Hybrid Cell Metabolism Observer for Bioreactor Yield Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biomanufacturing processes lack direct measurements of cell metabolism, leading to trial-and-error approaches for optimizing bioprocesses, difficulty in diagnosing abnormal metabolic operations, and limited ability to predict high-yield product formation, relying heavily on subject matter expert judgment.
Innovation Solution
A computer-implemented method using a hybrid model that includes a kinetic growth model and a metabolic condition model to predict biomaterial production, classify metabolic states, and provide real-time observability and control of bioreactor conditions, utilizing machine learning and statistical models to optimize process conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If indirect measurements of average cell metabolic behavior are used, then measurement cost and complexity are reduced, but measurement precision and ability to diagnose abnormal metabolic operations deteriorate
Solution Approach 1:
The patent introduces an artificial neural network as an intermediary computational system that processes indirect measurements (metabolite concentrations, process parameters) and transforms them into accurate predictions of direct metabolic states (specific consumption rates, specific production rates). The ANN acts as a mediator between easily measurable bulk parameters and difficult-to-measure cellular metabolic states, achieving high measurement precision without requiring complex direct measurement instrumentation.
2Reliability
If statistical regression models are used to predict titer and byproducts, then predictive capability is improved, but ability to understand process variables responsible for metabolic function deteriorates
Solution Approach 1:
The patent implements feedback by using the neural network to continuously predict metabolic rates and comparing these predictions with actual measured values. The system provides feedback on which process variables (substrate concentrations, process parameters) are most influential in determining metabolic state, enabling both accurate prediction and process understanding. The model identifies key input variables that drive metabolic behavior, preventing information loss about process mechanisms.
3Productivity
If Monod kinetics models are used to predict cell growth, then growth rate prediction is improved, but ability to consider current cell state deteriorates
Solution Approach 1:
The patent transitions from static Monod kinetics models to a dynamic neural network model that adapts to changing cell states in real-time. The neural network is trained on data spanning different growth phases and metabolic conditions, enabling it to dynamically adjust predictions based on current substrate concentrations, process parameters, and metabolic rates. This dynamic approach captures the evolving cellular state throughout the bioprocess, unlike fixed-parameter kinetic models.
4Adaptability or versatility
If trial and error approaches are used to design process operations, then process development flexibility is maintained, but development time and resource consumption increase
Solution Approach 1:
The patent applies preliminary action by training the neural network model in advance using historical process data and metabolic measurements. Once trained, the model provides immediate predictions of metabolic rates and optimal process conditions without requiring real-time trial and error experimentation. The preliminary modeling work enables rapid what-if analysis and process optimization, dramatically reducing development time while maintaining flexibility through the model's ability to evaluate multiple process scenarios.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Techniques for predicting an amount of at least one biomaterial produced or consumed by a biological system in a bioreactor are provided. Process conditions and metabolite concentrations are measured for the biological system as a function of time. Metabolic rates for the biological system, including specific consumption rates of metabolites and specific production rates of metabolites are determined. The process conditions and the metabolic rates are provided to a hybrid system model configured to predict production of the biomaterial. The hybrid system model includes a kinetic growth model configured to estimate cell growth as a function of time and a metabolic condition model based on metabolite specific consumption or secretion rates and select process conditions, wherein the metabolic condition model is configured to classify the biological system into a metabolic state. An amount of the biomaterial based on the hybrid system model is predicted.