The in-depth study algorithm developed at the Massachusetts Institute of Technology (MIT) has discovered new antibiotics that can treat more than 30 drug-resistant diseases. Among the microorganisms that antibiotics can target, Acinetobacter baumannii, it has infected thousands of U.S. troops fighting in Iraq.
In 2017, the World Health Organization (WHO) named Acinetobacter baumannii a bacterium that poses the greatest threat to human health because of its resistance to antibiotics. According to the Financial Times, new antibiotics can also be directed against tuberculosis and C. difficile, a bacteria that can damage the intestines and cause diarrhea.
The algorithm, which discovered a new antibiotic, was trained in the analysis of structures of more than 2,500 molecules to evaluate their antibacterial properties. The AI then searched a library of 100 million molecules to predict their ability to fight specific pathogens.
The algorithm was developed by researcher Regina Barzilai from MIT, which specializes in in-depth applications for the study of chemistry and oncology.
“The question remains whether machine learning tools really do something useful in healthcare and how we can develop them to become workhorses in the pharmaceutical industry,” said Barzilai. “Our study shows how this tool can be adapted.”
The launch of the instrument on the market will be “fraught with significant difficulties”, however, scientists note that the discovery of drugs is one of the most promising applications of AI.