Digital Twin Calibration for Natural System-of-Methods: Cell Custom Manufacturing Course of
Authors: Fuqiang Cheng, Wei Xie, Hua Zheng
Abstract: Biomanufacturing innovation will depend on an atmosphere pleasant design of experiments (DoE) to optimize processes and product top quality. Standard DoE methods, ignoring the underlying bioprocessing mechanisms, often bear from a shortage of interpretability and sample effectivity. This limitation motivates us to create a model new optimum learning technique which will info a sequential DoEs for digital twin model calibration. On this study, we ponder a multi-scale mechanistic model for cell custom course of, usually often known as Natural Methods-of-Methods (Bio-SoS), as our digital twin. This model with modular design, composed of sub-models, permits us to mix data all through quite a few manufacturing processes. To calibrate the Bio-SoS digital twin, we take into account the suggest squared error of model prediction and develop a computational technique to quantify the have an effect on of parameter estimation error of specific individual sub-models on the prediction accuracy of digital twin, which can info sample-efficient and interpretable DoEs.
