Cell Culture Device Control With Adaptive Machine Learning Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing culture processes lack suitable control mechanisms to respond to changes in culture conditions over time, particularly in the upstream phase of cell proliferation.
Innovation Solution
A control method for a culture device that involves acquiring process and analytical variables at predetermined timings, deriving a new set variable through machine learning, and recalculating the culture environment using a function updated with the latest variables to maintain a suitable culture environment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional culture control methods are used, then the culture process is simple and easy to operate, but the culture environment cannot adapt to changes over time and production quality deteriorates
Solution Approach 1:
The patent implements a feedback control system that continuously monitors process variables (pH, dissolved oxygen, temperature) and analytical variables (cell concentration, product concentration) and uses this information to dynamically adjust culture conditions. The control device calculates optimal setpoints based on actual measurements, creating a closed-loop system that adapts to changes in real-time.
Solution Approach 2:
The patent transitions from static culture control to dynamic control by continuously updating setpoints based on real-time data. The control system adjusts culture parameters dynamically throughout the process, allowing the culture environment to adapt to changing conditions such as cell growth phases, substrate depletion, and product accumulation.
2Productivity
If real-time monitoring and machine learning control are implemented, then production quality and productivity improve, but measurement and detection difficulty increases
Solution Approach 1:
The patent employs a multi-functional control device that integrates multiple monitoring and control functions into a single system. The control device handles process variable monitoring, analytical variable measurement, machine learning model execution, and setpoint adjustment all through one integrated platform, reducing the complexity of detecting and measuring multiple variables.
Solution Approach 2:
The patent introduces machine learning models as intermediaries between raw sensor data and control decisions. The machine learning algorithms process complex patterns from process and analytical variables, transforming raw measurements into actionable insights that guide control adjustments, thereby simplifying the detection and measurement complexity.
3Reliability
If continuous culture monitoring is performed, then culture quality is maintained, but energy consumption and operational time increase
Solution Approach 1:
The patent implements periodic sampling and monitoring rather than continuous uninterrupted monitoring. The control device performs measurements and data processing at optimized intervals based on the culture process phase, maintaining culture quality while reducing energy consumption compared to continuous monitoring.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
To provide a control method for a culture device which allows culture to be performed in a more suitable culture environment. A control method for a culture device 1 is a control method for a culture device to input a set variable SV for setting a culture environment and control the culture device performing culture, the control method including: a first step S1 of acquiring, at a predetermined timing, a process variable PV of a culture environment with the set variable SV and an analytical variable AV of a culture solution c obtained by culturing; a second step S2 of deriving by machine learning a function F for calculating a new set variable SV for resetting the culture environment using the process variable PV and the analytical variable AV acquired at the predetermined timing, and a third step S3 of calculating the new set variable SV on the basis of a desired target variable TV of the culture solution by using the function F, inputting the new set variable SV, and performing culture, wherein the function F used in the third step S3 is a new function F derived as needed by machine learning, with a latest process variable PV and a latest analytical variable AV that are acquired at a predetermined timing being included.