Development of a chemometric platform for mid-infrared fermentation monitoring

Abstract Industrial bioprocess development still depends heavily on off-line analytics such as HPLC, which, while accurate, are slow and limit real-time decision-making. Process Analytical Technologies (PAT) can overcome this barrier, with mid-infrared spectroscopy (mIR) offering fast, sensitive, and cost-effective monitoring. Yet, most applications rely on complex chemometric tools such as Partial Least Squares (PLS), restricting adoption by non-specialists and limiting systematic evaluation of data sources. Here, we present a streamlined platform for rapid spectral analysis and chemometric model construction to monitor glucose consumption and citramalic acid production during continuous E. coli fermentations. The workflow integrates baseline correction, optimal wavenumber region selection, and correlation of solute concentrations with the area-under-the-curve (AUC), enabling straightforward model generation without advanced statistical expertise. Comparative evaluation across pure solutions, synthetic mixtures, and fermentation broths revealed continuous cultures as the most robust calibration dataset. While the baseline/AUC method showed slightly higher mean absolute error (~ 0.65 g L⁻ 1 ) than PLS, it provided accurate, accessible, and fast predictions consistent with HPLC values. This study demonstrates a practical and user-friendly mIR-based platform/methodology that lowers technical barriers to PAT implementation, accelerating adoption of real-time monitoring in industrial continuous bioprocesses.