For the first time, humans will Review a drug created using AI

[ad_1]

The process of creating a drug is long and expensive. Almost 15 years are necessary for a molecule to become a "real medicine" according to the National Institute of Health and Medical Research (Inserm). What if artificial intelligence was the solution to reduce this duration? On January 30, 2020, the British start-up Exscientia announced the start of clinical trials of the first drug "created" using artificial intelligence (AI) in just 12 months.

Working in partnership with the Japanese company Sumitomo Dainippon Pharma, the start-up has therefore considerably reduced the time required to develop a new treatment and start phase 1 trials (human Review). Addressing the BBC, the general manager of Exscientia, Professor Andrew Hopkins said that "if AI was used to diagnose patients and analyze data and scanners, this is the first direct use of AI in the creation of a new drug".

First tests at Japan

Called "DSP-1181", this molecule aims to treat people affected by compulsive behavior disorders (OCD), that is to say repetitive and unreasonable but irrepressible behaviors. The first clinical trials will take place in Japan, before being extended if the tests prove positive. The goal of phase 1 is to observe the evolution of the molecule in the body as a function of time and its toxicity.

The algorithms created by Exscientia have generated tens of millions of different compositions of the drug. The programs then sorted out all of these combinations to keep only the one having a beneficial effect on the brain receptor known since 2016. "There are billions of decisions to be made to find the right molecules, which makes it very difficult to design a drug"said Professor Hopkins.

The potential of AI for medical research

It is not the first time that the potential of AI has been raised for the manufacture of new drugs. In early September 2019, the American biotechnology start-up Insilico Medicine in collaboration with researchers from the University of Toronto successfully identified a potential treatment for fibrosis in just 46 days.

More specifically, 21 days were devoted to the creation of 30,000 models of molecules. A deep learning model then analyzed previous research work and patents concerning molecules known to act against the protein by giving priority to new structures that can be synthesized in the laboratory. Then six molecules were synthesized in the laboratory and two of them were injected into stem cells. The most effective molecule was then tested on a mouse.

Exscientia is already looking into other drugs to cure cancer or cardiovascular disease. In the near future, the goal would be to use AI almost systematically to reduce the time and cost of designing a drug. In fact, the computing capacities of the AI ​​make it possible to process a quantity of data unreachable by Man. For Professor Hopkins, "by the end of the decade, all new drugs could potentially be created by AI".

[ad_2]