New articles are coming out almost every day chronicling how the integration of artificial intelligence (AI) is revamping the way we do healthcare. One of the latest from the Massachusetts Institute of Technology (MIT) shows AI’s promising potential use case to help develop new antibiotics that could successfully fight hard-to-treat infections.
According to the article, “it is estimated that drug-resistant bacterial infections cause nearly 5 million deaths per year” across the globe, with that number expected to double by the year 2050. In response, researchers at the MIT Antibiotics-AI Project have been working on ways to harness the tech to address this growing problem.
Using generative AI, the team designed “more than 36 million possible compounds and computationally screened them for antimicrobial properties,” with the most promising compounds being structurally distinct from preexisting antibiotics. They eventually ended up with two compounds that could be chemically synthesized, with one being “very effective at killing N. gonorrhoeae in a lab dish and in a mouse model of drug-resistant gonorrhea infection.”
In a second round of tests, the researchers used generative AI to design molecules, allowing the tech more free rein with “no constraints other than the general rules of how atoms can join to form chemically plausible molecules.” They were then able to successfully chemically synthesize and test 22 of them, with six demonstrating strong capabilities to fight S. aureus (MRSA) that was grown in a lab dish, and one clearing the infection in a mouse model.
Additional testing is needed, but these findings are a strong indication of a promising future for AI use cases in the antibiotic drug creation space.
Last year, researchers at Stanford Medicine and McMaster University also published similar findings in a study utilizing generative AI. They began by creating a unique model, called SyntheMol, to “design new molecules that have never been seen in nature,” generating a step-by-step recipe of constructing the building blocks needed to create the compounds as well. In just a little under nine hours, the model was able to produce 25,000 possible antibiotics.
After filtering out the results that were too similar to preexisting compounds, the researchers were able to synthesize the top 58 compounds with the highest potential to kill resistant strains of A. baumannii, “one of the leading pathogens responsible for antibacterial resistance-related deaths.” By the end, six were able to effectively kill the strain in a lab, and also “showed antibacterial activity against other kinds of infectious bacteria prone to antibiotic resistance, including E. coli, Klebsiella pneumoniae and MRSA.”
As a final, comparative component of the study, the Stanford/McMaster researchers also utilized the standard computational approach where individuals employ “algorithms to scroll through existing drug libraries, identifying those compounds most likely to act against a given pathogen.” The results found from this technique barely “just scratched the surface in finding all the chemical compounds that could have antibacterial properties.”
This finding alone illustrates how AI applications are being used to completely transform the way we serve Americans throughout our healthcare system. The results of these studies offer promising foundations for the future of fighting antibiotic resistance, benefiting overall health outcomes and hopefully lowering the substantial number of related deaths per year.
As physicians’ AI adoption numbers are growing at substantial rates, with nearly two-thirds reporting usage in 2024 (a 78% increase from those surveyed in 2023), the healthcare industry will continue to greatly evolve in the coming years. These vast adoptions and resulting consequential health benefits cannot continue without maintaining an innovation-friendly landscape. As American lives continue to be positively impacted by this technology, it is key that policymakers look to the healthcare industry to set responsible guardrails that allow for the safe and continued innovation and deployment of artificial intelligence tools.
Image via Unsplash.