Hybrid modeling framework for process analytical technology: Application toBordetella pertussiscultures

Abstract

In the process analytical technology (PAT) initiative, the application of sensors technology and modeling methods is promoted. The emphasis is on Quality by Design, online monitoring, and closed-loop control with the general aim of building in product quality into manufacturing operations. As a result, online high-throughput process analyzers find increasing application and therewith high amounts of highly correlated data become available online. In this study, an hybrid chemometric/mathematical modeling method is adopted for data analysis, which is shown to be advantageous over the commonly used chemometric techniques in PAT applications. This methodology was applied to the analysis of process data of Bordetella pertussis cultivations, namely online data of near-infrared, (NIR), pH, temperature and dissolved oxygen, and off-line data of biomass, glutamate, and lactate concentrations. The hybrid model structure consisted of macroscopic material balance equations in which the specific reactions rates are modeled by nonlinear partial least square (PLS). This methodology revealed a significant higher statistical confidence in comparison to PLSs, translated in a reduction of mean squared prediction errors (e.g., individual root mean squared prediction errors calibration/validation obtained through the hybrid model for the concentrations of lactate: 0.8699/0.7190 mmol/L; glutamate: 0.6057/0.2917 mmol/L; and biomass: 0.0520/0.0283 OD; and obtained through the PLS model for the concentrations of lactate: 1.3549/1.0087 mmol/L; glutamate: 0.7628/0.3504 mmol/L; and biomass: 0.0949/0.0412 OD). Moreover, the analysis of loadings and scores in the hybrid approach revealed that process features can, as for PLS, be extracted by the hybrid method.

Hybrid semiparametric systems for quantitative sequence-activity modeling of synthetic biological parts

“…PLS is frequently used for process interpretation, namely to identify the most influential input variables for the target output. There is, however, some controversy regarding the efficiency of such analysis ( 36 ) particularly for sparse data sets. For the present study, it would be interesting to identify the sequence positions associated with higher TM residuals.…”

“…For the present problem, there are two main sources of knowledge: (i) a priori knowledge regarding the thermodynamics of the mRNA and ribosome complex formation and (ii) a sequence-activity data set that fully reflects all mechanisms involved in protein translation, many of which still lacking mechanistic interpretation. A key rule in hybrid modeling is that reliable mechanistic knowledge has priority over heuristic or empirical knowledge (35). We have thus adopted a hybrid structure that gives priority to the TM (Figure 1).…”

A hybrid real-time tidal prediction mechanism based on harmonic method and variable structure neural network

“…It has also been discovered that the combination of first principles knowledge and ANNs can better structure the process operation space (Fiedler and Schuppert, 2008), thus generate better calibration properties, better extrapolation capabilities and enhanced statistical confidence of the model estimates comparing to pure ANNs or other data-driven techniques (von Stosch et al, 2011;Ömer Faruk, 2010).…”

An efficient method for the numerical integration of measured variable dependent ordinary differential equations

“…Near InfraRed (NIR)), the application of interpolation techniques might not even be appropriate, because (i) the error introduced by the interpolation technique can be large; and (ii) the number of measured variables is large, wherefore interpolations come with a significant time and computational burden. In order to avoid (repetitive) interpolations, fixed step size methodologies find application in which the integration step-size is determined by the sampling frequency of the measured variables z (Schubert et al, 1994;van Can et al, 1996;von Stosch et al, 2012). However, it is well known that fixed step size methodologies are, likewise, not computationally efficient.…”