Artificial intelligence has been used to identify a new class of antibiotics capable of killing bacteria that cause many drug-resistant infections.
The new antibiotic was identified by researchers at MIT and McMaster University from a library of nearly 7,000 potential drug compounds.
The researchers used the machine learning models they trained to assess whether the compounds would inhibit the growth of Acinetobacter baumannii.
James Collins of MIT’s Institute for Medical Engineering and Science and Department of Bioengineering said the study supports the idea that “artificial intelligence could significantly accelerate and expand our research on novel antibiotics.”
“I’m delighted that this work shows that we can use artificial intelligence to help fight problematic pathogens like Acinetobacter baumannii.”
Acinetobacter baumannii is often found in hospitals and can cause pneumonia, meningitis and other serious illnesses.
Jonathan Stokes, assistant professor of biochemistry and biomedicine at McMaster University, said Acinetobacter can survive for long periods of time on hospital doorknobs and equipment, and can pick up antibiotic resistance genes from the environment.
“It’s really common now to find A. baumannii isolates that are resistant to almost all antibiotics.”
Elon Musk says AI isn’t ‘necessary for anything’
AI holds key to stunning breakthrough that lets paralyzed people walk again
Daily Podcast: Can Artificial Intelligence Really Take Your Job?
The researchers plan to use their model to identify potential antibiotics for other types of drug-resistant infections and hope to develop such compounds for use in patients.
Their research was published in Nature Chemical Biology.
AI is also being used in the fight against breast cancer, helping scientists develop a model that can predict whether an aggressive branch of the disease will spread.
The AI model detects changes in the lymph nodes of women with triple-negative breast cancer — one of the first places breast cancer can spread is in the lymph nodes under the ipsilateral arm, and in these cases, patients may need more intensive treatment.
Dr Anita Grigoriadis, who leads the Breast Cancer Research Unit at King’s College London, said the development would give doctors “another tool in their arsenal to help prevent secondary breast cancer”.
“By demonstrating that changes in lymph nodes can predict whether triple-negative breast cancer will spread, we are learning more and more about the important role that the immune response can play in understanding patient outcomes,” she said.
The researchers tested their AI model on more than 5,000 lymph nodes donated to the biobank from 345 patients, and the model was then able to determine the likelihood of breast cancer spreading by analyzing the immune response.
Around 15% of breast cancers in the UK are triple-negative, accounting for around 25% of breast cancer deaths.