Can artificial intelligence put an end to animal testing?

From animal lovers to lab technicians, no one enjoys subjecting animals to scientific tests.

But it is necessary and unavoidable as it helps ensure that drugs and other substances are ultimately safe for human use.

Researchers have long sought to find non-animal alternatives. AI systems are currently working to accelerate this step.

One of the AI applications used in this field is easy and simple, yet it is said to have proven its effectiveness; therefore, using it to handle, compile, and analyze all current and available global test results will help prevent the need for unnecessary new tests.

And this is useful because it can be difficult for scientists to sift through decades of data to find exactly what they are looking for and analyze it, says Joseph Manubay, a senior research analyst at the American nonprofit organization, Physicians for Responsible Medicine.
He adds, “I am very excited to apply AI models like ChatGPT to extract and compile all this available data and make the most of it.”

Thomas Hartung, a professor of toxicology at Johns Hopkins University in the United States and director of the Center for Alternatives to Animal Testing, says, “Artificial intelligence is as good as, or better than, humans at extracting information from scientific papers.”

And when it comes to current animal testing, Professor Hartung says that the need to test new chemicals is one of the main reasons. With more than 1,000 new compounds entering the market each year, there is a lot to test.

Hartung adds that trained AI systems have recently begun to have the capability to determine the toxicity of any new chemical substance.
The professor explains that the availability of tools through which we can press a button and get an initial assessment, which gives us some indicators that “this is a problem,” would be extremely useful.

Professor Hartung adds that while software systems have long been used in toxicology, artificial intelligence provides a “huge leap forward” in terms of power and precision.

He says, “This suddenly creates opportunities that didn’t exist before,” adding that artificial intelligence is now involved in every stage of toxicity testing. In fact, artificial intelligence is primarily used in the production of new drugs.

And AI systems are not perfect in determining chemical safety, of course. One of the problems it faces is the phenomenon known as data bias.

One example of this is if the AI system and its algorithm were trained using health data primarily from one ethnic group.

The danger lies in the fact that its calculations or conclusions may not be entirely suitable for people from different ethnic backgrounds.

But as Professor Hartung points out, testing human drugs on animals can sometimes be useless.

For example, the arthritis drug Vioxx passed animal testing, was then approved for human use, but was eventually withdrawn from the market after studies showed that long-term use of the drug by humans increased the risk of heart attacks and strokes.

On the other hand, some widely used drugs may fail animal tests, such as the pain reliever aspirin, as it is toxic to mouse embryos.

Professor Hartung concludes that in a number of cases, artificial intelligence has already proven to be more accurate than animal tests.
One of the AI projects built to try to replace the need for animal testing in the future is called AnimalGAN. This program, developed by the U.S. Food and Drug Administration, aims to accurately determine how mice interact with any specific chemical.

The artificial intelligence was trained using data from 6,442 real mice across 1,317 therapeutic scenarios.

A similar international project called Virtual Second Species is focused on creating a virtual dog powered by artificial intelligence, trained using data derived from the results of previous dog tests.

Kathy Vickers, head of the innovation department at the National Centre for the Replacement, Refinement and Reduction of Animals in Research in the UK, which is part of the plan, explains that new drugs are currently being tested for the first time on rats and dogs to check for potential toxicity before starting human trials.

The main challenge facing AI testing is obtaining regulatory approval. Dr. Kathy acknowledges that “full acceptance of it will take some time.”

However, Emma Grange, Director of Scientific and Regulatory Affairs at the advocacy group Cruelty Free International, says that every effort should be made to ensure the gradual phasing out of animal testing.
She adds, “At the moment, it is unclear how or whether new technologies like artificial intelligence can actually contribute to ending animal testing, rather than just reducing or improving such tests.”

Emma explains, “But we know that using animals as models to protect human health and the environment is outdated science, and we hope that artificial intelligence will eventually play a role in moving away from using animals in any test or experiment.”

However, Kirsten Kleinschmidt-Dör, chief veterinary officer at the German pharmaceutical company Merck, one of the sponsors of Virtual Second Species, says that animal testing cannot disappear overnight.

She adds, “The use of animals is necessary for good reasons, and it is mandatory in many aspects.” But we believe in a future where we will identify better solutions free from animal testing for the unresolved problems we need today.