Chemical products, cosmetics… Tests carried out on animals on rats or rabbits in order to verify the safety and side effects of a chemical component are the subject of heated debate. American researchers at Johns Hopkins University in Baltimore believe they have found the solution: an artificial intelligence capable of accurately determining the toxicity of a chemical. Their study was published in the specialized journal Toxicological Sciences. Their artificial intelligence was designed to be able to recognize and analyze tens of thousands of chemical components on nine types of tests, from inhalation damage to aquatic ecosystems, reports the Nature magazine. According to toxicologist Thomas Hartung, this computer model would be more efficient, less expensive but also more ethical than some animal experiments consisting of throwing compounds in the eyes of rabbits to see if they are irritating or feeding the rats with chemicals. .
The potential of AI to develop
“With the power of big data, we can produce a tool that is more predictive than many tests on animals”, assures Thomas Hartung, quoted in Nature. In the United States, researchers are teaming up to carry out a larger study which consists of testing the effectiveness of artificial intelligences on 40,000 chemicals previously tested on rats. If the potential of AI is defended by scientists, there is still a long way to go before algorithms can completely replace animal testing methods.
Within the European Union, the European Chemicals Agency (ECHA) encourages companies to avoid animal testing using methods based on laboratory cell analysis when possible, says Nature. But Mike Rasenberg, ECHA’s IT Assessment Manager remains cautious. According to him the study shows “a good initiative”, says Rasenberg, but “scientifically, there is a lot of work to be done.” “No one wants animal testing, but we can’t do all of the tests yet. toxicologies with a computer. “
Read also
Cancer: a patient takes the US court against the Round up
Glyphosate found even in beer
