Abstract

The Digital Twin: Toward Bioreactor Modeling Using Deep Analytics

Paul Goodwin, Science Director, GE Healthcare

A bioreactor is vessel for large-­scale biological reactions. GE Healthcare, Life Sciences is a major supplier of bioreactors for culturing of cells and especially mammalian cells. The bioreactor must provide for the complete nourishment, gas exchange, temperature control, and homeostasis that would normally be provided by a complete organism. Most current bioreactors provide for relatively basic sensing and control of the cell culture medium. Common sensors include temperature, pH, and dissolved oxygen and most control systems regulate temperature, pH and gas exchange through PID (Proportional-­Integral-­Derivative) control. A number of studies have demonstrated that improved cell viability and production can be achieved by more thorough sensing and modeling of the bioreactor system. Improvements in sensors, complex system modeling, and control technology create an opportunity for improved bioreactors capable of delivering consistent, predictable results even in complex culturing conditions like the autologous cell cultures required for cell-­based therapies such as immunotherapy and regenerative medicine. To this end, GE is developing new modeling methods to understand not only average batch regulation but complete modeling of individual. This modeling of every batch we refer to as the Digital Twin. By creating a Digital Twin of the bioreactor we will be able to improve both adaptive and predictive control of processes through in silico modeling to deliver improved outcomes for our customers.