Innovation in the fast lane

Innovation in the fast lane

Corning and Merck are two years into joint effort to make best use of data science, one of the hottest fields in research today.

In a world where the pace of innovation is faster than ever, the most successful companies are those able to solve complex business and customer problems quickly.

Data science can help them do just that – and Corning is vastly accelerating the way it puts this powerful set of technology tools to work. A collaboration with Merck (known as MSD outside the U.S. and Canada) has resulted in both companies advancing their mutual capabilities in data science, advanced analytics, and digital platforms.

Two years into their collaboration, Corning and Merck have established seven robust project teams who are harnessing the power of data science to creatively address a variety of critical business issues.

One project workstream, for example, focuses on adding predictive modeling to their arsenal of research techniques. Predictive modeling can give scientists speedy options in understanding how glass chemistry affects physical properties. They can get information about specific scenarios within seconds, as opposed to waiting days for traditional experiment results.

In another collaborative project, Corning and Merck partners are developing new Natural Language Processing (NLP) models. These versatile tools can help researchers quickly access vast amounts of detailed information – whether in text or numerical form – without having to scan full journals, reports, or patent documentation.

“It’s traditionally very tedious work. Using this platform can give us a tremendous advantage,” said Corning data scientist Zhang Liu, who has been part of the collaboration since its early days. Developing the application in-house, he added, is much less costly than using third-party software and helps protect vital information at the same time.

The collaboration brings together about 50 people from both companies. Team members exchange experiences and fresh points of view that can help solve problems in new and efficient ways. Their learnings, the companies believe, will have a positive impact in the next era of research and manufacturing.

Workshops and an annual Data Science Symposium draw hundreds of researchers from Corning and Merck, extending the workstream learnings deep into the companies’ scientific communities.

The most recent symposium -- held virtually on Oct. 25, 26, and 27 -- focused on analytics-ready data and featured keynote talks from two distinguished professors and thought leaders in the data science field.

Dr. Michel Dumontier, from Maastricht University in the Netherlands, emphasized the importance of industry-wide standards in documenting data-driven research. Since 2014, he has led an effort to adopt the FAIR standards digital content – named for nomenclature and practices that make data findable, accessible, interoperable, and reusable. Knowledge structured by these common standards can exponentially accelerate breakthrough discoveries across many fields, he said.

Dr. Michael Stonebraker, adjunct professor at MIT, advised the group to invest in machine learning to do collection, cleaning, and integration of vast amounts of data. This frees data scientists and engineers to focus on analysis and application of their learnings: “Clean data is power.”

Collaboration manager Michelle Pastel of MT&E said the symposium helped further energize data science professionals from both companies – and that bodes well for the momentum of the joint effort.

“We’re creating a larger resource pool of big ideas, and we’re more quickly solving problems we have in common,” she said.

“It links the best parts of both of us, and we’re helping each other advance.”