How AI Is Used To Design And Develop A Drug In 21 Days

A health startup from Hong Kong, in a remarkable development, has used AI (Artificial Intelligence) to design and develop a drug in 21 days. The discovery proved to be potential for fibrosis treatment as well as other diseases as it can help speed up the time it normally takes to design drugs for different incurable illnesses.

The AI system is called GENTRL or generative tensorial reinforcement learning. It only took the system 21 days to come up with six promising fibrosis treatments.

In a paper which was published in the Nature Biotechnology journal, Insilico Medicine is quoted, saying,

The new method, compared with tradition drug discovery, is more promising. The tradition method starts with thousands of small molecules testing just to find some lead-like molecules. Moreover, out of 10 molecules, only one could pass the clinical trials carried out in human patients.

GENTRL, using a similar technique utilize by DeepMind in order to beat human GO players, can quickly create new molecular structures which have specified properties.

The source code of GENTRL has been made available by Insilico as opensource on GitHub, a Microsoft-owned repository.

According to Insilico Medicine’s CEO, Alex Zhavoronkov, these first six molecules’ development is only the beginning.

He added that:

The new technology made used of the 2016 groundbreaking academic research of Insilico about accelerating drug discovery using AI techniques including GAN or Generative Adversarial Networks and RL or Generative Reinforcement Learning.