The spectre of a post-antibiotic era, where common infections could once again prove lethal, is rearing its head.

Indeed, the World Health Organization (WHO) projects antimicrobial resistance (AMR) will cause 10 million annual deaths by 2050 and inflict significant economic harm.

Compounding this, the new antibiotic pipeline has nearly run dry, with almost no new antibiotics discovered in the last few decades.

Into this void, bacteriophage therapy is re-emerging, revitalized by artificial intelligence (AI) and modern microbiology.

These bacteriophages – viruses that infect and replicate only in bacterial cells, that are known as ‘phages’ for short – were first identified and named by Félix d’Herelle, a French-Canadian scientist at the Pasteur Institute in Paris, in 1917.

Phages are the most abundant biological entities on Earth. They exclusively target and destroy bacteria, rendering them harmless to any type of cells – human or otherwise. In fact, the human body itself hosts a greater number of phages than its own cells.

The therapeutic principle of phage therapy involves isolating specific phages from the environment, identifying those effective against a particular pathogenic bacterium (such as Salmonella), and then cultivating these selected phages to high concentrations for administration.

However, this early momentum shown in d’Herelle’s work was soon overshadowed, with the arrival of penicillin, discovered by Alexander Fleming in 1928, ushering in the antibiotic era. These new "wonder drugs" were broad-spectrum, relatively straightforward to manufacture and delivered remarkable clinical outcomes, quickly capturing the medical and industrial focus.

The adoption of phage therapy then faced significant practical impediments due to the technological limitations of the early to mid-20th century. Limited technology made it difficult to consistently isolate specific phages, as identifying the correct phage for a particular bacterial strain was an incredibly time-consuming manual process.

This, alongside further challenges in standardizing treatments for reliable effects and scaling up production, presented a major hurdle. Additionally, existing regulatory frameworks, designed for chemical drugs, were ill-suited for these biological agents.

Consequently, as antibiotic development flourished, phage research and application in many Western countries receded. Yet, the current, escalating global crisis of antimicrobial resistance, aggravated by the critically dry pipeline for new antibiotic discovery, has forcefully brought phages back into the scientific spotlight.

Currently, finding the right phage for a patient is a painstaking manual process. Scientists must perform a “phagogram” – a series of lab tests that are like trying countless keys in a single lock. They physically mix each available phage from their library with the patient’s specific bacterial infection to see which one works.

Finding the best available phage candidate for a single patient would require hundreds of thousands of manual tests. But laboratories have the capacity to perform a few hundred at most, making it one of the main obstacles preventing phage therapy from breaking through as mainstream medicine.

However, the latest advances in artificial intelligence (AI) could help address such challenges. Scientists are soon expected to be able to predict how well a phage will work solely from its genetic code, marking a fundamental shift from manual to digital phagogram models.

Accordingly, we are getting close to developing genuinely personalized and highly effective ways to combat infections through bacteriophage therapy, offering a much greater chance of success, particularly where previous treatments have fallen short.

Moreover, AI’s understanding of phage genetics and predicted interactions extends beyond identifying optimal single phages, as it can help design phage “cocktails” – i.e. combinations of phages needed to best eliminate bacterial targets.

These will be crucial in developing further treatments as bacteria are genetically very diverse, with a single species often encompassing numerous strains and having their own susceptibilities. Moreover, their inherent adaptability means resistance to single therapeutic agents is an ever-present threat.

AI algorithms, however, can already help address such challenges by constructing and speeding up the process of multi-phage cocktails with considerable sophistication. The outcome is a powerful therapeutic, designed not only to be effective against a wider array of bacterial species, but also to significantly curtail the emergence of resistance.

This is precisely the challenge we are tackling at Phagos. As our CTO, Adèle James, explains: “The synergy between artificial intelligence and microbiology is fundamentally reshaping phage therapy. We're now looking at delivering highly targeted solutions against evolving bacterial infections in a matter of weeks instead of years. This far outpaces the typical decade required for new drug development.”

However, these sophisticated AI models can only be achieved and scaled up thanks to equally impressive strides in modern microbiology and bioengineering. Sophisticated automated high-throughput screening platforms enable the rapid assessment of countless phage-bacteria interactions. This wealth of data is then efficiently processed by advanced bioinformatics tools, which help identify the most potent and specific phages from these extensive screenings.

At the same time, genomic tools map out the DNA of these selected phages, giving us a full picture of their strengths. Finally, automated liquid-handling systems make the processes of testing and preparing for manufacturing much faster and more precise, paving the way for scalable production.

The convergence of AI and advanced microbiology is not merely theoretical; it is yielding tangible results that signal a shift away from over-reliance on conventional antibiotics. This "phage renaissance" is already demonstrating practical value in critical areas and points towards a more resilient future against bacterial threats.

In agriculture, where AMR poses a significant threat to our food production systems, phages offer a targeted method to control bacterial diseases. For instance, phages have shown to be very effective at combatting pathogens like E. coli in poultry farms. This also significantly reduces the need for antibiotics, leading to safer food and a diminished AMR footprint from this sector.

When it comes to human health, phages are providing a crucial new line of defence against multi-drug resistant infections where other treatments have failed. A notable 2024 study in Nature Microbiology, for instance, reviewed 100 consecutive cases of bacteriophage treatment for challenging infections. The findings reported clinical improvement in 77% of these difficult-to-treat infections, with complete bacterial eradication in 61%.

For phage therapy to truly advance, tailored regulations are essential to ensure both safety and patient access. Continued R&D investment is also vital to drive the science, support robust clinical trials, and eventually scale production.

Crucially, collaboration across academia, industry and regulatory bodies will be key to standardizing approaches and integrating phages into our core strategies against antimicrobial resistance.

With thanks to Ilias Theodorou, Marine Feyereisen, Ana Flack and Andrea Di Gioacchino for their contributions.