Answers are grounded in this paper and cite their sources.
Background Degree of hydrolysis (DH) is a critical control parameter in whey protein hydrolysate (WPH) production because it strongly influences nutritional and functional properties. Conventional DH reference methods are accurate but relatively slow and labour‐intensive, limiting near real‐time process monitoring. Fourier‐transform infrared (FT‐IR) spectroscopy has shown promise for protein characterisation, but robust industrial‐scale, at‐line DH prediction without sample preparation remains insufficiently documented. Aim To evaluate the feasibility and predictive performance of rapid, high‐throughput FT‐IR spectroscopy for at‐line DH determination during full‐scale industrial WPH production without sample preparation. Methods WPH samples were collected during full‐scale production of four commercial product types. FT‐IR spectra were acquired at‐line using a MilkoScan FT3, while reference DH was determined using the O‐phthaldialdehyde (OPA) assay. Partial least squares regression (PLS) models were developed using cross‐validation and tested on an independent hold‐out validation set. Major Findings Across 80 industrial samples spanning DH from 4.8% to 22.5%, FT‐IR–PLS models achieved high predictive accuracy. The best model provided an external validation error (RMSEP) of approximately 0.78% DH, with strong agreement between predicted and reference DH values and minimal bias across multiple product types. Scientific or Industrial Implications At‐line FT‐IR prediction of DH can substantially reduce analytical lead‐time and manual laboratory workload, enabling faster feedback for process decisions. Implementation as a process analytical technology tool could improve batch consistency, reduce over‐processing, increase yield and support more efficient industrial control of enzymatic hydrolysis.
No takes yet. Share an insight, caveat, or question.
Holme et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: