Here's how researchers used AI to find an antibiotic against a superbug

Scientists from the United States and Canada have used artificial intelligence (AI) to discover a new antibiotic that is potent enough to kill a superbug, which represents a significant advancement for the use of AI in the field of medicine.

Superbugs are microorganisms that can survive a variety of different antibiotics. According to the US Centres for Disease Control and Prevention (CDC), these drug-resistant bacteria infect more than 2 million people annually and cause at least 23,000 deaths in the US.

McMaster University in Canada and Massachusetts Institute of Technology (MIT) in the US participated in the study, “Deep learning-guided discovery of an antibiotic targeting Acinetobacter baumannii,” which was published on May 25 in the journal Nature Chemical Biology.

The World Health Organisation (WHO) named the bacterium as one of the most dangerous antibiotic-resistant bacteria in the world in 2017. A. baumannii is “notoriously difficult to eradicate, and can cause pneumonia, meningitis, and infect wounds, all of which can lead to death,” according to the University of McMaster. It stated that “A. baumanni is typically located in healthcare facilities where it can survive on surfaces for extended periods of time.”

The WHO’s list of superbugs highlighted bacteria with the innate ability to develop novel drug-resistant strategies and the ability to spread genetic material to other bacteria.

Medicines known as antibiotics are used to both prevent and treat bacterial infections. According to the WHO, bacteria can become resistant to antibiotics when they adapt to their use. The capacity of medications to treat prevalent infectious diseases is ultimately threatened by this.

It warns against excessive medication intake without a doctor’s prescription for treating common illnesses. “Where antibiotics can be bought for human or animal use without a prescription, the emergence and spread of resistance are made worse,” it states.

According to the WHO, infections like pneumonia, TB, and foodborne illnesses are becoming more difficult to cure with current drugs due to rising anti-bacterial resistance.

Finding the best antibacterial substances to combat bacteria can be a time-consuming and challenging procedure. Because the idea of AI is founded on the process of machines being fed vast amounts of data and training themselves to detect patterns and solutions based on them, this is where algorithms come into play.

According to MIT, the researchers first tested around 7,500 different chemical compounds on A. baumannii cultured in a lab dish to see which ones could slow the bacterium’s development.

They then supplied the machine-learning model the structures of each molecule. Additionally, they indicated to the model whether or not each structure could stop bacterial development. This made it possible for the algorithm to discover chemical characteristics linked to growth inhibition.

The model was trained before being utilised by the researchers to examine a batch of 6,680 chemicals. This study produced a few hundred findings in less than two hours. Focusing on substances with structures different from those of known antibiotics, the researchers selected 240 of these for experimental testing in the lab.

Nine antibiotics were discovered as a result of those investigations, one of which was extremely potent and successful at eliminating A. baumannii. Abaucin is the name given to this.

One thought on “Here’s how researchers used AI to find an antibiotic against a superbug”

Leave a Reply