This is the authors' abstract. We don't add key points for this paper.
Mathematical models of bioreactors are powerful tools that aid in the analysis and prediction of process operation. However, the complex behavior of microorganisms makes modelling of biological processes a particularly challenging task, especially in the early developmental stages when data and knowledge are scarce. As a result, bioreactor models may perform poorly due to structural errors or high uncertainty in their parameterization. Here, we present a method for automated dynamic model regression based on a Bayesian approach that can be applied in the operation of laboratory robotic platforms to perform both parameter estimation and state predictions for a given experimental design. Starting with wide distributions over parameters (prior knowledge), the model is updated as new data is generated and is then used to predict the evolution of the experiment. The proposed method is tested with data from several parallel cultivations from a 24 mini-bioreactors platform containing an Escherichia coli strain operating in fed-batch mode. The results highlight both the versatility of the approach to estimate parameter distribution as well as to predict the state evolution. • Automated regression of bioreactor models is applied to parallel cultivations performed in robotic platforms. • Bayesian methods foster iteratively model improvement as new data are available. • Process System Engineering tools can take advantage of probabilistic predictions of bioreactor models. • Bayesian model regression is experimentally tested with data from two parallel fermentation experiments.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Luna et al. (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: