Rate-Based Information Processing Approaches for Bioprocess Development Along QbD Principles

Patrick Sagmeister                  patrick.sagmeister@tuwien.ac.at

Patrick Wechselberger            patrick.wechselberger@tuwien.ac.at

Christoph Herwig                    christoph.herwig@tuwien.ac.at

Details on the topics presented can be found in following contributions:

Sagmeister P, Wechselberger P, Herwig C. 2012. Information Processing: Rate-Based Investigation of Cell Physiological Changes along Design Space Development. Pda J. Pharm. Sci. Technol. 66:526–541.

Sagmeister P, Wechselberger P, Jazini M, Meitz A, Langemann T. 2013. Soft sensor assisted dynamic bioprocess control: Efficient tools for bioprocess development. Chemical Engineering Science. 96: 190-198.

Wechselberger P, Sagmeister P, Herwig C. 2012. Real-time estimation of biomass and specific growth rate in physiologically variable recombinant fed-batch processes. Bioprocess Biosyst. Eng. (online first)

Wechselberger P, Sagmeister P, Herwig C. 2013. Model-based analysis on the extractability of information from data in dynamic fed-batch experiments. Biotechnol. Prog. 29:285–296.

Abstract

Regulatory initiatives summarized under the term quality by design (QbD) promote science and risk-based pharmaceutical bioprocess development strategies. During upstream process development as well as manufacturing a plethora of on- and off-line data is collected but typically not efficiently exploited and used for the demonstration of process understanding. Here we want to present recent developments in the field of data processing (quality check and condense your data into information) as well as multivariate information processing (extract knowledge from your information) of fermentation data. On the basis of the presented concepts a novel roadmap for data- and information processing of fermentation data is presented.

1)      Soft sensing- Real-time sensing of physiological information

Soft sensors are process analytical devices that use readily available on-line signals to calculate important process descriptors and hence grant on-line information on the state of the process. Next to on-line applications, first principle soft sensors can be run on historical data sets to extract non-measured variables such as the biomass concentration, specific rates or yield coefficients that stand in connection to cell physiology. Applications, limits and practical aspects of the use of soft-sensors for on- and offline use are demonstrated.

2)      Information quality control: Reconciliation as a statistical tool to ensure bioprocess information quality

Data Reconciliation refer to statistical approaches to remove measurement error out of data through the application of defined constrains. Little prior knowledge is required and the signal to noise ration of the extracted information can be greatly enhanced. Furthermore, a statistical test can be applied to detect gross measurement errors. This is especially useful for further information processing approaches (e.g. multivariate data analysis), since outliners can be efficiently detected and removed on a sound scientific basis. Applications, limits and practical aspects of the use of data reconciliation for data quality control are discussed.

3)      Information Processing: Multivariate rate-based investigation of cell physiology

The application of multivariate techniques on fermentation data is still a challenging task. The use of processed data (specific rates and yield coefficients) instead of raw data for multivariate data analysis is highly promising, since variables have a direct physiological interpretation and the number of variables can be reduced prior to multivariate data analysis. The general concept of rate based multivariate data analysis is presented and applications are discussed.

4)      A rate based data to knowledge roadmap for upstream bioprocess development

The tools mentioned above are combined to a novel general concept for the a) mining of data and  information, b) rate based processing, c) statistical assessment as well as d) multivariate processing of information. The presented roadmap for process knowledge extraction is complementary to process development and continuous improvement along QbD principles.