Multi-herb formulations modulating gut microbiota
A systematic review and data-driven analysis
Bibliographic Data
| ID | 19406344 |
|---|---|
| Authors | Min-Seok Cho (Kyung Hee University), In-Seon Lee, In‐Seon Lee (0000-0001-9275-3111, Kyung Hee University), Jae‐Joong Kim (0000-0002-2714-2282, Kwangwoon University), Junsuk Kim (0000-0002-7252-3420), Jae-Woo Park (0000-0002-1862-7317, Kyung Hee University), Jinsung Kim (0000-0001-8968-416X, Kwangwoon University), Seok-Jae Ko (0000-0002-9902-0938, Kyung Hee University, corresponding author) |
| Year | 2026 |
| Volume | 370 |
| Pages | 122082 |
| Publication date | 2026-11-01 |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Journal of Ethnopharmacology (JOURNAL) |
| Journal identifiers | ISSN: 0378-8741 • E-ISSN: 1872-7573 |
| Publisher | Elsevier BV (PUBLISHER) |
| DOI | 10.1016/j.jep.2026.122082 |
| PMID | 42320776 |
| OpenAlex | W7165123461 |
| Language | EN |
| References cited | 39 |
ETHNOPHARMACOLOGICAL RELEVANCE: Multi-herb formulations, characterized by their complex synergistic compositions, are widely used in traditional medicine to modulate the gut microbiota. However, identifying reproducible herb-microbiota associations across disparate clinical settings remains a significant methodological challenge. AIM OF THE STUDY: This study aims to systematically synthesize human clinical evidence to map the modulation patterns of multi-herb formulations on the gut microbiota and to identify the herbal components associated with reported directional microbial shifts. MATERIALS AND METHODS: We conducted a systematic review and data-driven analysis of 29 clinical trials involving 954 participants in multi-herb formulation groups. To integrate findings from heterogeneous clinical settings, we employed a binarization strategy focused on statistically significant directional shifts (+1 for increase, -1 for decrease). An extreme gradient boosting (XGBoost) learning framework combined with SHapley Additive exPlanations (SHAP) was used to deconstruct these formulations and explore the predictive importance of individual constituents. To ensure the highest level of scientific integrity and prevent data leakage, the model's generalizability was rigorously validated using Leave-One-Study-Out (LOSO) cross-validation at the independent study level. RESULTS: The LOSO validation yielded a mean accuracy of 0.84 and a macro F1-score of 0.42, indicating limited but informative cross-study pattern recognition despite the inherent heterogeneity of clinical data. Our analysis identified recurrent directional associations: formulas containing Scutellaria baicalensis Georgi were associated with reported reductions in genus-level taxa such as Escherichia-Shigella within neuropsychiatric disease contexts. Formulas containing Zingiber officinale Roscoe were associated with reported increases and decreases in selected genus-level taxa across heterogeneous disease contexts. CONCLUSIONS: This study provides a comprehensive, evidence-based map of how multi-herb formulations modulate the human gut microbiota. By prioritizing rigorous validation and accounting for the complexity of synergistic preparations, we have identified hypothesis-generating patterns that transcend individual study variations. These findings provide a realistic foundation for future high-resolution metagenomic research and the development of standardized ethnopharmacological therapies
Clinical trial · Disease · Generalizability theory · Generalization · Gut flora · MEDLINE · Precision medicine · Scutellaria baicalensis · Barrier Structure and Function Studies · Gut microbiota and health · Vagus Nerve Stimulation Research
| Highly cited | No |
|---|