Your AI applications are only as strong as your data. This is why Elsevier strives to make data not only align with FAIR usability principles but also be maximized for trustworthiness and relevance. In this way we can confidently offer the full potential of data science in an easy-to-use manner – from data acquisition and preparation, to AI application and interpretation.
We believe the power of data should be
for everyone. And that’s why Elsevier is applying its long experience in
curating and deep indexing scientific information to provide Data-As-A-Service
(DaaS) to help researchers turn data into actionable insights.
Ted Slater is Elsevier’s Senior Director for Product Management PaaS. As one of the original authors of the seminal paper from 2016 that set out the guiding principles for FAIR data, he can also be considered a data evangelist. “Apps and algorithms will come and go,” says Ted. “But quality data is forever – and for that reason it should be treated as a first-class citizen.”
“Elsevier really has a hilarious amount
of content. In most cases, more data means more confusion. But by following
FAIR, we are at least making this vast amount of data easy-to-use,” says Ted. “However,
R&D people and organizations need more than that: they need the data to be trustworthy
and relevant – and here again, Elsevier can bring a lot to the table.”
Data’s Holy Trinity:
1. Useable: sharing the love for the ages
Elsevier is committed to FAIR data. By following the principles of Findability, Accessibility, Interoperability, and Reusability, we aim to optimize the reuse of data. After all, if you are going to go through all the effort of cleaning up and preparing a data set, you may as well do it for posterity (or even just for a bump in the road further down your own pipeline). And by offering datasets that are already FAIR, they are interoperable with each other and any other any other additional FAIR data sources that come along (including your own).
2. Trustworthy: quantity + quality
Elsevier has been a trusted gatherer of content for over 140 years – and of digital data for over 20 years. We have access to a vast amount of the world’s published scientific literature. After all, you need a lot of trusted quality data to support analyses and build more precise and trustworthy models – ones that don’t generate bias. And yes, since we know how to take care of our own sensitive data, we also know how to take care of yours.
3. Relevant: finding what you are looking for
But collecting data remains a random affair unless you
curate it. “If you are fishing for an answer related to cancer, it’s not likely
you will find it in a data set covering the geothermal properties of Iceland,”
notes Ted. “You want to know you have a fairly good chance that a particular
data set has the answer you are looking for.” And that’s why Elsevier works on provisioning FAIR data sets by therapeutic areas.
While we’ve always had a small army of very smart people poring over and
collating the latest scientific information, we are increasingly applying
machine learning methods to support this small army. It means we can now collate
relevant info from 16,000 periodicals as opposed to the 350 we covered just a
few years ago.
Accelerating R&D with AI
This holy trinity of data comes together in a range of Elsevier products and services, such as our next-level chemical database Reaxys. It certainly acts as the guiding force and strength of our award-winning cloud-based platform Entellect for data-driven R&D in the pharmaceutical industry. Using clear, vetted and contextualized data, the customizable platform can apply predictive and exploratory analytics for a whole range of workflows, including drug repurposing, preclinical drug safety and target scoring.
“There are only winners when you make
first-class data easier to use,” says Ted. “It goes beyond plug-and-play. It’s
about plug-and-let’s-make-the-world-a-healthier-place.”
From initial target selection to staying
attuned to the changing marketplace, Elsevier is ready to be a partner.
