Google AI Challenges Big Pharma in Drug Discovery

Google AI Challenges Big Pharma in Drug Discovery

Alphabet’s subsidiary DeepMind (the company that unites Google’s businesses) has announced that it has made significant progress in artificial intelligence for drug development. It can shorten the process from months to days.

The artificial intelligence of the company has solved the problem with the so-called “protein folding” that scientists have been trying to figure out for about 50 years. The structure of proteins is key to determining their properties and plays an important role in making the right treatment.

[ READ: Google pays 50% of a deceased employee’s salary to a spouse for ten years ]

It is very difficult to predict these folds, and in traditional scientific methods, the process takes even years. Subsequently, the Folding@Home project emerged, through which volunteers from around the world share some of the system resources on their computers to perform the calculations. Thus, the process is shortened from years to months.

DeepMind’s new AlphaFold development manages to shorten the work to a few days. The company says that artificial intelligence can correctly predict the structure of proteins, with an accuracy of atomic width.

The algorithm can constantly self-learn and improve its accuracy, using what has been learned so far and the real results it has achieved. Thus, the accuracy of its forecasts is constantly increasing.

Understanding the process of protein folding will play a critical role in medicine. This will allow scientists to find treatments faster for many more diseases. It will also enable them to understand more quickly how diseases are spread and transmitted and what to do to alleviate conditions such as allergies and more. It will also be possible to prevent “folding errors” that can lead to various health problems.

All of these are opportunities to learn about the process. DeepMind’s algorithm will help by shortening the time and thus allowing scientists to test faster and more accurately many more options. The algorithm needs to be further refined, currently scoring 87 points out of the required 90, and scientists will be able to accept the data as “competitive” with the information from traditional experiments.

DeepMind will continue working on the project, hoping to achieve even greater accuracy in the near future. During this time, more possible applications of development will be sought, what else can it be useful for except the development of drugs, since there are many proteins in the body that are still relatively unknown to scientists.

Sign up for our newsletter to get the best of The Sized delivered to your inbox daily.