Bacterial infections are becoming increasingly difficult to treat due to the rise of multidrug-resistant bacteria, among which Acinetobacter baumannii holds a prominent and worrisome position. This unassuming bacterium, typically found in soil and water, is a master of survival, frequently developing new defense mechanisms against antibiotics and evading medical intervention. It can cause a variety of infections, including in the blood, urinary tract, lungs, and open wounds, and its insidious nature allows it to inhabit a patient without exhibiting any immediate signs of infection.
The World Health Organization sounded the alarm back in 2017, recognizing this bacterium as a ‘critical’ threat and emphasizing the urgent need for innovative antibiotics to combat it. Answering this call, a team of scientists used an unconventional ally in their search for a solution: artificial intelligence (AI).
Traditional methods of antibiotic discovery have proven insufficient in the face of the resistant A. baumannii. To surmount this challenge, the scientific team trained a machine-learning model to explore an extensive library of 7,500 potential molecular drug compounds. The AI’s task was to determine the potential of each compound to inhibit the growth of A. baumannii.
The result was an antibiotic breakthrough. The AI singled out an antibacterial compound known as abaucin, demonstrating its effectiveness specifically against a range of organisms, including A. baumannii. This discovery signaled a critical step forward in the battle against antibiotic-resistant bacteria.
Subsequent tests in mice supported the AI’s recommendation, as abaucin successfully treated wound infections caused by A. baumannii. This achievement bolstered the argument for the use of AI in antibiotic discovery, as it showcased the technology’s ability to expedite and broaden the search for novel drugs.
This is not the team’s first foray into AI-assisted antibiotic discovery. In a previous experiment, the team used machine learning to identify a molecule dubbed ‘halicin’, which demonstrated effectiveness against E. coli and various other antibiotic-resistant bacterial species.
Following the success of that experiment, the team turned its focus to what they perceive to be public enemy number one in the realm of multidrug-resistant bacterial infections: Acinetobacter baumannii.
Professor James Collins from MIT, a co-author of the paper, lauded the role of AI in the project. “I’m excited that this work shows that we can use AI to help combat problematic pathogens such as A. baumannii,” he said in a press statement.
The breakthrough study, co-authored by Jonathan Stokes, an assistant professor at McMaster University, was published in the journal Nature Chemical Biology. The research establishes a promising foundation for the continued use of machine learning in the quest to discover antibiotics capable of tackling challenging multidrug-resistant pathogens.