Researchers at Goethe University Frankfurt and Philipps University of Marburg, working with the Fraunhofer Institute for Translational Medicine and Pharmacology (ITMP), have developed a generative AI tool designed to reduce the number of animals required in preclinical drug testing.
The system, called genESOM, is trained on small experimental datasets and uses that learning to generate additional synthetic data points that accurately reflect the characteristics of real laboratory results. According to the researchers, the tool could reduce the number of laboratory animals needed during testing of new active ingredients by between 30 and 50%.
Regulatory and commercial drivers
Prof. Jörn Lötsch, a data scientist at Goethe University, said the tool could “make an important contribution to reducing the number of animal experiments in large areas of preclinical research.”
Scrutiny of animal testing’s translatability to human outcomes continues to grow within the scientific community. Regulators and industry stakeholders are increasingly focused on alternative testing methods, both for ethical reasons and to address high dropout rates in clinical trials, which prolong and raise the cost of drug development.
Government bodies in several markets are actively supporting this. The German federal body BfR3R, which coordinates the development of animal testing alternatives in Germany, is among those promoting replacement and reduction approaches across preclinical research.
The full study, authored by Jörn Lötsch, Benjamin Mayer, Natasja de Bruin, and Alfred Ultsch, is published in Pharmacological Research.
