AI news ·
Penn Medicine's AI Breakthrough Accelerates Antiviral Drug Discovery Using Minimal Data
Penn Medicine researchers developed an AI method to rapidly identify antivirals using limited data. This approach sped drug discovery against EV71, a virus causing hand, foot, and mouth disease.
Penn Medicine Researchers Develop AI Technique for Rapid Antiviral Discovery Using Limited Data
Researchers at the University of Pennsylvania’s Perelman School of Medicine have developed an artificial intelligence method capable of efficiently identifying antiviral drug candidates, even when starting with a small dataset. This approach combines AI algorithms with traditional laboratory techniques to identify compounds active against human enterovirus 71 (EV71), a common cause of hand, foot, and mouth disease.
The study, published in Cell Reports Physical Science, describes how the team trained a machine learning model using only 36 small molecules. The AI was programmed to detect molecular shapes and chemical features linked to antiviral effects, scoring compounds based on their potential to inhibit EV71.
To validate the AI predictions, researchers tested eight shortlisted compounds in cell-based experiments. Five of these compounds demonstrated the ability to slow viral replication, a success rate notably higher than typical results from unguided screening methods. This highlights the AI’s capability to streamline early drug discovery stages.
This AI-driven approach can reduce the discovery timeline from months to days, an advantage in scenarios with limited time, budget, or experimental data. EV71 infections vary from mild symptoms like rash and fever to severe neurological issues, especially in young children and immunocompromised individuals. Currently, no FDA-approved antivirals target EV71.
Further computer simulations explored how the effective compounds interact with the virus, revealing binding to specific sites. These interactions could prevent the virus from changing shape and entering host cells, offering insight into how these molecules might serve as treatments.
The researchers suggest this method could serve as a blueprint for faster antiviral drug development, helping accelerate responses to future outbreaks caused by enteroviruses, emerging respiratory pathogens, or resurging viruses such as polio.
This project involved collaboration with Procter & Gamble and Cornell University. Funding came from the Langer Prize, the National Institutes of Health, and the Defense Threat Reduction Agency. An invention disclosure related to this work has been filed.