Towards complete automation of mammalian cell culture perfusion processes

We have developed a framework that is capable of providing complete automation of mammalian cell perfusion processes. This framework includes an expert system that provides low-level control of the process based on data obtained from online and offline sensors and maintains system integrity by checking for failure and taking the necessary corrective action. In addition, it can also make several high level decisions such as switching operating modes during the perfusion process based on an extensive knowledge database of the process. The capabilities of the expert system can be considerably enhanced through the generation or high quality process data in real time. One such approach we have taken is the real-time computation of specific rates and metabolic fluxes resulting in a physiological state vector of high information content. We have also developed an in-situ microscope for reliable in-situ monitoring of the cell density, one of the most critical parameters for perfusion process control. These advances enable advanced and intelligent control of the perfusion process. The other component of process automation includes applications such as bioreactor sampling and subsequent sample manipulations. To achieve this objective, a prototype of a robotic system was developed that is able to draw samples from the bioreactor and perform certain offline operations with the sample.