Methods of determining viable cell count and impedance-based biosensors for the same

A system comprising:

• a hollow fiber bioreactor comprising:
• an inlet configured for a continuous perfusion of a culture medium containing cells;
• an outlet configured for a removal of waste products;
• and
• a bioreactor extending from the inlet to the outlet, the bioreactor having an exterior wall defining a bioreactor volume therewithin, the exterior wall accommodating semi-permeable hollow fibers in the bioreactor volume, the bioreactor volume having an intracapillary (IC) volume component within the hollow fibers and an extra capillary (EC) volume component outside the hollow fibers;
• a trans-wall impedance-based biosensor comprising:
• an IC probe in fluid communication with the IC volume component;
• and
• one or more EC probes in fluid communication with the EC volume component;
• an impedance data acquisition system configured to acquire impedance data from the trans-wall impedance-based biosensor;
• and
• a data processing system configured to calculate a viable cell count (VCC) of cells in the bioreactor based on the impedance data;
• wherein the data processing system is further configured to:
• process the impedance data of a cell suspension component and an electrode polarization component;
• extract physics-based cell suspension parametric features from the impedance of the cell suspension component;
• extract physics-based electrode polarization parametric features from the impedance of the electrode polarization component;
• predict the VCC with an equation in a physics-scaled model including a first calibration coefficient and variables derived directly from theoretical calculations;
• predict the VCC with an equation in a gray-box model utilizing a power-law between the variables of the physics-scaled model, which includes the first calibration coefficient and gray-box calibration coefficients for each of the variables reflecting the power-law;
• predict the VCC with an equation in an enhanced gray-box model comprising the equation of the gray-box model multiplied with one or more extracted features, each extracted feature raised to an enhanced gray-box calibration coefficient, the one or more extracted features selected from a group consisting of the physics-based cell suspension parametric features and the physics-based electrode polarization parametric features;
• evaluate an accuracy of each of the models against image-based viable cell count data;
• train each of the models with the image-based viable cell count data;
• and
• output the calculated VCC of the trained enhanced gray-box model.