Functional Data Explorer Adds a New Dimension to Sequential Experimental Design ( 2019-EU-45MP-078 )
Feb 11, 2019 4:33 PM
| Last Modified: Mar 19, 2019 11:27 AM
Level: Intermediate Job Function: Analyst / Scientist / Engineer Martin Owen, Consultant, Insight by Design
This presentation shows how we have applied Quality by Design principles to assess risk, identify potential failure modes and determine a control strategy. We are doing this to optimize the continuous manufacture of a hot-melt extrusion process to produce medicines. Previously we have shown how sequential experimental design can be used to understand a process and then demonstrate consistency, based on single-point values (e.g., absorbance at a single wavelength). Using Functional Data Explorer, we can now model data over a range of wavelengths (i.e., spectral data) and/or over time. This capacity to model an additional dimension enables us to gain additional insight and understanding by grouping and characterizing similar spectra or time profiles. Different examples of functional data will be explored, showing the impact of observation points that are distant from other observations or data with high variability. We aim to show aspects of pre-processing data and how filters can help zoom in on areas of significance. Continuous processing presents new challenges for pharmaceutical manufacturing, and we consider both "in-line" and "off-line" monitoring techniques in the context of sequential experimental design and monitoring of stabilized processes.