AI-designed virus injected into human patient in experimental treatment

A biotechnology startup has used artificial intelligence to design a bacteriophage, a virus capable of infecting and destroying bacteria, for the treatment of a human patient with a persistent infection caused by antibiotic-resistant Escherichia coli (E. coli).

Radical Numerics developed the virus variants using its AI model, Omnii, in what the company describes as the first clinical trial of phage therapy designed with this technology. The announcement was published last week, marking a potential step toward developing personalized treatments for infections that no longer respond to conventional antibiotics.

Researchers used Omnii to analyze information about the resistant bacterial strain and design genetic variants of a T7 bacteriophage capable of attaching to the bacteria and delivering genetic material into their cells. The infected bacteria can then produce additional phages, which may spread to and destroy other bacteria.

AI aims to accelerate personalized treatments

Phage therapy has been investigated and used for more than a century, but finding a naturally occurring virus capable of targeting a specific bacterial strain can take considerable time. Radical Numerics says its approach aims to speed up the process by designing customized bacteriophages in the laboratory for patients who have few remaining treatment options.

The company declined to disclose the patient’s identity, the treating physicians or the patient’s clinical progress, citing privacy concerns. It confirmed that the trial had been authorized by the U.S. Food and Drug Administration (FDA).

However, the treatment’s effectiveness remains uncertain. Results have not yet been published, and scientists have cautioned that laboratory performance does not necessarily predict how well a bacteriophage will work inside a human body.

Michael Baym, a microbiologist at Harvard University, described the prospect as remarkable if successful, while emphasizing the need to remain cautious until the treatment is completed. Brandon Berryhill, an evolutionary microbiologist at Uppsala University in Sweden, similarly noted that laboratory tests cannot fully establish a phage’s effectiveness against an infection in a real-world setting.

If the approach proves successful, Radical Numerics plans to explore additional applications of its technology, including designing cancer vaccines and identifying emerging viruses that could pose future pandemic risks.

For now, the case represents an early milestone in AI-assisted personalized medicine, rather than proof that AI-designed viruses can reliably treat resistant infections.