Share this story!
- MIT researchers use AI to identify potential antibiotics.
- Compounds effective against MRSA, showing low toxicity to human cells.
- New method offers insights into AI's decision-making in drug discovery.
AI and antibiotic discovery
Researchers at the Massachusetts Institute of Technology (MIT) have applied artificial intelligence, specifically deep learning, to discover potential antibiotic candidates. These compounds have demonstrated effectiveness in combating methicillin-resistant Staphylococcus aureus (MRSA), a bacterium responsible for severe infections and numerous deaths annually.
A notable aspect of this study, published in Nature, is the researchers' ability to decipher the decision-making process of the deep-learning model. This breakthrough allows a more targeted approach in designing new drugs.
The battle against MRSA
MRSA, a significant health threat causing skin infections and potentially fatal sepsis, has been a focus of MIT's deep learning research. The AI models employed in this study have been trained with extensive datasets, enabling them to predict compounds' antimicrobial activity and human cell toxicity. This dual prediction capability is crucial in identifying substances that can combat bacteria while being safe for humans.
The study involved screening about 12 million compounds, identifying several promising candidates against MRSA. Two of these compounds, tested in lab dishes and mouse models, showed a substantial reduction in MRSA populations, offering hope for effective treatment options.