Science

A research team compressed the gut microbiome into one number — here is what it tracked

Medically reviewed by Longevity Benchmark Editorial Last reviewed 2026-08-24

A Microbial Health Index built from 128 bacterial variants correlated with diet, fitness and lipids in 893 adults. No accuracy figures were reported.

August 24, 2026 · Science · Gut health · Microbiome testing

What was reported

Writing in Medical Xpress about her group's own paper in Gut Microbes, Yael Haberman — a professor of pediatric gastroenterology at Tel Aviv University — describes a Microbial Health Index, or MHI, built as the log ratio of 97 health-associated bacterial variants to 31 disease-associated variants. The group applied it to 893 individuals who underwent routine annual Executive Medical Checkups at a tertiary medical center, profiling stool by 16S sequencing of the V4 region and pairing it with food-frequency and lifestyle questionnaires, electronic medical record data, and cardiac stress testing. Higher index values went with "high fiber intake, adherence to a Mediterranean diet and better physical fitness," including a positive correlation with METs on stress testing, plus improved blood lipid profiles. Ultra-processed food consumption and higher subclinical inflammation went with lower scores.

The short version

Read this as support for the concept and silence on the products. Compressing a microbiome into a single ratio produced a number that moved in the directions you would expect, which is a real and non-trivial result — plenty of composite scores fail that test. But every measure the index tracked with is something a reader can already assess directly: what they eat, how fit they are, their lipids, their inflammatory markers. The source reports no accuracy figures at all, and it is not a consumer product. For anyone weighing a microbiome test, this changes the question, not the answer.

What it means for you

The commercial microbiome tests we review sell a score. That is the actual product — a number, a grade, or a set of subscores, derived from sequencing a stool sample and compared against a reference population the company assembled. The persistent difficulty in evaluating them is that the reference sets and the scoring algorithms are proprietary, so the scores cannot be independently checked and generally have no published validation of their own. Research like this is the closest thing available to an outside read on whether the underlying idea holds up, which is why it is worth a briefing even though no product is involved.

What it supports is modest and worth stating precisely: in one cohort, a transparently defined ratio of named bacterial groups moved together with diet quality, measured fitness and lipid profiles. That argues the microbiome carries a legible signal about lifestyle. It does not argue that the signal adds anything on top of measuring the lifestyle. If a score is high largely because someone eats fiber, follows a Mediterranean pattern and has good cardiorespiratory fitness, a person who already knows those three things about themselves has learned little from the sequencing. That is the central question for a buyer, and this study does not attempt to answer it — the analysis reported here is about associations, not about incremental value over cheaper information.

The second thing to hold onto is what is absent. There is no reported AUC, no sensitivity or specificity, no threshold at which a score means one thing rather than another. Hypertension and metabolic syndrome come up as examples of what the group looked at, but a composite that correlates with a condition across a group of people is categorically different from a test that classifies an individual — and only the second is what a consumer test implicitly promises when it returns a personal score. Our page on whether gut health tests are accurate walks through why that gap is so persistent in this category, and how clinicians actually test gut health covers what gets ordered when a specific problem is being investigated.

Finally, note the population. These were adults presenting for annual executive health checkups at a tertiary center — a self-selected group with the means and inclination to buy comprehensive screening, which is unusually similar to the readership of a site like this one, and unlike the general population. That makes the findings more transferable to our readers than most cohort work, and simultaneously means they should not be read as describing everyone. If you are choosing between the major products, our Viome versus ZOE comparison and the best gut health test guide take up the practical question of what each one actually returns for the money.

What this cannot tell you

These are correlations reported in a single cohort of 893 people, described by the study's own senior author in a general-audience article; the paper itself sits behind a paywall, and the available source does not state whether participants were followed over time or measured at one point. Correlation does not establish that changing the microbiome changes health, or that the direction runs from bacteria to outcome rather than from diet and exercise to both. No accuracy or significance figures were published in the source we read, so none appear here. The index is a research tool, not a test you can order, and nothing in it indicates whether any testing is appropriate for you — that is a question for a clinician who knows your history.

How this fits what we already know

This sits alongside a pattern we keep documenting across categories: a composite score is easy to build and hard to prove useful. Our briefing on epigenetic clocks underperforming conventional risk factors found the same shape of problem in biological-age testing, where a sophisticated molecular score lost to a tape measure and a smoking history at predicting who got sick. The microbiome field is earlier and the present study is a constructive step in it — a defined, inspectable index rather than a proprietary one. The standard we apply on the gut health hub stays the same either way: a score has to beat the cheaper thing you could have measured instead, and until someone shows that, breadth of sequencing is not the same as depth of insight.

Sources

Haberman Y., "A gut microbial health index as a composite biomarker for evaluating overall health," Medical Xpress, August 23, 2026, medicalxpress.com (accessed August 24, 2026). Author-written account of the study; source for the index definition, cohort size, sequencing method and every association described above.
Gut Microbes, 2026. DOI 10.1080/19490976.2026.2719055 (paywalled; not accessible to us at the time of writing, so no figures are taken from it).

Frequently Asked Questions

What is the Microbial Health Index?

It is a research score, not a product. As described by its senior author, the MHI is calculated as the log ratio of 97 health-associated bacterial variants to 31 disease-associated bacterial variants found in a stool sample, reducing a whole microbiome profile to a single number. It was reported in Gut Microbes in 2026 from a cohort of 893 adults, using 16S sequencing of the V4 region.

Does this mean a consumer microbiome test can score my gut health?

It does not establish that. The study reports that a composite index built by researchers tracked with diet, fitness and lipid measures in one cohort. It says nothing about the proprietary scores sold by consumer testing companies, which use their own reference sets and their own algorithms and are generally not published in a form that can be checked. A research index performing sensibly is evidence that the general approach is not meaningless; it is not evidence that any particular product works.

What did a higher score correlate with?

The author describes higher index values going with high fiber intake, adherence to a Mediterranean diet, better physical fitness — including a positive correlation with METs measured on cardiac stress testing — and improved blood lipid profiles. Lower scores went with ultra-processed food consumption and higher levels of subclinical inflammation. These are correlations reported in one cohort. The source we read does not publish correlation coefficients, p-values or accuracy figures, so we are not quoting any.

Can a score like this diagnose anything?

Nothing in the available source supports that. No diagnostic accuracy is reported — no AUC, no sensitivity or specificity, and no threshold separating people with a condition from people without one. Hypertension and metabolic syndrome are mentioned as examples of what the group examined, but a composite index correlating with a condition across a group is a long way from a test that classifies an individual. Any question about a specific condition belongs with a clinician.

Should this change whether I buy a microbiome test?

It gives you a sharper question to ask rather than an answer. The measures this index tracked with — what you eat, how fit you are, your lipids, your inflammatory markers — are all things you can already measure directly, most of them more cheaply and more reliably than by sequencing stool. So the question for any microbiome product is what its score tells you that those measurements do not, and what you would do differently on receiving it. Our reviews cover how the major products answer that.