Bioprocess CQA Prediction Using State-Space and ML Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current bioprocess monitoring and control systems are limited by the slow measurement of critical quality attributes (CQAs), which are typically determined after the process is complete, leading to delayed corrective actions and inefficient quality assurance.
Innovation Solution
A method using a machine learning model trained on state variables and process conditions to predict CQAs in real-time, combining kinetic growth and material balance models to simulate and control bioprocesses, enabling timely adjustments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If offline measurement methods are used to determine CQAs, then measurement accuracy can be maintained, but measurement time is excessively long and quality determination is delayed until after process completion
Solution Approach 1:
The patent replaces slow offline laboratory measurement systems with rapid online analytical instruments (spectroscopy, chromatography, mass spectrometry) that provide real-time or near-real-time CQA data. This substitution of measurement technology enables both high precision and fast response, resolving the contradiction between measurement accuracy and measurement speed.
Solution Approach 2:
The patent introduces intermediate predictive models and sensors that estimate CQAs based on process parameters and intermediate measurements. These intermediaries provide real-time quality predictions without requiring complete process completion, enabling timely quality determination while maintaining accuracy through validated prediction algorithms.
2Reliability
If corrective actions are taken after process completion based on quality determination, then quality issues can be identified, but corrective actions are too late to prevent defects
Solution Approach 1:
The patent implements real-time feedback loops where CQA measurements and predictions are continuously fed back to process control systems. This enables dynamic adjustment of process parameters during operation to maintain quality within specifications, transforming quality assurance from a post-process check to an active control mechanism that prevents defects.
Solution Approach 2:
The patent uses real-time CQA monitoring and predictive modeling to identify quality deviations early in the process, enabling corrective actions to be taken before defects become permanent. This preliminary intervention prevents quality issues rather than merely detecting them after completion.
3Ease of operation
If multivariate statistical models are used to identify CPP ranges, then process conditions can be monitored, but the connection between CPPs and CQAs is not well understood leading to unnecessarily constraining limits
Solution Approach 1:
The patent transforms the relationship between CPPs and CQAs by introducing real-time measurement and prediction capabilities that reveal actual causal relationships. This enables dynamic adjustment of CPP limits based on real-time process state and quality predictions, replacing static, overly conservative limits with adaptive, scientifically-grounded ranges that maintain quality while improving operational flexibility.
Data Source
Figure 1
Figure 2A~2B
Figure 2C~2D
AI summary
Methods for monitoring, controlling, optimising and simulating a bioprocess comprising a cell culture in a bioreactor are provided. The methods comprise obtaining the values of one or more state variables of a state space model at one or more maturities, and predicting the value of one or more critical quality attributes of a product of the bioprocess using a machine learning model trained to predict the value of the one or more critical quality attributes based on input variables comprising values of the one or more state variables or variables derived therefrom, at one or more maturities. The state space model comprises a kinetic growth model representing changes in the state of the cell culture and a material balance model representing changes in the bulk concentration of one or more metabolites in the bioreactor. Systems, computer readable media implementing such methods, and methods for providing tools to implement such methods, are also provided.