Multivariate statistical process control of an industrial-scale fed-batch simulator

Multivariate statistical data analysis of cell‐free protein synthesis toward monitoring and control

“…x, and its reconstructed value obtained using the MSDA model. The confidence limits for the charts were calculated using the bootstrapresampling technique used in Duran-Villalobos et al, 16 which infers confidence intervals from an empirical distribution function.…”

“…MSDA techniques have been applied successfully to pharmaceutical production processes for optimization, monitoring, online control and detection of sensor faults in seed, batch, and fed-batch cultivations. 10,[13][14][15][16][17][18][19][20][21] For example, the capabilities of MSDA in a cell-culture process for small scale (2 L) and large scale (2000 L) batches were published in Kirdar et al 22 This work aimed to evaluate whether MSDA was able to characterize the process through the analysis of process parameters such as CO 2 , O 2 , glucose, pH, lactate, ammonium ions, purity, viable cell density, viability, and osmolality. The proposed methodology included analyzing various control charts to identify fault conditions, with their results demonstrating that MSDA could be used as a tool for extracting knowledge from such processes.…”

“…An alternative approach to modeling multidimensional data is to use multivariate statistical models, such as those commonly referred to as multivariate statistical data analysis (MSDA). MSDA techniques have been applied successfully to pharmaceutical production processes for optimization, monitoring, online control and detection of sensor faults in seed, batch, and fed‐batch cultivations 10,13‐21 . For example, the capabilities of MSDA in a cell‐culture process for small scale (2 L) and large scale (2000 L) batches were published in Kirdar et al 22 This work aimed to evaluate whether MSDA was able to characterize the process through the analysis of process parameters such as CO 2 , O 2 , glucose, pH, lactate, ammonium ions, purity, viable cell density, viability, and osmolality.…”

Combining model structure identification and hybrid modelling for photo‐production process predictive simulation and optimisation

“…The nonlinearity of this mismatch is greatly reduced compared to the original bioprocess, as the general process behaviour has been described by the kinetic model. For example, a number of previous studies have shown that after subtracting the mechanistic process trajectory, even a simple data‐driven model, for example, partial least squares (PLS) which is mainly used for linear systems, can well capture the processes physically undetermined behaviour (do Carmo Nicoletti & Jain, 2009; Duran‐Villalobos, Goldrick, & Lennox, 2020). As a result, it is not necessary to embed sophisticated data‐driven models such as ANNs into a hybrid model.…”

History and Evolution of Modeling in Biotechnology: Modeling & Simulation, Application and Hardware Performance

“…The authors reported an almost 5-fold increase of biomass yield in <20 batches [87] . In 2020 the same authors applied their previously developed multivariate statistical process model to “a realistic industrial-scale fed-batch penicillin simulator” for batch to batch optimization of penicillin yield by adjusting the glucose feed rate [88] .…”