MicroAge
A machine-learning model that estimates biological age from the composition of the saliva microbiome — a very early, emerging approach to aging biomarkers, distinct from the mortality-focused OMAA Score.
Number of studies
2
Safety
High
Time to effects
Not applicable — this is an analytical/research tool, not an intervention producing effects over time.
Who it's for
Table of contents
TL;DR
A machine-learning model that estimates biological age from the composition of the saliva microbiome — a very early, emerging approach to aging biomarkers, distinct from the mortality-focused OMAA Score.
- →A non-invasive method of data collection (a saliva sample instead of blood)
- →Shows that oral microbiome composition carries a measurable signal linked to age and health status in a large, diverse population
- →A potentially useful research direction for further differentiating aging biomarkers independent of blood and DNA
| Tool type | Machine-learning model estimating biological age from the saliva microbiome |
|---|---|
| Research level | Early-stage — a single publication (2025), no independent replication yet |
| Data source | 16S rRNA analysis of the saliva microbiome, over 8,000 microbial features |
| Training sample size | 4,532 healthy people from six continents |
| Key metric | MicroAgeGap — the difference between predicted and chronological age |
| Differs from | OMAA Score — a separate oral-microbiome-based indicator linked to mortality and frailty |
| Status | Research tool, not a validated commercially available clinical test |
Understand
Overview
MicroAge is the name of a machine-learning model, published in 2025 in the journal iMetaOmics, that uses saliva microbiome composition (16S rRNA gene analysis) to estimate a person's biological age and detect deviations suggesting poorer health. The model was trained on saliva samples from 4,532 healthy people across six continents, using presence and abundance data from more than 8,000 microbial features. It is one of the first 'clocks' of this kind based not on blood or DNA, but on the oral microbiome.
It's worth distinguishing MicroAge right away from the related but distinct concept of the OMAA Score (Oral Microbiome Aging Acceleration Score) — both use the oral microbiome, but serve different purposes. The OMAA Score is an 'aging acceleration' indicator calculated as the difference between predicted and actual age, directly linked to mortality and frailty in a specific research cohort. MicroAge is a broader predictive model, tested primarily for accuracy in age estimation and detecting disease states (the so-called MicroAgeGap — the difference between predicted and chronological age as a signal of health disruption), rather than as a ready-made mortality-risk indicator. In other words, these are two different analytical tools built on a similar type of data, not synonyms.
This field of research is at a very early stage. A single publication, even a methodologically well-designed one, is not the same as a validated, reproducible clinical biomarker. People interested in the biology of aging and the microbiome can treat MicroAge as an interesting research direction showing that oral bacterial composition (and earlier, gut bacterial composition) carries an age-related signal — but not as a validated test available today for self-assessing one's own biological age.
Mechanism of action
The composition of the microbiome — both gut and oral — changes with age in a partly predictable way: certain bacterial groups systematically gain abundance while others decline, and overall microbial diversity shifts as well. Models like MicroAge exploit this relationship statistically — a machine-learning algorithm (in this case compared against methods such as LASSO and elastic net) analyzes thousands of microbial features from a saliva sample and learns to recognize the pattern typical of a given chronological age across a large healthy population.
Once trained, the model can be applied to a new sample and generate a predicted 'microbiome age.' The difference between this value and a person's chronological age (MicroAgeGap) is interpreted as a potential health signal — in the study described, a positive gap was associated with a pro-inflammatory profile, and centenarians consistently had a lower predicted microbiome age than their chronological age would suggest. A key limitation: the model, which explained most of the variance in age in the healthy population (R² ≈ 0.80), performed noticeably worse in populations with disease (R² ≈ 0.16) — showing that the predictive power of such models depends heavily on clinical context and is not universal.
Saliva sample collection
Non-invasive sample collection and 16S rRNA gene sequencing to determine oral microbiome composition.
Microbial feature extraction
Analysis of the presence and abundance of thousands of bacterial taxa as model input data.
Age prediction via the ML model
An algorithm trained on a large, healthy population estimates 'microbiome age' from a pattern typical of a given age group.
MicroAgeGap interpretation
The difference between predicted and actual age is analyzed as a potential, preliminary health signal — not a ready-made diagnosis.
