Editor’s note: Find the latest COVID-19 news and guidance in Medscape’s Coronavirus Resource Center.
Machine learning has come a long way in the quarter-century since a computer nicknamed Deep Blue shocked the world by beating chess champion Garry Kasparov. Today, when our smartphones have far more computing power than Deep Blue, scientists have trained their sights on even bigger opponents, including potentially fatal illnesses like cancer, heart disease, and COVID-19.
When supercomputers hunt for new drug cocktails to treat these conditions, scientists can feed the machines mountains of data from decades of studies to help inform the analysis. But the coronavirus is still too new and mutating too rapidly for scientists to turn to these usual strategies.
Researchers at the Massachusetts Institute of Technology have a new way to address the lack of data on the new virus. They’re training computers to run algorithms patterned after signaling networks in the human brain. Like the brain, these neural networks can “learn” and adapt to rapidly changing information, forging new connections on the fly.
To identify drug combinations that might work against COVID-19, the investigators are asking their computer neural network to assess two things at once.
One of those is to search for drug pairs that will be more powerful antivirals together than either drug on its own. This concept of two medicines being more effective in concert is known as “drug synergy.”
The computer also looks for parts of a disease that the drugs target, such as proteins or genetic mutations linked to a condition. The idea behind these two approaches is that the machines can “learn” which drug cocktails might have the most antiviral power.
In their study,published inthe Proceedings of the National Academy of Sciences the MIT scientists reveal two potential drug cocktails they found using this approach. One combines remdesivir, which the FDA already approved to treat COVID-19, and reserpine, a medication for high blood pressure. The other pairing is remdesivir and an experimental drug called IQ-1S, one of a family of medicines used to treat autoimmune diseases like rheumatoid arthritis.
These drug cocktails haven’t yet been proven effective against COVID-19 in human trials. But the study results can help drug developers pinpoint which combinations might make the most sense to test as they search for new treatments.
Proceedings of the National Academy of Sciences: “Deep learning identifies synergistic drug combinations for treating COVID-19.”
News release, MIT.