No abstract

An integrated online multivariate statistical process monitoring (MSPM), quality prediction, and fault diagnosis framework is developed for batch processes. Batch data from I batches, with J process variables measured at K time points generate a three-way array of size I × K × J. Unfolding this three-way array into a two-way matrix of size IK × J by preserving the variable direction is advantageous for developing online MSPM methods because it does not require estimation of future portions of new batches. Two different multiway partial least squares (MPLS) models are developed. The first model (MPLSV) is developed between the data matrix (IK × J) and the local batch time (or an indicator variable) for online MSPM. The second model (MPLSB) is developed between the rearranged data matrix in the batch direction (I × KJ) and the final quality matrix for online prediction of end-of-batch quality. The problem of discontinuity in process variable measurements due to operation switching (or moving to a different phase) that causes problems in alignment and modeling is addressed. Control limits on variable contribution plots are used to improve fault diagnosis capabilities of the MSPM framework. Case studies from a simulated fed-batch penicillin fermentation illustrate the implementation of the methodology.

The manufacture of biotherapeutic proteins consists of complex upstream unit operations requiring multiple raw materials, analytical techniques, and control strategies to produce safe and consistent products for patients. Raman spectroscopy is a ubiquitous multipurpose analytical technique in biopharmaceutical manufacturing for real-time predictions of critical parameters in cell culture processes. The accuracy of Raman spectroscopy relies on chemometric models that need to be carefully calibrated. The existing calibration procedure is nontrivial to implement as it necessitates executing multiple carefully designed experiments for generating relevant calibration sets. Further, existing procedure yields calibration models that are reliable only in operating conditions they were calibrated in. This creates a unique challenge in clinical manufacturing where products have limited production history. In this paper, a novel machine-learning procedure based on just-in-time learning (JITL) is proposed to calibrate Raman models. Unlike traditional techniques, JITL-based generic Raman models can be reliably used for different modalities, cell lines, culture media, and operating conditions. The accuracy of JITL-based generic models is demonstrated on several validation studies involving real-time predictions of critical cell culture performance parameters, such as glucose, glutamate, glutamine, ammonium, lactate, sodium, calcium, viability, and viable cell density. The proposed JITL framework introduces a paradigm shift in the way industrial Raman models are calibrated, which to the best of authors' knowledge have not been done before. K E Y W O R D Sbiopharmaceutical manufacturing, generic models, machine-learning, Raman spectroscopy

Raman spectroscopy is a multipurpose analytical technology that has found great utility in real‐time monitoring and control of critical performance parameters of cell culture processes. As a process analytical technology (PAT) tool, the performance of Raman spectroscopy relies on chemometric models that correlate Raman signals to the parameters of interest. The current calibration techniques yield highly specific models that are reliable only on the operating conditions they are calibrated in. Furthermore, once models are calibrated, it is typical for the model performance to degrade over time due to various recipe changes, raw material variability, and process drifts. Maintaining the performance of industrial Raman models is further complicated due to the lack of a systematic approach to assessing the performance of Raman models. In this article, we propose a real‐time just‐in‐time learning (RT‐JITL) framework for automatic calibration, assessment, and maintenance of industrial Raman models. Unlike traditional models, RT‐JITL calibrates generic models that can be reliably deployed in cell culture experiments involving different modalities, cell lines, media compositions, and operating conditions. RT‐JITL is a first fully integrated and fully autonomous platform offering a self‐learning approach for calibrating and maintaining industrial Raman models. The efficacy of RT‐JITL is demonstrated on experimental studies involving real‐time predictions of various cell culture performance parameters, such as metabolite concentrations, viability, and viable cell density. RT‐JITL framework introduces a paradigm shift in the way industrial Raman models are calibrated, assessed, and maintained, which to the best of authors' knowledge, have not been done before.

A morphologically structured model is proposed to describe penicillin production in fed-batch cultivations. The model accounts for the effects of dissolved oxygen on cell growth and penicillin production and variations in volume fractions of abiotic and biotic phases due to biomass formation. Penicillin production is considered to occur in the subapical hyphal cell compartment and to be affected by availability of glucose and oxygen. As it stands, the model provides a wide range of applicability in terms of operating conditions. The model has been tested for various conditions and has given satisfactory results. A series of glucose feeding profiles have been considered to demonstrate the capabilities of the proposed model. It is concluded that the model may be valuable for the interpretation of experimental data collected specifically for metabolic flux analysis during fed-batch cultivation because the elements of measured specific production rates are determined from measurements of the concentrations of the components and their mass balances. The proposed model may be further used for developing control strategies and model order reduction algorithms.