Evidence: early-stage — based on 2 studies in this database.
Benefits
Common myths
MythMicroAge and the OMAA Score are the same test under two names.
FactThese are two distinct models built on a similar type of data (oral microbiome) but with different purposes: MicroAge estimates biological age and health deviations (MicroAgeGap), while the OMAA Score is an aging-acceleration indicator directly linked to mortality and frailty in a specific cohort.
MythSince the model predicts age well in a healthy population, it can already be treated as a ready-made diagnostic test.
FactThe model's performance dropped markedly in populations with disease (R² ≈ 0.16 versus ≈ 0.80 in the healthy population), showing the tool needs further validation before clinical use can be discussed.
Check your profile
Not sure which supplements actually make sense for you?
Answer a few short questions about your lifestyle, diet, sleep, and goals. VitMode will build your profile and show supplements worth considering — with reasoning and evidence strength.
Recommendations take your answers and the strength of the scientific evidence into account. A supplement's popularity has no bearing on whether it gets recommended.
Practice
Frequently asked questions
It's a machine-learning model (published in 2025 in iMetaOmics) that estimates biological age from saliva microbiome composition, trained on a sample of over 4,500 healthy people.
Both use the oral microbiome, but MicroAge is a general model for predicting age and health status (MicroAgeGap), while the OMAA Score is a separate indicator of aging acceleration directly linked to mortality and frailty in the studied cohort.
There is currently no validated, commercially available clinical test based on this specific model — it's a research tool described in a single scientific publication.
Yes, separate 'clocks' based on the gut microbiome (e.g., the so-called gAge) are being developed in parallel, also linking bacterial composition to age and health status, but these are different models from MicroAge.
No — it's a general statistical signal that in the study was associated with a pro-inflammatory profile or reduced immune function, not a diagnosis of a specific disease.
What to combine with
Good combinations
Gut Microbiome — Both areas — the oral and gut microbiome — are studied as independent sources of age-related signal, though they involve different bacterial ecosystems in the body
Safety
Side effects & contraindications
Possible side effects
Contraindications
No significant contraindications at typical doses.
Is it worth taking?
Who it's for
- People following research on aging biomarkers and the microbiome
- Science-minded readers interested in new methods of assessing biological age
Not for
- No significant contraindications at typical doses.
Evidence
Worth knowing
The MicroAge model was trained on saliva samples from 4,532 healthy people across six continents, using over 8,000 microbial features.
Centenarians in the study consistently had a lower predicted microbiome age (MicroAgeGap) than their chronological age would suggest.
Studies
Saliva MicroAge: A salivary microbiome based machine learning model for noninvasive aging assessment and health state prediction
Early-stage evidenceXu T, Niu Y, Deng C, Cheung Y, Li Y, Hu Z, Sun S, Chen Y, He F, Yang G, Chen F, Duan C, Huang Y, Deng X · iMetaOmics · 2025
The publication introducing the MicroAge model — a machine-learning model trained on 4,532 saliva samples from six continents that predicts biological age from the oral microbiome; performance (R² ≈ 0.80 in the healthy population) dropped to R² ≈ 0.16 in populations with disease.
View studyA gut aging clock using microbiome multi-view profiles is associated with health and frail risk
Early-stage evidenceWang H, Chen Y, Feng L, Lu S, Zhu J, Zhao J, Zhang H, Chen W, Lu W · Gut Microbes · 2024
A separate but conceptually related model (gAge) based on the gut microbiome, showing that a similar approach to 'microbiome clocks' recurs across different bacterial ecosystems in the body and correlates with health status and frailty risk.
View studySources & bibliography
Citations are illustrative for this demo version and require full bibliographic verification by the editorial team before production publication.
Compare with similar entries
About the authors of this entry
Author
dr Anna KowalczykEditor-in-Chief, Molecular Biology
Anna studied molecular biology at the University of Warsaw, then spent eight years after her PhD in a lab researching the mechanisms of cellular aging and autophagy. She stumbled into science journalism almost by accident — frustrated by how easily her field's findings get oversimplified in the media, she started a blog explaining the biology of aging in plain language. That blog became the seed of VitMode. Today Anna oversees the entire editorial process, holding every piece to the same rigor her old lab demanded: primary sources, methodology checks, and honesty about the limits of the evidence. Outside work, she's a dedicated boulderer.
