Biomedical Engineering Reference
In-Depth Information
account for interactions between parameters. Multivariate experiments at small scale can
define a more meaningful space for parameters that may interact (Fig. 2.4c). Such a space
can be generated with an efficient number of experiments using design of experiment
(DOE) approaches [36, 37]. As with univariate small-scale experiments, the validity of the
scale-down and the linkage of measured responses to product performance should be
described. In addition to these considerations, the modeling used in DOE approaches
should be justified.
An important advantage of generating a multidimensional design space is described
in Fig. 2.5. An empirically derived manufacturing process is static with locked-in
parameters (Fig. 2.5a). Any variability in process inputs is transferred to the product
Figure
2.5. A design space allows for a dynamic approach to manufacturing that transfers
variability from important product attributes to process parameters. (a) The traditional para-
digm for pharmaceutical manufacturing utilizes a fixed process. Thus, any variability in inputs
may result in variable product. (b) In a dynamic manufacturing process, input variability can be
monitored either directly or through product impact. Information regarding this variability can
then be used to adjust the process parameters to compensate and produce high-quality product.
Aknowledge-richdesign space that has been exploredduringdevelopment and throughout the
product life cycle will allow the process flexibility necessary to compensate for variable process
inputs. This figure was adapted from a diagram by Jon Clark.
