Beyond the model: data infrastructure as the foundation for autonomous virtual laboratories.

Area of Science:

• Biotechnology
• Artificial Intelligence
• Data Science

Background:

• AI and machine learning are revolutionizing biotechnology.
• Current impact is limited by inadequate data infrastructure, with data quality, standardization, and interoperability as key bottlenecks.

Purpose of the Study:

• To review the limitations of current data infrastructure in AI-enabled biomanufacturing.
• To advocate for a data-centric approach to overcome these challenges and advance the field.

Main Methods:

• Review of current literature and practices in AI and biomanufacturing.
• Analysis of bottlenecks in the design-build-test-learn cycle.
• Identification of essential data practices for AI advancement.

Main Results:

• Data quality, standardization, and interoperability are critical limitations.
• The manual test-to-learn data ingestion step introduces significant latency.
• Curated repositories, consistent metadata, real-time validation, and efficient learning strategies are essential.

Conclusions:

• A data-centric strategy is necessary for AI-enabled biomanufacturing.
• Addressing data limitations will enable scalable, reliable, and autonomous virtual laboratories.
• Adherence to Findable, Accessible, Interoperable, Reusable (FAIR) principles is vital.