A few years ago, the idea of using viruses to treat bacterial infections might have sounded futuristic to many. Adding artificial intelligence to the mix makes it seem even more like science fiction. However, phage therapy (using viruses to kill harmful bacteria) is making a remarkable comeback after being largely abandoned for decades. Researchers are now combining this century-old concept with cutting-edge AI to address one of today’s most urgent health crises: antibiotic-resistant bacteria.
Why Antibiotics Aren’t Enough Anymore
Antibiotics revolutionized medicine in the 20th century, making it possible to treat infections that were once deadly. However, over time, many bacteria have developed resistance to these drugs. Some strains, known as “superbugs,” are now so resistant that even our strongest antibiotics can’t stop them. This growing resistance has pushed scientists to explore alternative treatments, and phage therapy is one of the most promising options.
Phage therapy was first developed in the 1920s, but it fell out of favour in the 1930s when antibiotics became widely available. Now, with antibiotic resistance on the rise, researchers are reexamining phages as a powerful tool to combat infections.
What Makes Phage Therapy Special?
Phages are natural predators of bacteria. They work by infecting and destroying specific bacterial cells while leaving human cells completely unharmed. Unlike antibiotics, which often target a broad range of bacteria (including the helpful ones in our bodies), phages are highly selective. This precision makes them particularly valuable when dealing with resistant bacteria.
However, there’s a catch: not all phages work on all bacteria. Finding the right phage to target a specific bacterial strain can be time-consuming and challenging, especially in a clinical setting where quick action is needed.
The AI Breakthrough: Matching Phages to Bacteria
To solve this challenge, researchers from several top institutions, including the Institut Pasteur and Université Paris Cité, developed an artificial intelligence model that can quickly identify the best phage or combination of phages for a particular bacterial infection. Here’s how it works:
- Building a Phage-Bacteria Interaction Map: Scientists started by analyzing 403 strains of Escherichia coli (E. coli) and 96 different phages. They studied 350,000 interactions, observing which phages were effective at killing specific bacteria.
- Decoding the Bacterial Genome: By examining bacterial DNA, the researchers identified specific genetic markers that predict how well a phage will work. They discovered that the bacteria’s surface receptors—essentially the entry points for phages—are key to determining a phage’s success.
- Training the AI Model: The team used this data to train an AI system that can analyze a bacterial genome and recommend the most effective phages. This process eliminates guesswork and speeds up the selection of the right phages.
Promising Results
The AI model has already delivered impressive results. In initial tests, it correctly predicted which phages would work in 85% of cases. When the system was tested on new strains of E. coli responsible for pneumonia, it successfully designed tailored phage “cocktails” that eliminated the bacteria in 90% of cases.
This success shows that AI can transform phage therapy into a practical, fast, and highly effective treatment option. Hospitals could soon use this technology to personalize phage therapy for patients with antibiotic-resistant infections.
What’s Next?
While these findings clearly opened a new chapter, there is still work to be done. Scientists need to test how well phages perform in real-world conditions, such as inside the human body, where factors like the immune system and bacterial environments can affect treatment. Researchers are also working to expand this method to target other dangerous bacteria beyond E. coli.
If successful, AI-guided phage therapy could revolutionize the way we treat infections. It offers hope not just for combating superbugs but also for moving toward a future of personalized medicine, where treatments are tailored to each patient’s unique needs.
A New Era in Infection Treatment
The combination of phage therapy and artificial intelligence is a powerful example of how modern innovation can be combined with age-old technology to solve current problems. As we face the growing threat of antibiotic resistance, this dual innovative approach could pave the way for safer, more effective treatments, bringing us one step closer to a world where superbugs are no longer a threat.
Reference article link Gaborieau, B., Vaysset, H., Tesson, F. et al. Prediction of strain level phage–host interactions across the Escherichia genus using only genomic information. Nat Microbiol 9, 2847–2861 (2024). https://doi.org/10.1038/s41564-024-01832-5
