Artificial intelligence has identified a new class of antibiotics against methicillin-resistant Staphylococcus aureus (MRSA), a bacterium that has become resistant to treatments. This bacterium is annually responsible for the deaths of millions of people worldwide. This discovery could be a game-changer. This isn’t the first time that AI has come to the aid of medicine.
In a groundbreaking achievement in medicine, 21 researchers from the Massachusetts Institute of Technology (MIT) utilized artificial intelligence to discover an antibiotic effective against Staphylococcus aureus, a bacterium that had previously shown resistance to existing medications.
In a statement, James Collins, Professor of Medical Engineering and Science at MIT, stated:
“The insight here was that we could see what was being learned by the models to make their predictions that certain molecules would make for good antibiotics. Our work provides a framework that is time-efficient, resource-efficient, and mechanistically insightful, from a chemical-structure standpoint, in ways that we haven’t had to date.”
Their findings were published in the journal Nature. The lead authors of the study are Felix Wong, a postdoc at IMES and the Broad Institute of MIT and Harvard, and Erica Zheng, a former Harvard Medical School graduate student.
Serious Infections
Golden Staph (Staphylococcus aureus) is responsible for thousands of deaths each year. 35,000 people die from it annually in the European Union, and 80,000 people are infected in the United States. This bacterium causes skin infections and can lead to pneumonia and even septicemia.
Until now, antibiotics have been ineffective against this bacterium. However, the discovery of a new antibiotic, the first in 60 years, could be a game-changer.
Deep Learning Models
How did researchers use AI to make this discovery? Under the supervision of Collins, the scientists have used deep learning models to identify the chemical structures of antimicrobial activity and generate predictions.
According to Felix Wong,
“What we set out to do in this study was to open the black box. These models consist of very large numbers of calculations that mimic neural connections, and no one really knows what’s going on underneath the hood.”
The researchers first trained an advanced deep-learning model with significantly enlarged datasets. They created this training data by assessing around 39,000 compounds for their antibiotic efficacy against MRSA. Then, they inputted this data, along with details regarding the chemical structures of the compounds, into the model.
Wong added:
“You can represent basically any molecule as a chemical structure, and also you tell the model if that chemical structure is antibacterial or not. The model is trained on many examples like this. If you then give it any new molecule, a new arrangement of atoms and bonds, it can tell you a probability that that compound is predicted to be antibacterial.”
Two Promising Antibiotic Candidates
To refine the selection of potential drugs, the researchers utilized three additional deep-learning models, training them to anticipate the toxicity of compounds on distinct human cell types.
They integrated this data with predictions of antimicrobial activity. This enabled them to identify compounds effective against microbes with minimal harm to human cells. The team screened around 12 million commercially available compounds using these models, pinpointing candidates from five distinct classes predicted to be active against MRSA based on chemical substructures.
By purchasing and testing about 280 compounds against MRSA in a lab dish, the researchers identified two promising antibiotic candidates from the same class. In mouse models of MRSA skin and systemic infections, both compounds reduced the MRSA population tenfold.
AI and Medicine Hand in Hand
This isn’t the first time that AI has come to the aid of medicine. Last March, Canadian researchers succeeded in discovering a new antibiotic capable of killing a deadly superbug responsible for pneumonia, Acinetobacter baumannii, thanks to AI.
Scientists initially trained the AI by analyzing 6,680 unknown compounds. This resulted in a list of 240 compounds identified by the AI, which doctors then tested until they arrived at an experimental antibiotic named abaucin. This new antibiotic shows promise in combating the superbug, and a medication could potentially be available by 2030.