121 publications on this site
Medical review
dr Piotr ZielińskiEndocrinologist
Piotr has practiced endocrinology for more than fifteen years, mostly in male hormonal disorders and metabolic health. He joined VitMode as a scientific consultant because, as he jokes, he got tired of explaining the same testosterone questions at every appointment and decided to write the answers down properly, once. He reviews content on hormone therapy, supplement pharmacology and drug interactions, making sure articles never turn into encouragement to self-supplement in situations that genuinely need diagnostics and medical supervision. His professional motto — "evidence first, enthusiasm second" — has come up more than once with a patient who arrived with a supplement plan they found online.
174 publications on this site
Related entries
4.2Biological Age Clocks
An umbrella term covering very different methods for estimating the 'true' pace at which an organism is aging — from well-validated second-generation epigenetic clocks to far less proven commercial tests based on glycans or the microbiome. Not all of them carry a comparable level of evidence.
4.0OMAA Score
The OMAA Score (Oral Microbiome Aging Acceleration) is a numerical index calculated as the difference between the age predicted from an oral microbiome sample and a person's actual age — in the study that introduced it, it was linked to elevated mortality and frailty risk.
4.6Gut Microbiome
Trillions of bacteria, viruses, and fungi colonizing the gut aren't a passive "passenger" in the body — they're an active ecosystem influencing digestion, immunity, and even mood, with an increasingly well-documented, though still not fully mapped, role in whole-body health.
4.1Oral Microbiome Aging
The composition of bacteria colonizing the mouth changes in a predictable way with age — predictable enough that researchers have started building models to estimate biological age from a simple saliva sample.
4.2Organ Biological Age
The concept that individual organs — the heart, brain, liver, or kidneys — can age at markedly different rates within the same person, measurable through analysis of tissue-specific plasma proteins rather than a single, averaged 'biological age' score.
4.7HbA1c (Glycated Hemoglobin)
A biomarker reflecting average blood glucose over the past 2–3 months — the gold standard for diagnosing and monitoring diabetes, far more stable than a single glucose measurement.
4.1Epigenetic Age Acceleration
The difference between an age estimated from a DNA methylation pattern and chronological age — one of the most statistically well-documented aging biomarkers, strongly linked to mortality in population studies, though still of limited diagnostic value for a single individual.
Related articles
Poradniki5 Biomarkers Worth Knowing Before 40 — Signals Before Symptoms Appear
A standard "once-a-year blood count" often misses early signs of metabolic and hormonal issues. Here are five markers worth asking your doctor about, before anything starts to hurt.
August 1, 2026
PoradnikiProbiotics and Antibiotics: Do They Really Prevent Diarrhea?
This is one of the best-studied uses of probiotics — and one of the few cases where science points to a specific, best strain instead of a vague "probiotics help." We check how much the risk of antibiotic-associated diarrhea really drops, who benefits most, and which strain wins in head-to-head comparison.
August 22, 2026
PoradnikiCollagen and Skin Health: What Do Two Meta-Analyses Show?
Collagen powder or capsules are one of the fastest-growing segments of the beauty supplement market — and, rare for this category, one of the few where the popular promise ("firmer, better-hydrated skin") has earned solid confirmation in two independent, well-designed meta-analyses. This is a rare situation in supplement science, worth a closer look precisely because the evidence here is unusually consistent.
August 25, 2026
PoradnikiTaurine and Longevity — What Do We Know, and What Don't We Know Yet?
In 2023, the journal Science published a headline-making study: declining taurine levels with age may drive aging, and supplementing it extended lifespan in mice and improved health in monkeys. Headlines quickly turned this into a "fountain of youth," but the study's own authors end their paper by saying that "clinical trials in humans seem warranted." We explain exactly what was shown in animals, what was merely correlated in humans, and why those two things are not the same.
August 25, 2026
Comments (2)
- KW
Kasia W. 2 weeks ago
Very clearly explained, especially the interactions section — I hadn't seen it laid out this well anywhere else.
- MT
Marek T. a month ago
Are you planning to update this with the newest study from this year? I saw an interesting meta-analysis.