Microbiome Data as a Nutrition Input
Whether gut microbiome measurements can guide food choices: what the main studies found, where they disagree, and why a stool report is not a diet prescription.
Content last updated . This page has not been medically reviewed.
What other studies add
- A large UK study (PREDICT 1, 1,002 people) found that the gut microbiome explained about 7.1% of the variation in blood fat response after a meal and 6.0% of the variation in blood sugar response, while meal makeup mattered more for blood sugar (15.4%). [3] Microbes were therefore one modest influence among several. A publisher correction exists for this paper, so check the journal version. [4]
- A US replication (327 people) found that a model built on the Israeli cohort predicted Midwestern participants' responses better than carbohydrate or calorie counts alone, and performance improved when the model was retrained on both groups. [5] An erratum has been published for this paper too. [6]
- Weight-loss prediction is inconsistent. A post hoc analysis of 52 completers reported that people with a high Prevotella-to-Bacteroides ratio lost more weight on a higher-fiber diet. [7] In a later trial of a plant-based diet, baseline ratio showed no relationship with weight change. [8] Label: mixed.
- A 10-week trial of personalized plans (100 adults) built on metabolic and genetic clusters found no difference in fat mass versus a generic healthy diet. [9]
Why microbiome results are hard to use
Stool results depend on the lab. In a comparison of 15 laboratories, variability depended most on the type and origin of the sample, followed by DNA extraction, sample handling and the analysis software. [10] A single result therefore cannot be assumed to be comparable between companies. See Commercial Stool Tests and Microbiome Tests: What They Can and Cannot Tell You.What has not been shown
- That changing a diet to match a microbiome profile improves long-term health outcomes.
- That a "good" or "bad" microbe list is dependable. See Why There Is No Single Universal Healthy Microbiome.
- That consumer microbiome reports add accuracy beyond ordinary dietary guidance. For a plain-language view, see Microbiome-Based Recommendations.
Frequently asked questions
Can a stool test tell me which foods to eat?
Not reliably. The best-known studies used microbiome data as one of several inputs to predict blood sugar response, not to prescribe a diet for health. [1] Lab-to-lab differences also limit comparability. [10]
How much does the microbiome explain about food responses?
In a UK study of 1,002 people, the microbiome explained about 6.0% of variation in blood sugar response after meals, compared with 15.4% for meal makeup. [3]
Does a high Prevotella-to-Bacteroides ratio predict weight loss?
Results conflict. One post hoc analysis said it did, and a later trial found no relationship. [7] [8]
Has microbiome-guided eating been shown to improve long-term health?
No long-term outcome evidence was found in the sources read for this page. Proper evidence not found; checked PubMed for trials and reviews of microbiome-guided diets.
More details: evidence, limits and cautions
The details behind this page: cautions, limits of the evidence and where sources disagree. Open any note to read it.
How to read the evidence labels
Labels used on this page: promising but limited, mixed, insufficient evidence and marketing claim.
What the research tested
In 2015 an Israeli team followed 800 people with glucose sensors, recorded 46,898 meals, and built a machine-learning model that combined blood tests, diet habits, body measures, activity and gut microbes to predict each person's glucose response to real meals. [1] The model's microbiome inputs were one part of a larger set, so the study does not show that microbes alone drive the predictions. Label: promising but limited. Details of the glucose work are on Personalized Glycemic Predictions.A 2017 crossover trial compared sourdough whole-grain bread with industrial white bread for one week each. On average there was no clear difference in clinical measures, but individuals responded differently, and the lower-response bread could be predicted from microbiome data collected before the trial. [2] This was a small, short study in which the prediction was about glucose response to bread, not about long-term health. Label: promising but limited.
Safety note
References
- Zeevi et al., Cell 2015, Personalized Nutrition by Prediction of Glycemic Responses. Read 2026-10-22 (abstract on PubMed).
- Korem et al., Cell Metabolism 2017, Bread Affects Clinical Parameters and Induces Gut Microbiome-Associated Personal Glycemic Responses. Read 2026-10-22 (abstract on PubMed).
- Berry et al., Nature Medicine 2020 (PREDICT 1), Human postprandial responses to food and potential for precision nutrition. Read 2026-10-22 (abstract on PubMed).
- Publisher Correction to Berry et al., Nature Medicine 2020. Read 2026-10-22 (abstract on PubMed).
- Mendes-Soares et al., Am J Clin Nutr 2019, Model of personalized postprandial glycemic response developed for an Israeli cohort predicts responses in Midwestern American individuals. Read 2026-10-22 (abstract on PubMed).
- Erratum to Mendes-Soares et al., Am J Clin Nutr 2019. Read 2026-10-22 (abstract on PubMed).
- Hjorth et al., Int J Obesity 2019, Prevotella-to-Bacteroides ratio predicts body weight and fat loss success on 24-week diets. Read 2026-10-22 (abstract on PubMed).
- Frontiers in Nutrition 2023, Supplementation with inulin-type fructans (PREVENTOMICS secondary analysis). Read 2026-10-22 (abstract on PubMed).
- Clinical Nutrition 2022, A double-blinded, randomized, parallel intervention to evaluate biomarker-based nutrition plans for weight loss: the PREVENTOMICS study. Read 2026-10-22 (abstract on PubMed).
- Microbiome Quality Control (MBQC) project consortium, Nature Biotechnology 2017. Read 2026-10-22 (abstract on PubMed).