A Comprehensive Guideline for Hybrid Modeling of Engineered Microbial Processes

Authors:
  • WXWeibo Xia
    Weibo Xia
  • JZJun-Jie Zhu
    Jun-Jie Zhu
  • JCJun Cao
    Jun Cao
  • ZRZhiyong Ren
    Zhiyong Ren
  • HYHeyang Yuan
    Heyang Yuan

Abstract
[This work has been published in Water Research (2025). DOI: https://doi.org/10.1016/j.watres.2025.124559] Hybrid modeling integrates the strengths of mechanistic and data-driven approaches to overcome their respective limitations, offering a promising framework for predicting and understanding complex engineered microbial processes. However, its application in this field remains underdeveloped. To assess the current state of hybrid modeling, an extensive literature review was conducted and yielded only 34 relevant articles over the past 30 years, with 26 hybrid models in 24 articles published in the last five years. This review systematically examined key processes in hybrid modeling, including data collection, data processing, model construction, and model validation, and identified critical challenges that were often not properly conducted. Issues such as hyperparameter tuning and data leakage were particularly overlooked in the reviewed literature, which could compromise model performance and reliability. To address these challenges, a standard protocol is proposed to mitigate common pitfalls and enhance model robustness. This review provides a comprehensive analysis and practical recommendations, laying the groundwork for future studies to advance hybrid modeling of engineered microbial processes.