The research team at MIT has announced a breakthrough in the design of a drug to combat methicillin-resistant Staphylococcus aureus (MRSA) bacteria in a petri dish, utilizing design and predictive efficacy from an artificial intelligence model.
The deep learning model was trained on a database of 39,000 chemical structures to generate predictions on which substances would exhibit antimicrobial properties. Additionally, the team created three more models to predict substances that may be harmful to humans. Following this, the models were run against a library of 12 million compounds to identify potential drugs.
The results of the predictions indicated that there were five groups of substances that could be viable for use. Consequently, the research team procured all 280 substances for experimental testing in petri dishes and discovered that two of them showed significant potential as medications.
MRSA infections affect approximately 80,000 people a year in the United States, with a mortality rate of around 10,000 individuals annually. The research team has now passed on the data to Phare Bio, a non-profit medical organization, for further analysis and preparation for clinical trials.
TLDR: MIT researchers have successfully designed a drug to combat MRSA bacteria using artificial intelligence models. After training the deep learning model on a database of chemical structures, they identified potential compounds and tested them in petri dishes. Two substances showed promise as medications. The research team has now shared their findings with Phare Bio for further analysis and clinical trial preparation.
Leave a Comment