[ad_1]
The approach is based on two popular AI techniques: generative accusatory networks, and reinforcement learning.
The news: A team from AI Pharma start-up, Insilico Medicine, collaborating with researchers at the University of Toronto, spent 21 days creating 30,000 models of molecules that target a protein linked to fibrosis (tissue scarring). They then synthesized six of these molecules in the laboratory, then tested two cells, the most promising being tested in mice. The researchers concluded that the protein is potent against the protein and that it has "drug-like" qualities. In total, the process took only 46 days. The search was published in Nature Biotechnology this week.
The method: The system examines previous research and patents for molecules known to act against the drug target, and then prioritizes new structures that can be synthesized in the laboratory. It's similar to what a chemist could do when looking for new treatments – but much faster.
The context: Putting a new drug on the market is extremely costly in time and money: it can take 10 years and cost up to $ 2.6 billion, the vast majority failing the test stage, according to the Tufts Center for the Study of Drug Development. It is therefore not surprising that so much work is being done on artificial intelligence to speed up the process. Deepmind is one of the companies that is exploring pharmaceutical research as a potential future for its algorithms.
A word of warning: Although the research looks promising, it is still a proof of concept. We are far from drugs designed by AI, let alone sold to patients. We explored the question in This article from our TR10 number earlier this year.
S & # 39; register right here for our daily newsletter The Download to receive your dose of the latest news from the world of emerging technologies.
[ad_2]
Source link