Application of Artificial Intelligence–Based Methods for Foam Monitoring and Control in Fermentative Bioprocesses

Foam formation remains a major operational challenge in fermentative bioprocesses, affecting process stability, product quality, and industrial scalability. In the last decade, artificial intelligence (AI)–based approaches, including computer vision (CV), machine learning (ML), and deep learning (DL), have been increasingly applied for automated foam monitoring and control. This systematic review and meta-analysis were conducted in accordance with PRISMA 2020 guidelines to critically evaluate AI-driven methods for foam detection, quantification, and control in fermentation systems. Systematic searches of PubMed, Scopus, Web of Science, and IEEE Xplore identified 566 records published between 2015 and 2025, of which 167 studies met the inclusion criteria and 89 were eligible for quantitative synthesis. Random-effects meta-analysis demonstrated that deep learning–based models significantly outperformed classical image processing approaches for foam-related tasks, with pooled coefficients of determination reaching R² = 0.88 (95% CI: 0.84–0.92) for foam height or volume prediction. Substantial heterogeneity was observed across studies (I² = 72%), reflecting variability in experimental scale, datasets, and validation protocols. Risk of bias assessment using QUADAS-AI and PROBAST-AI revealed moderate methodological concerns in model validation and dataset representativeness, while overall certainty of evidence was rated as low to very low according to GRADE criteria, reflecting methodological limitations in model validation and dataset heterogeneity. These findings indicate that AI-based foam monitoring enables accurate, non-invasive, and real-time assessment in fermentative bioprocesses, but also highlight the need for standardized datasets, transparent validation strategies, and tighter integration with process control frameworks to support reliable industrial deployment.