professor Harting added that while software systems have long been used in toxicology, AI is providing a “huge leap forward” in both power and accuracy.

“It's suddenly creating opportunities that weren't there before,” he says, adding that AI is now involved in every step of toxicology testing. AI is even being used to develop new drugs in the first place.

Of course, AI systems are not perfect in determining chemical safety. One problem is a phenomenon known as data bias.

An example of this is if an AI system and its algorithms are trained primarily using health data from an ethnic group.

But as Professor Harting points out, testing human drugs on animals is sometimes of little benefit.

For example, the arthritis drug Vioxx passed the animal testing phase, only to be advanced and eventually withdrawn from sale after studies showed that long-term use in humans could cause heart attacks. The risk of falls and stroke increases.

On the other hand, some widely used drugs may have failed animal tests, such as the pain reliever aspirin, which is toxic to rat embryos.

Professor Harting concluded that in many cases AI is already proving more accurate than animal testing.

An AI project called AnimalGAN is being built to try to replace the need for animal testing in the future. Developed by the US Food and Drug Administration, the software aims to be able to accurately determine how mice will react to any given chemical.

A similar international project called Virtual Second Spaces is creating an AI-powered virtual dog, which is being trained using data from historical dog test results.

Cathy Vickers is head of innovation at the UK's National Center for the Replacement, Refinement and Reduction of Animals in Research, which is part of the scheme.

She explains that new drugs are currently tested on both rats and dogs to test for possible toxicity, before human trials begin.

One of the biggest challenges for AI testing going forward is getting regulatory approval. Dr Vickers acknowledged that “full acceptance will take time”.

However, Emma Grange, director of science and regulatory affairs at the pressure group Cruelty Free International, says every effort should be made to ensure animals are kept out of the testing phase.

“At this point, it's not clear how new technologies like AI can actually contribute to eliminating animal testing, rather than reducing or improving such testing,” she says.

“But we know that using animals as models for human health and environmental protection is old science, and hope that eventually AI can play a role in the transition away from using animals in any test or experiment. Is.”

However, Kirsten Kleinschmidt Doer, chief veterinary officer at German pharmaceutical company Merck, says animal testing cannot disappear overnight. His firm is one of the sponsors of Virtual Second Spaces.

“The use of animals is important and for good reasons, mandatory in many aspects,” she says. “But we believe in a future where we identify better, free animal testing solutions to the unsolved problems we need today.”