Different modeling approaches for inline biochemical monitoring over the VLP-making upstream stages using Raman spectroscopy.

[Display omitted] • Monitoring of Rabies virus-like particles production using online Raman spectra. • Nonlinear models showed a better fit than linear ones in biochemical monitoring. • The models can be used in rabies virus-like particle commercial applications. • Variables absolute error was like those for equivalent offline work. • Model accuracies were alike chemometrics models for other cells hosts. This work aimed to set inline Raman spectroscopy models to monitor biochemically (viable cell density, cell viability, glucose, lactate, glutamine, glutamate, and ammonium) all upstream stages of a virus-like particle-making process. Linear (Partial least squares, PLS; Principal components regression, PCR) and nonlinear (Artificial neural networks, ANN; supported vector machine, SVM) modeling approaches were assessed. The nonlinear models, ANN and SVM, were the more suitable models with the lowest absolute errors. The mean absolute error of the best models within the assessed parameter ranges for viable cell density (0.01–8.83 × 106 cells/mL), cell viability (1.3–100.0 %), glucose (5.22–10.93 g/L), lactate (18.6–152.7 mg/L), glutamine (158–1761 mg/L), glutamate (807.6–2159.7 mg/L), and ammonium (62.8–117.8 mg/L) were 1.55 ± 1.37 × 106 cells/mL (ANN), 5.01 ± 4.93 % (ANN), 0.27 ± 0.22 g/L (SVM), 4.7 ± 2.6 mg/L (SVM), 51 ± 49 mg/L (ANN), 57 ± 39 mg/L (SVM) and 2.0 ± 1.8 mg/L (ANN), respectively. The errors achieved, and best-fitted models were like those for the same bioprocess using offline data and others, which utilized inline spectra for mammalian cell lines as a host. [ABSTRACT FROM AUTHOR]