RoboRXN: IBM’s drug-making lab powered by AI and cloud technology

Hyperaxion Aug 29, 2020

Created by IBM, the RoboRXN system will optimize research and production times for new chemical substances.

The production of drugs and materials that require chemical tests is very complex. On average, it takes at least 10 years to discover and market a new item, and estimated production costs are around US$10 million.

However, the technology company IBM recently launched the RoboRXN system, which aims to optimize research and production times for new drugs, allowing scientists to run chemistry experiments remotely.

RoboRXN: IBM's drug-making lab powered by AI and cloud technology
(Credit: IBM).

The system combines artificial intelligence (AI), cloud technology, and automation to produce molecules in a laboratory with as little human intervention as possible.

According to Teodoro Laino, a researcher at the company, the objective of the new product is to facilitate scientific research and reduce the production time for new substances.

During social distancing, for example, researchers would not need to leave their homes to proceed with their studies, since, thanks to cloud technology, it is possible to use the system remotely.

It all started three years ago, when the company started developing RXN for Chemistry, an online artificial intelligence service that helps scientists predict results of chemical reactions.

It uses a machine learning algorithm, converting reagents and reactants into products. Prediction accuracy increases as it is fed with more data.

Based on the same approach, RoboRXN is able to provide general guidelines for chemical procedures, as well as the correct steps for each procedure.

(Credit: IBM).

According to a note from IBM, the challenge was to design “an algorithm that specifically extracts the synthesis information for organic chemistry and converts it into a structured and automation-friendly format.”

The system is purely data-driven. The main advantage of this approach is that, to improve the production of any product, it is only necessary to feed the machine learning model with more data.


Leave a Reply

Notify of