Data Backup Digest

Launched in 2013, Facebook AI has been spearheading research and development into the field of AI for quite some time. Their work has paid off in numerous ways, but the latest headlines might be the best yet.
According to the recent news, the team with Facebook has joined forces with the German Research Center for Environmental Health to pioneer a brand new, AI-driven tool aimed at analyzing data and better predicting new treatments for disease.

More specifically, it's focused on accelerating drug discovery – a traditional medical process that involves testing new drug combinations that may have the potential of fighting off advanced diseases – which often involves a lot of monotonous work on the part of scientists and medical researchers. Thanks to their highly advanced AI platform, however, things just got a whole lot easier.

Discovering New Drug Cocktails

Often referred to as drug cocktails, these treatments are designed to target the most aggressive diseases and cancers appearing in patients today. Thanks to their new tool, known as the Compositional Perturbation Autoencoder, or CPA for short, researchers will be able to model dosages, timings, and other advanced forms of treatment.

Dr. Fabian Theis, director of the Institute of Computational Biology with the German Research Center for Environmental Health, explained the need for such technology in a recent interview by saying: "Our field has been successful in putting together cell atlases for different organs. This search space – across cell types, drug combinations as well as patient variation – is incredibly large, and can never be explored in full experimentally, so machine learning is crucially needed here."''

An Open-Source Project

Available immediately, the CPA uses historical observation data about past drug combinations and their effects on cell types. This lets the tool predict drug behavior on a molecular level while using next-gen self-supervision protocols to observe cells that have been treated with various drug combinations, ultimately letting the tool formulate a prediction regarding the potential effects. In short, it helps researchers narrow down the billions of potential drug combinations – via various simulations – in a matter of hours instead of days, weeks, or even years.

But perhaps the best news of all is the fact that the CPA happens to be an open-source project. It was released with a handy API, complete with a Python package for use by developers and those who are interested in contributing to the project. Details of their work are expected to be published in medical journals within the near future.

A recent blog, written by Facebook program manager Anna Klimovskaia and her partner, research scientist David Lopez-Paz, described the team's ultimate goal by stating: "Our hope is that pharmaceutical and academic researchers as well as biologists will utilize [CPA] to accelerate the process of identifying optimal combinations of drugs for various diseases. In the future, [CPA] could not only speed up drug repurposing research, but also — one day — make treatments much more personalized and tailored to individual cell responses, one of the most active challenges in the future of medicine to date."