70NANB24H071

Within the biopharmaceutical sector, there exist the need for a contactless multiplex sensor, which can accurately detect cell viability and metabolite levels in real time for precise feedback control of a bioreactor environment. The use of hyperspectral imaging (HSI) allows for efficient, fully contactless collection of large spectral datasets for metabolite quantification. Here, we report the development of an interpretable AI/deep learning system a convolution metabolite regression (CMR) approach that detects cell viability, glucose, and lactate concentrations using label-free contactless HS images of media samples containing Chinese hamster ovary (NISTCHO) cell grown in an expansion bioreactor. We propose using a dataset of <500 HS images of a microfluidics chip with flowing CHOH cells and spent media online in order to predict NISTCHO cell viability, total cell count, glucose, and lactate based using the CMR model and a training set of ground truth readings of each of these material attributes and process parameters in real-time. Collectively, this work will advance the safe and efficient adoption of contactless AI/deep learning sensing systems for a fine control of a variety of bioreactor environments.