Assessment of the structure and predictive ability of models developed formonitoring key analytes in a submerged fungal bioprocess using near-infrared spectroscopy

The robustness of models developed for the near-infrared spectroscopic pred iction of mycelial biomass, total sugars, and ammonium in a submerged Penic illium chrysogenum bioprocess was assessed by rigorously challenging them w ith artificially introduced analyte and background matrix variations, so th at analyte concentrations were varied in an invariant matrix and vice versa , The models were also challenged by using a data set from a process operat ed at a different scale from that used in the original model formulation. S imple univariate and bivariate linear regression models, and partial least- squares (PLS) models with as few; factors as three and four, performed suff iciently well for predicting analyte concentrations and were robust with re spect to the matrix variations tested, Howe c er, models based on relativel y weaker absorptions, or those that were likely to be influenced by stronge r absorbers present in the same matrix, were vulnerable to changes in the m atrix, ri change in the stale of operation affected models that would he in fluenced by biomass, possibly due to an influence of tile morphology of the mycelial biomass. An analysis of the loading vectors of some PLS models re vealed details that were useful in understanding the type of information mo deled and the behavior of these models to the variations tested.