AI pinpoints new anti-aging drug candidates
More than 70% of the drugs identified by artificial intelligence extended the lifespan of C. elegans worms.
May 29, 2025
LA JOLLA, CA—Everyone wants to slow down aging—but despite decades of research, truly effective anti-aging drugs remain elusive. Most drug discovery efforts for aging have focused on targeting single biological pathways, an approach that often falls short against the complexity of the aging process.
Now, scientists at Scripps Research and the biotechnology company Gero have used artificial intelligence to identify drugs that combat aging by targeting multiple age-related biological pathways at once. In their new study, published in Aging Cell in May 2025, more than 70% of the anti-aging drugs identified by an AI tool significantly extended the lifespan of the microscopic worm Caenorhabditis elegans. This also demonstrates that AI can successfully guide the design of drugs that work through this complex mechanism.
“This study shows that artificial intelligence can help us go beyond the traditional ‘one-drug, one-target’ mindset,” says co-senior author Michael Petrascheck, professor at Scripps Research. “By embracing the complexity of polypharmacological targeting, we were able to identify compounds that produce stronger and more reliable effects on lifespan than anything we’ve seen in previous screens.”
The research team embraced an approach known as polypharmacology—the idea that many of the most effective medicines in use today work by interacting with multiple proteins at once. Often, this can be undesired, leading to side effects. But when it comes to conditions like aging that affect many biological systems at once, Petrascheck and his colleagues thought hitting different targets at once could be an advantage—as long as those targets were carefully selected.
“Living systems are incredibly resilient—both to damage and to interventions. Think of a machine with many backup systems: turning off just one switch rarely does much. But if you press the right combination of switches, you might get a major change,” says co-senior author Peter Fedichev of Gero.
The group used a type of AI tool known as a machine learning network. They gave the network access to previous studies on C. elegans’ longevity, as well as databases of known drug mechanisms. They focused on drugs known to target specific proteins—dopamine, serotonin and histamine receptors—linked to aging processes. The model then identified existing drugs that it predicted to act on all three targets simultaneously.
When the researchers tested 22 of the compounds identified by the model in worms, 16 of them extended lifespan. One, a novel compound not currently in clinical use, increased the lifespan of C. elegans by 74%. Several FDA-approved drugs, including two antipsychotic medications used to treat schizophrenia, also performed especially well.
The new data will likely not lead to the immediate human use of any existing drugs to combat aging—clinical trials to study their full effects would be needed—but verifies both the overall benefit of polypharmacology in anti-aging as well as the key involvement these targeted brain pathways have in aging.
The team adds that the results also pave the way for an expanded use of artificial intelligence in polypharmacological design. While the current study focused on existing drugs, future studies could ask AI models to design completely novel drugs that aim to simultaneously target multiple pathways. Ultimately, the work could help lay the foundation for next-generation therapies that address age-related diseases—such as Alzheimer’s, cardiovascular disease or frailty—not by targeting a single mechanism, but by nudging multiple biological systems at once.
In addition to Petrascheck and Fedichev, authors of the study, “AI-Driven Identification of Exceptionally Efficacious Polypharmacological Compounds That Extend the Lifespan of Caenorhabditis elegans,” include Khalyd Clay of Scripps Research; and Konstantin Avchaciov, Kirill Denislov and Olga Burmistrova of Gero.
This work was supported by funding from the National Institutes of Health (R01 AG067331 and R01 AG069206).
For more information, contact press@scripps.edu