The COVID-19 pandemic unveiled the crucial need for fast methods of creating and producing new vaccines, but GSK was already on the case. In 2019—before the world knew about SARS-CoV-2—GSK teamed up with Atos and Siemens to create a digital twin of vaccine development and manufacturing. This virtual replica, said Sandrine Dessoy, digital innovation lead at GSK, is “a dynamic tool—a simulation of the vaccine process.”
This digital twin can be used in two ways: offline indeed as a simulator of the vaccine process or online to monitor and control a vaccine process based on data from sensors. Dessoy described the online use as “a really new way to get advanced control of the process.” To accomplish that, Dessoy and her colleagues must build digital twins for each step in making a vaccine and combine them to simulate the entire production line.
In addition to the many parts of producing a vaccine, building a digital twin involves many parts of GSK.
“It’s really the result of the collaboration between IT and business and science,” said Loredana Vagaggini, senior product owner in the vaccine tech R&D team at GSK. “So, we really had the mindset of supporting the business, and we tried to put in place the best architecture we could to connect all the different components of digital twins. You can imagine there are tons of systems we need to connect from process analytical technologies to read the data from the sensors.”
Perhaps surprisingly, the urge to use too much technology created one challenge. “One of the complexities is finding a balance between introducing technology and not overwhelming with too much technology,” Vagaggini explained. “Plus, you need to find a way to support digital twins and to allow them to be easily moved from the research labs to manufacturing.”
GSK’s digital twin already makes a difference. “We see the benefits in development already, because we want to bring a model that is a standard way of working,” Dessoy said. “We want the scientists to first work on the computer with simulation and then take the best process to the lab to check the production. We are not completely there, but we started by introducing this model and we are getting really great feedback from our people.”
