Machine Learning Could Speed Drug Development

Google’s DeepMind, the artificial intelligence arm of Alphabet Inc., has shown promise in predicting the shapes of proteins, the building blocks of disease. A recent Bloomberg report notes that this could lead to more rapid development of drugs.

DeepMind beat experienced biologists from 50 of the top labs around the world in predicting the shapes of proteins — an incredibly complex process required to find ways for medicines to attack disease. DeepMind used the latest neural-network algorithms and its limited experience in protein folding, the process by which proteins get their 3D shape, to deliver its results.

Biologists who took part in the program were stunned by the results. Still, DeepMind’s simulation doesn’t yet produce atomic-level resolution that is important for drug discovery, and it may be years before it’s known whether this type of software can consistently spot promising therapies that researchers might have otherwise missed, the report explains.