Radical Numerics, an AI lab focused on bioscience, said the treatment is believed to be the first time a cocktail containing AI-designed bacteriophages has been given to a human. Phages are viruses that specifically infect bacteria and are increasingly sought as an alternative to antibiotics as drug-resistant infections spread.
The company publicly unveiled its AI model, Omnii-1, this week. Using the model, researchers designed viable phages and tested them against bacteria isolated from the antibiotic-resistant infection afflicting their patient. "The cocktail containing Omnii-1-designed phage variants eliminated the patient's antibiotic-resistant isolate in vitro," the company said in its write-up, adding that the investigational cocktail was subsequently administered under compassionate use.
The write-up stops there. Radical Numerics published no results from the human administration, citing the need to protect the patient's privacy, and warned that observations from a single patient do not establish that the therapy is generally safe or effective. It did not say what infection the patient had, though the research describes testing Omnii-1's ability to predict phages that infect Pseudomonas aeruginosa, a bacterium that can cause resistant infections in burn wounds and persistent lung infections in cystic fibrosis patients. The model "performed competitively" against a specialised model at predicting phage-host interactions, according to the company, though nothing in the write-up indicates the patient was infected with Pseudomonas.
The work builds on earlier research, including a Stanford University and Arc Institute project in which AI-designed phages targeting E. coli outperformed a biological predecessor in laboratory tests. Radical Numerics said its result marks "an important milestone in translating AI-designed biology from computational predictions and laboratory validation to clinical administration." The company hopes the work will lead to a process for designing and building phages targeted to individual bacterial infections. The Register reported that Radical Numerics did not respond to questions for this story.
