Using artificial intelligence to find drugs

[zakiyatasnim]

In December 2019, researchers at Drexel University College of Medicine tried to find out whether a cream with the immunosuppressant rapamycin could slow the aging of human skin.

Experiments are becoming more and more common, which means that researchers are making more and more efforts to explain whether it is possible to "break" diseases associated with aging and slow their onset - from cancer and arthritis to dementia and heart disease.


Using artificial intelligence, scientists are discovering promising hospital compounds.

Why it's important: The cost of developing a mexico number data new drug starts at $2.5 billion. One reason is the difficulty of finding promising molecules.
Key players: Insilico Medicine, Kebotix, Atomwise, University of Toronto, BenevolentAI, Vector Institute.
Availability: 3-5 years.
Researchers estimate that there are about one decilion molecules that could be used to create drugs. That's more than all the atoms in the solar system, meaning there are virtually limitless possibilities for chemists - they just have to find the right ones.

Chemists are helped by machine learning tools that analyze databases of existing molecules and their properties and suggest new combinations. Thus, machine learning speeds up and reduces the cost of finding new drug candidate molecules.

In September, a team of researchers from Hong Kong-based company Insilico Medicine and the University of Toronto demonstrated the work of machine learning technologies - they synthesized several potential drugs that were found using algorithms.

Using deep learning and generative models, the researchers identified about 30,000 new molecules with desirable properties and selected six for synthesis and testing. One of them showed promise after successful animal tests.

Read the continuation of the article in the second part .