Facts.AI (AI) whose use will be a game changer in the field of medicine.The technology is currently helping scientists develop the 60-year-oldThe firstNew antibiotics.
The use of Artificial Intelligence (AI) technology has led to a major breakthrough in the field of medicine, helping scientists discover new antibiotics against drug-resistant Staphylococcus aureus (MRSA).Discovery promises to be a turning point in the fight against antibiotic resistance.
"Our insight is that we can see how models can learn to predict the potential of certain molecules to act as effective antibiotics," said James Collins, a professor of medical engineering and science at the Massachusetts Institute of Technology (MIT).
"Our work provides a framework that is both efficient in saving time and resources from a chemical structure perspective and has mechanical insights that we have not had to date."
The results of this study have been published inNaturepublished on , co-authored by 21 researchers.
Research aims to 'open the black box'
The project team used deep learning models to predict the activity and toxicity of new compounds.
Deep learning involves using artificial neural networks to automatically learn and represent features from data without explicit programming.
This approach is increasingly being applied to drug discovery to accelerate the identification of potential drug candidates, predict their properties, and optimize the drug development process.
In this case, the researchers focused on methicillin-resistant Staphylococcus aureus (MRSA).
MRSA infections can range from minor skin infections to more serious and potentially life-threatening conditions such as pneumonia and bloodstream infections.
According to the European Center for Disease Prevention and Control (ECDC), nearly 150,000 MRSA infections occur annually in the European Union (EU), and antimicrobial drug-resistant infections result in nearly 35,000 deaths in the EU each year.
A team of MIT researchers trained an expanded deep learning model using an extended dataset.
To create the training data, approximately 39,000 compounds were evaluated for their antibiotic activity against MRSA. These resultant data and detailed information about the chemical structure of the compounds were then fed into the model.
Our goal in this study was to open the black box," said Felix Wong, a postdoctoral researcher at MIT and Harvard and one of the study's lead authors. These models consist of a lot of computations that simulate neural connections, and no one really knows what's going on underneath."
Discovery of new compounds
To select potential drugs, the researchers used three additional deep learning models. These models were trained to assess the toxicity of compounds on three different types of human cells.
By combining these toxicity predictions with previously identified antimicrobial activities, the researchers identified compounds that were effective against microbes and minimally harmful to humans.
Using this set of models, about 12 million existing compounds were screened.
The model identified compounds with predicted activity from five different classes based on specific chemical substructures within the molecule.
The researchers then obtained about 280 of these compounds and tested them against MRSA in a laboratory setting. This approach led them to identify two promising antibiotic candidates that belong to the same class.
In experiments with two mouse models -- one with MRSA skin infections and the other with MRSA systemic infections -- the compounds each reduced the amount of MRSA by 10-fold.