Figure 2: Typical approaches how
to conduct a biopharmaceutical process shown for a chromatographic step
exemplarily. It can be either (A) Based on Fixed time / volume for
loading and elution; (B) Threshold based e.g. stop loading or start
collecting the eluate at a certain UV absorbance; (C) Model based online
monitoring with sensors implemented before and / or after the column to
predict product quality and quantity in feed and / or eluate; (D) Model
predictive control. While (A) and (B) rely on time and material offline
analysis and do not allow in process control, (C) and (D) do need
offline analysis only for the model training and enable real-time
decision making and real-time release.
In this review, we will elaborate how we can arrive at a system for
monitoring and control to run a bioprocess, autonomously; we will mainly
on downstream processing. An ideal bioprocess from the point of view of
real-time monitoring, control and real-time release is outlined.