Interpretable Machine Learning Reveals Complementary Age-Related Signatures in the Oral and Gut Microbiome
- 1BRAC University
Abstract
Whether combining microbiome data from multiple body sites improves prediction, and whether different sites carry complementary or redundant information, are distinct questions that most studies conflate into a single accuracy metric. This work makes two contributions, one methodological and one biological, validated across two independent cohorts. In the primary cohort (Ferretti et al., 2018; n=44, adult vs. newborn classification), we show that a subject-matched fusion design combined with SHAP-based (SHapley Additive exPlanations) site attribution can detect complementary information between body sites even when no measurable accuracy gain results. This is a pattern that conventional model comparison would misread as a null result. Gut (stool) composition alone achieved near-perfect classification (area under the receiver operating characteristic curve, AUC = 1.00), and combined stool-oral models never exceeded this ceiling. A null baseline, bootstrap confidence intervals, and preprocessing sensitivity checks confirmed that this ceiling reflects genuine biological signal rather than an artifact. Despite the flat accuracy curve, SHAP analysis of the fused model showed that oral cavity features carried more total feature importance than stool features (58.1% versus 41.9%), indicating that the model draws on real, non-redundant information from both sites. We replicated this finding in an independent, larger cohort (Brito et al., 2016; n=140, sex classification), where oral cavity again dominated the fused model's SHAP importance (60.7% versus 39.3%), and a permutation test confirmed the asymmetry was statistically significant (p < 0.005). Biologically, the taxa driving these patterns include Malassezia restricta, Staphylococcus epidermidis, and Prevotella melaninogenica. These taxa behave in a manner consistent with their established roles as early colonizers of the neonatal gut, skin, and oral cavity, once their model-specific behavior is verified directly against abundance data rather than inferred from the literature alone. Together, these results support a model of oral-gut microbiome maturation as two distinct, complementary processes, and demonstrate that detecting this kind of relationship requires examining a model's internal reasoning rather than its accuracy alone. The site-attribution method is released as a standalone, reusable Python tool.
Related
- Interpretable Machine Learning Links Group-Specific Plasma Proteomics to Regional Structural Neuroimaging in Alzheimer's Disease
- Diabetes and all-cause mortality in hemodialysis: A comparative survival analysis with complementary machine-learning methods
- Sparse Machine Learning Pipeline with Stabl Identifies Cord Blood Multi-Omic Signatures of Bronchopulmonary Dysplasia
- Effects of drying time on physicochemical properties and volatile organic compounds of pork jerky revealed by GC-IMS and interpretable machine learning.
- An Interpretable EEG Machine Learning Prototype for Alzheimers Disease Classification
BibTeX
@misc{ruthbah2026interpretable,
title = {{Interpretable Machine Learning Reveals Complementary Age-Related Signatures in the Oral and Gut Microbiome}},
author = {C. A. Ruthbah and T. H. Sadi and N. E. S. Jahan and A. N. M. T. Adib},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.09.750358},
url = {https://www.biorxiv.org/content/10.64898/2026.09.09.750358v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Interpretable Machine Learning Reveals Complementary Age-Related Signatures in the Oral and Gut Microbiome
AU - C. A. Ruthbah
AU - T. H. Sadi
AU - N. E. S. Jahan
AU - A. N. M. T. Adib
PY - 2026
DA - 2026/09/09
PB - bioRxiv
DO - 10.64898/2026.09.09.750358
UR - https://www.biorxiv.org/content/10.64898/2026.09.09.750358v1
AB - Whether combining microbiome data from multiple body sites improves prediction, and whether different sites carry complementary or redundant information, are distinct questions that most studies conflate into a single accuracy metric. This work makes two contributions, one methodological and one biological, validated across two independent cohorts. In the primary cohort (Ferretti et al., 2018; n=44, adult vs. newborn classification), we show that a subject-matched fusion design combined with SHAP-based (SHapley Additive exPlanations) site attribution can detect complementary information between body sites even when no measurable accuracy gain results. This is a pattern that conventional model comparison would misread as a null result. Gut (stool) composition alone achieved near-perfect classification (area under the receiver operating characteristic curve, AUC = 1.00), and combined stool-oral models never exceeded this ceiling. A null baseline, bootstrap confidence intervals, and preprocessing sensitivity checks confirmed that this ceiling reflects genuine biological signal rather than an artifact. Despite the flat accuracy curve, SHAP analysis of the fused model showed that oral cavity features carried more total feature importance than stool features (58.1% versus 41.9%), indicating that the model draws on real, non-redundant information from both sites. We replicated this finding in an independent, larger cohort (Brito et al., 2016; n=140, sex classification), where oral cavity again dominated the fused model's SHAP importance (60.7% versus 39.3%), and a permutation test confirmed the asymmetry was statistically significant (p < 0.005). Biologically, the taxa driving these patterns include Malassezia restricta, Staphylococcus epidermidis, and Prevotella melaninogenica. These taxa behave in a manner consistent with their established roles as early colonizers of the neonatal gut, skin, and oral cavity, once their model-specific behavior is verified directly against abundance data rather than inferred from the literature alone. Together, these results support a model of oral-gut microbiome maturation as two distinct, complementary processes, and demonstrate that detecting this kind of relationship requires examining a model's internal reasoning rather than its accuracy alone. The site-attribution method is released as a standalone, reusable Python tool.
ER -