A comprehensive guideline for hybrid modeling of engineered microbial processes

Area of Science:

• Biotechnology and Environmental Engineering
• Computational Biology and Systems Biology

Background:

• Engineered microbial processes are vital for contaminant removal and product recovery.
• Controlling and modeling microbial populations and communities in these systems is challenging due to inherent variability.
• Hybrid models, combining mechanistic and data-driven approaches, show promise but are underdeveloped in this field.

Purpose of the Study:

• To review the current state of hybrid modeling in engineered microbial processes.
• To identify critical challenges and overlooked issues in hybrid model application.
• To propose a standardized protocol and recommendations for improving hybrid modeling.

Main Methods:

• Conducted an extensive literature review of hybrid modeling in engineered microbial processes over 30 years.
• Systematically examined data collection, processing, and model construction stages.
• Analyzed common pitfalls such as hyperparameter tuning and data leakage.

Main Results:

• Identified only 52 qualified articles, with 32 hybrid models in the last five years.
• Found critical challenges frequently overlooked, including missing hyperparameter tuning (21 studies) and data leakage concerns (16 studies).
• These issues significantly compromise model performance and reliability.

Conclusions:

• Hybrid modeling in engineered microbial processes requires standardized protocols to address overlooked challenges.
• Recommendations are provided for data collection, processing, and model construction to enhance reliability.
• Future work should focus on mitigating data scarcity and expanding hybrid modeling applications.