Harnessing near-infrared and Raman spectral sensing and artificial intelligence for real-time monitoring and precision control of bioprocess

Abstract

Effective monitoring and control of bioprocesses are critical for industrial biomanufacturing. This study demonstrates the integration of near-infrared and Raman spectroscopy for real-time monitoring and precise control of gentamicin fermentation. The orthogonal method reduced redundant features and improved spectral model performance by 9.2–100.4 % in terms of the coefficient of determination (R²). The combinatorial spectral model outperformed single-source models in external validation (R² > 0.99). An AI-based platform, combining dual-sensors data collection, ML-based prediction, and automated feeding control, was developed for fully automated fed-batch fermentation. This platform dynamically adjusted feeding rates, maintained low glucose concentrations (5 g/L) with accuracy and coefficient of variation below 2 %, and increased gentamicin C1a concentration (346.5 mg/L) by 33.0 % compared to traditional intermittent feeding. These findings underscore the transformative potential of combinatorial spectroscopy and machine learning for real-time bioprocess monitoring, offering a scalable solution for enhancing industrial fermentation efficiency and product titer.