Doctors may soon have a way to defeat MRSA, a deadly bacterium that is difficult to treat because of its resistance to some antibiotics. Researchers have used AI to create the first new antibiotics in 60 years.
AI identifies compounds to create the first new antibiotics in 60 years
In many fields, artificial intelligence (AI) is proving to be a game changer. This is certainly the case in fields that involve heavy research. One field experiencing major breakthroughs through the use of AI is medicine. As such, AI has helped scientists discover and create the first new antibiotics in 60 years.
MIT researchers have successfully used AI to identify candidates for creating a new class of antibiotics.
A team of 21 researchers, mostly from MIT and Harvard, worked on the study. Their results were published in the journal Nature.
The discovery of new compounds
The team employed a deep-learning model to predict the activity and toxicity of new compounds. In particular, the researchers focused on methicillin-resistant Staphylococcus aureus (MRSA).
To begin with, the deep-learning models were tasked with identifying chemical structures associated with antimicrobial activity. Then they sifted through millions of other compounds, generating predictions of which might have strong antimicrobial activity.
In creating the training data, the team had AI evaluate approximately 39,000 compounds for their antibiotic activity against MRSA.
Digging deeper, AI’s next job was to make predictions for which substructures of the molecule likely account for its antimicrobial activity.
In refining the selection from this large number, researchers employed three additional deep-learning models. These models were trained to assess the toxicity of compounds by focusing on only three distinct types of human cells.
The result allowed researchers to discover compounds that could kill microbes and have minimal adverse effects on the human body.
All total, researchers screened roughly 12 million compounds. All of these are commercially available. From all that data, the AI models identified compounds from five different classes. These were based on the chemical substructures within the molecules. In particular, those that were predicted to be active against MRSA.
AI narrowed it down to 280 compounds which were tested against MRSA grown in a lab dish. From there, researchers identified two from the same class which appeared to be very promising antibiotic candidates.
Read More: Bleach Can’t Kill This Common Superbug
New antibiotics effective against deadly MRSA bacterium
The new compounds discovered offer hope in the battle against the bacterium that is resistant to the current range of antibiotics.
In particular, these new compounds have demonstrated the ability to kill methicillin-resistant Staphylococcus aureus (MRSA).
The researchers conducted experiments involving two mouse models. One was focused on MRSA skin infection. The other focused on MRSA systemic infection. In both tests, each of the two new compounds reduced the MRSA population by a factor of 10.
Findings used to research drugs to fight other types of bacteria
The researchers are now using the new study’s findings to design additional drug candidates. They are using the models to seek compounds that can kill other types of bacteria.
“We are already leveraging similar approaches based on chemical substructures to design compounds de novo,” said Felix Wong, a postdoc at IMES and the Broad Institute of MIT and Harvard. “And of course, we can readily adopt this approach out of the box to discover new classes of antibiotics against different pathogens.”
MRSA: A deadly antibiotic-resistant infection
Staphylococcus aureus (S. aureus) is a common gram-positive bacterium found everywhere. It is commonly found on the skin or in the nose of many healthy people. Around 21-30% of the human population is estimated to be long-term carriers of S. aureus. It plays a role in a range of illnesses and minor skin infections.
A strain of S. aureus developed into Methicillin-resistant Staphylococcus aureus (MRSA). It creates a staph infection that is difficult to treat because of antibiotic resistance. This strain is becoming increasingly prevalent as a community-acquired infection found within institutions, such as hospitals, other healthcare facilities, and nursing homes. It can also spread in schools, workplaces, and other community settings.
It is estimated that more than 80,000 MRSA infections occur annually in the US, leading to 11,000 deaths each year, according to Children’s National Hospital. MRSA infections that reach the bloodstream cause numerous complications and fatalities. Between 10-30 percent of patients with MRSA die from the infection.
In the European Union, there are nearly 150,000 MRSA infections annually, which lead to 35,000 deaths, according to the European Centre for Disease Prevention and Control (ECDC).
Read More:Â New Cancer Treatment Uses Vibrations
The downside of all AI research
One downside to this study is common to work that employs artificial intelligence: Not knowing how AI reached its conclusions. Even the scientists who build AI cannot precisely explain how it works.
Researchers call the models AI develops “black boxes.” It refers to not knowing what the model based its predictions on.
In the case of developing new antibiotics, if the researchers knew how the models were making their predictions, it would be easier for them to identify or design additional antibiotics.
“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,” says Wong.
