Calibration transfer of near infrared spectrometers for the assessment of plasma ethanol precipitation process

Abstract

Plasma ethanol precipitation process is an important step in the biopharmacy production process, and near infrared spectroscopy (NIRS) is an effective process analytical technology (PAT) tool in detecting the key quality attribute: the total protein (TP) content. But establishing a robust model is pretty complex. The aim of this study was to explore the calibration transfer strategy between master and slave spectrometers with the biopharmacy. Firstly, 8 batches of ethanol precipitation process were experimented. The changing spectra were collected by two spectrometers, reference values were measured by biuret method. Then, calibration transfer based on the PRS algorithm was carried out. Finally, before establishment of the optimal TP content partial least squares regression (PLSR) model of calibration transfer, rank sample set partitioning based on joint x-y distance (Rank-SPXY) and spectral pretreatment methods were selected after comparison. The determination coefficients of prediction (Rp2) and root mean squares error of prediction (RMSEP) were 0.934 and 0.8395 g/L, respectively. The results showed that Rank-SPXY-PRS was an effective calibration transfer strategy with different NIRS spectrometers. To some extent, the predictive capability of PLSR model was improved after calibration transfer, and it could provide reference for the further research in biopharmacy production process.