Innovating cell culture process development with deep learning-powered robotic experimentation using the first...

Area of Science:

• Biotechnology
• Process Engineering
• Artificial Intelligence

Background:

• Traditional biologics process development is inefficient, relying on labor-intensive optimization.
• Current methods for antibody and recombinant protein production require extensive iterative cell culture optimization.

Purpose of the Study:

• To develop an autonomous laboratory framework, the Industrial Smart Lab Framework for Cell Culture (ISLFCC), to enhance cell culture processes.
• To accelerate biologics development and biomanufacturing through AI and robotic automation.

Main Methods:

• Implemented ISLFCC, combining deep learning (decoder-only transformer models) with robotic experimentation.
• Utilized an IoT system for data transmission from bioreactors to AI models and execution of automated actions.
• Employed AI to predict cell states and recommend optimal actions like nutrient feed and temperature adjustments.

Main Results:

• Achieved an average titer increase of 26.8% for three cell clones in a single batch.
• Maintained lactate levels below 1 g/L without late-phase rebound.
• Demonstrated enhanced reproducibility, data accuracy, adaptability, and scalability in 3 and 15 L bioreactors.

Conclusions:

• The ISLFCC framework offers a transformative, autonomous, and data-driven approach to biomanufacturing.
• This AI-driven methodology significantly accelerates biologics development compared to traditional empirical methods.
• Represents a paradigm shift towards automated, high-throughput cell culture in biopharmaceutical production.