Machine learning methods for small data and upstream bioprocessing applications: A comprehensive review.

Area of Science:

• Biotechnology
• Bioprocessing
• Machine Learning

Background:

• Acquiring large datasets for machine learning (ML) is challenging and expensive in biopharmaceutical upstream bioprocessing.
• Complex cell cultivation and optimization processes yield limited data, hindering ML applications.

Purpose of the Study:

• To review and classify ML methods that address small data challenges in bioprocessing.
• To provide practical guidance for applying ML in data-constrained biopharmaceutical research.

Main Methods:

• Literature review of ML methods for small data.
• Classification of methods into a taxonomy.
• Analysis of core concepts and effectiveness of each method.

Main Results:

• A taxonomy of ML methods for small data challenges was developed.
• Effectiveness of various ML techniques was evaluated using bioprocessing examples.
• Research gaps in ML for data-constrained bioprocessing were identified.

Conclusions:

• ML offers solutions for data scarcity in upstream bioprocessing.
• The reviewed methods and taxonomy provide actionable insights for researchers.
• Further research is needed to optimize ML in data-limited biopharmaceutical development.