AI drives MIT research into developing antibiotics against superbugs
Last year, researchers at the Massachusetts Institute of Technology (MIT), in the United States, launched an innovative challenge to an artificial intelligence system: to develop completely new molecules with the potential to become effective antibiotics.
After months of intensive training, artificial intelligence managed, in just one or two days, to generate more than 29 million new molecules, a feat that marks a new era in drug discovery.
Traditionally, the identification of new drugs is a time-consuming and complex procedure. However, the application of artificial intelligence is beginning to revolutionize this panorama, promising to significantly speed up the process.
Research at MIT focuses on fighting infections that have become resistant to antibiotics, responsible for more than a million deaths annually worldwide. The capacity of existing medicines does not keep up with the evolution of these threats.
More on this story: Artificial intelligence recreates the face of a victim in Pompeii who tried to escape Vesuvius
MIT professor of medical engineering James Collins highlighted the imbalance between the increase in resistant pathogens and the decrease in the development of new antibiotics over the decades.
Artificial intelligence advances in creating antibiotic compounds
A recent study, published in the journal “Cell”, is part of the Collins laboratory’s Antibiotics-AI project and demonstrates the crucial role of artificial intelligence in achieving important advances in medicine.
The team synthesized some of the AI-generated compounds and managed to eradicate a resistant infection in a mouse using one of them. In additional tests with another method of generating molecules, the positive results were repeated, solidifying the feasibility of developing fully AI-designed medicines to treat serious infections.
Learn more: Artificial intelligence generates innovative viruses with the potential to combat resistant bacteria
The current challenge in antibiotic development
The conventional method for creating new antibiotics involves testing compounds individually or analyzing soil samples for promising molecules. This process has proven to be inefficient for current needs.
Since the 1980s, the FDA (United States Food and Drug Administration) has approved a few dozen new antibiotics, most being small modifications of existing drugs.
The development of new medications requires considerable time, and human trials are still in the early stages and will take an extended period of time.
See Also
Attract prosperity: check out the lucky numbers for each sign to bet on
Researchers identify doctor in Pompeii through his surgical instruments
Luigi Mangione admits to having shot CEO and pleads guilty at federal level
Collins explained that the last few decades have been marked by a “vacuum of discoveries”, where new antibiotics emerged, but they were very similar to the previous ones, just adapted versions.
On the same topic: Scientists reverse superbug resistance to vancomycin in groundbreaking discovery
The situation is worsened by economic issues, according to Collins. Developing an antibiotic has a similar cost to that of a medicine for cancer or hypertension, for example.
The difference lies in the time of use: antibiotics are used for a few days, while medicines for cancer or hypertension can be consumed for months or years. Consequently, new antibiotics generate a fraction of the profit, resulting in few options emerging on the market for complex infections.
Artificial intelligence in modern medicine
Collins’ laboratory has been dedicated to studying antibiotics for approximately 20 years. Six years ago, the team decided to focus on artificial intelligence as the main tool for discovery. Initially, AI was employed to analyze libraries of known compounds, which led to the identification of molecules capable of fighting infections in innovative ways.
A non-profit organization, Phare Bio, was created from the project with the aim of bringing the most promising compounds to market. The expectation is to begin clinical trials with halicin, a drug originally developed to treat diabetes in 2009, but which the team later discovered to have a strong antibiotic action.
More on this story: Superbug discovered in deep cave in Bulgaria resists next-generation antibiotics
While new drug development continues to take time and human trials take time, AI is already accelerating the early phase of discovery. This advance contributes to reducing costs and increasing the chances of success.
Collins stated that artificial intelligence has made it possible to explore a much broader chemical spectrum than that available in traditional libraries, opening up paths to completely new molecules.
According to him, the same approach can be adapted to other areas of medicine, as all the AI methods used in the project can be easily applied to the development of other types of treatments.
