A team of researchers from the Massachusetts Institute of Technology (MIT), headed by Professor Alán Aspuru-Guzik has now developed an artificial intelligence program that can recommend a molecular structure that combines properties of medications.
In an article published by Nature World News, the question of what would happen if you cross ibuprofen with aspirin is raised, and with this new AI technology, we might actually be able to answer this question, and many other similar scenarios.
The primary purpose of the program is to find new drug compounds, as researchers have up until now, relied on software that needs to go through “huge pools of candidate molecules with the help of rules crafted by chemists to predict or identify useful structures”, Nature World News writes.
This technology, however, is highly dependent on the assistance of humans, precision of simulations and the required processing power, whereas this new AI is independent and applies deep learning instead of lengthy simulations. The AI will apply its own experience from training machine-learning algorithms with information from being exposed to countless drug-like molecules.
Aspuru-Guzik predicts that this development will also help improving chemists as it explores things intuitively by applying the chemical knowledge required, such as chemists do. The success of the AI seems promising after being trained on 250 000 drug-like molecules and being able to generate conceivable new structures with the help of the knowledge of properties of current drugs and combining them, Nature World News writes.
There is even potential for the software to identify specific properties of molecules, such as solubility, and Aspuru-Guzik promises that the chemical knowledge of the AI will improve once its fed more data.
This software has a large potential, Vijay Pande from Stanford University says, and explains how it can be applied to optimise or discovering drug candidates, or in other fields such as catalysts or even solar cells.
This article was first found at: http://www.natureworldnews.com/articles/31480/20161109/mit-researchers-develop-artificial-intelligence-that-can-search-for-new-drug-compounds.htm