Biomedical Engineering Reference
In-Depth Information
and gases. These analytical devices are being combined with statistical techniques made
available by the discipline of chemometrics to gain new insights and better control of
these complex biological processes. Other tried and true techniques are also increasingly
used to gain process understanding. An assortment of chromatographic techniques such
as gas and liquid chromatography are being coupled with a wide array of available
detectors such as UV, fluorescence, refractive index, light scatter, charged aerosol
detection, and mass spectrometry. Other nonsolid phase separation techniques are seeing
increasing use due to the rapidity of their measurements such as flow injection analysis.
Off-line spectroscopic techniques such as NMR, while still extremely useful, may
benefit from the coupling of these devices to some of the latest generation of sterile
sampling devices in combination with HPLC.
Data manipulation techniques such as signal preprocessing are frequently neces-
sary to reduce background noise and optimize spectroscopic data for use in correlation
building. The “scrubbed” data are then combined with other critical process parameter
data (mechanical, chemical, and biochemical) collected by modern control systems
such as dissolved oxygen, pH, sparge rate, nutrient attributes such as carbon source,
vitamins, and certain beneficial trace elements. Often, detrimental metabolic indicators
such as ammonia or acetate levels may be included in the data set to determine if buildup
of the undesirable by-products should be more tightly regulated. The combined data set
is then analyzed by using multivariate techniques such as principal components analysis
(PCA), discriminant and cluster analysis, and partial least squares (PLS) analysis to
understand the interdependencies of critical process parameters and potentially their
effects on critical quality attributes. Multivariate exploration with correlations and
cluster analysis enables looking at many variables at the same time. It is simply a set of
statistical tools to look at continuous variables when they are considered as responses
with no factors or independent variables. Used within the framework of continuous
improvement, process analytical tools and multivariate techniques fit in well with
the process improvement principles of six sigma and lean manufacturing and are
enabling process scientists and engineers to work toward better defined and better
controlled processes.
Relatively new fields such as surface plasmon resonance (SPR), microcapillary
and nanofluidic arrays, and neutron reflectometry hold the potential to push the
boundaries of process understanding further. Even tools not necessarily designed for
bioprocess monitoring, such as hydrogen sensors or silicon dioxide monitors, are being
applied to biologics processes and may lead to deeper understanding of biological
processes.
Clearly, with many of these techniques, the ability to deconvolute the data and
present them in a manner that is easily understandable in a manufacturing environment is
critical to using these technologies. Integration of computerized laboratory systems is
one of the components necessary to achieve real-time model building and automated
process control systems. Supervisory Control and Data Acquisition (SCADA) systems,
programmable logic controllers (PLCs), laboratory information management system,
data warehouses, and adaptive multivariate analysis process tools are critical to
achieving the type of control found in other industries such as the petrochemical industry
or the food processing industry.
