Chimica Oggi-Chemistry Today
Agro FOOD Industry Hi Tech
p. 38-43 / PAT/QbD
Model maintenance: the unrecognized cost in PAT and QbD
*Corresponding author
Eigenvector Research, Inc.,
Wenatchee, WA 98801, USA

KEYWORDS: Multivariate calibration, regression modeling, model updating.
ABSTRACT: Multivariate calibration, classification and fault detection models are ubiquitous in QbD (Quality by Design) and PAC and PAT (Process Analytical Chemistry and Technology, respectively) applications. They occur in both the development of processes and their permissible operating limits, (i.e. models for relating the process design space to product quality), and in manufacturing (i.e. models used in monitoring and control). Model maintenance is the on-going servicing of these multivariate models in order to preserve their predictive and diagnostic abilities. It is required because of changes to either the sample matrices or the instrument response. The goal of model maintenance is to sustain or improve models over time and react to changing conditions with the least amount of cost and effort. A model maintenance roadmap is presented. It includes procedures for determining when model maintenance is required, the probable source of the model/data mismatch, and the best approaches for bringing model performance back to acceptable levels.

Model maintenance can be roughly defined as the on-going upkeep of (primarily) multivariate calibration and fault detection models in order to preserve their predictive and diagnostic abilities. The goal of model maintenance is to preserve or improve models over time and address changing conditions with the least amount of cost and effort. As noted in ASTM E2891-13 (1), model maintenance should be considered a core activity of the overall modelling activity in pharmaceutical and manufacturing operations. Regrettably, in spite of this, model maintenance is often not planned for in advance in QbD and PAT applications. Instead, it is frequently assumed that models will work indefinitely. However, it is an unfortunate fact that seemingly innocuous changes to sample matrices and/or instrument performance can sometimes significantly degrade the performance of multivariate models. The lack of a plan to deal with this when it occurs exacerbates the problem.

This document outlines the reasons model maintenance is required and some approaches for dealing with it. A roadmap is presented which can serve as a template for developing a more detailed plan for specific applications.

Why is model updating necessary?
The reasons that calibration models need updating can be roughly divided into two cases. The first is when the calibration set needs to be expanded in range or dimension. In these cases, nothing has changed with the response of the instrument to specific analytes. Movement of the data to a range beyond the original calibration data, (i.e. when analyte concentrations move outside their previous extremes), requires extrapolation from the model, which may lead to less accurate predictions. A more serious problem is the addition of new analytes or other previously unseen variations which expand the dimension of the data and can make the original models biased. In order for multivariate models to ignore irrelevant variation, they must have samples exhibiting this variation in the calibration data. Thus, when the data range is expanded or new variations are added, new samples must be added to the calibration data which exhibit these variations.

The second case is when the samples...In order to continue reading this article please register to our website – registration is for free and no fees will be applied afterwards to download contents.

Teknoscienza Mobile
Teknoscienze - Viale Brianza 22, Milano - ITALY - P.IVA 06817720151
Teknoscienze S.r.l. © 2012 - tutti i diritti riservati - web terms & conditions - privacy - Credits