VOC Analysis: From Laboratory Research to Home Microbiome Testing

The idea that you could learn something meaningful about a microbial ecosystem by analyzing the gases it produces once sounded implausible to analytical chemists trained on metagenomics and sequencing data. It was, on the face of it, too simple. Over the past fifteen years, that skepticism has largely been replaced by a substantial peer-reviewed literature. When Garner and colleagues ran headspace solid-phase microextraction and GC-MS on stool from healthy donors and patients with gastrointestinal disease, they identified 297 distinct volatile compounds, 44 of which were shared by 80 percent of subjects (Garner et al., FASEB Journal, 2007). That study established that fecal headspace is a rich, structured chemical space rather than undifferentiated odor.

Volatile organic compound analysis has moved from a niche laboratory technique into one of the more interesting frontiers in precision medicine. What began with expensive, room-sized instruments has evolved into a plausible path toward at-home monitoring. This article traces that journey: the analytical chemistry, the landmark studies with their actual reported performance, the miniaturization problem, and the honest limits of where the technology stands.

From Industrial Detection to Medical Diagnostics

Volatile organic compounds have been detected since the early days of modern chemistry, but for most of that history the purpose was industrial: air quality, manufacturing contaminants, gas leaks. Gas chromatography, developed in the 1950s, became the standard for separating and identifying volatile compounds, and coupling it to mass spectrometry (GC-MS) gave scientists the ability to identify compounds with molecular precision.

In the 1990s researchers began asking a different question: could disease states be read from the volatile compounds emitted by biological samples? This was not simple repurposing. Industrial VOC detection is concerned with contamination thresholds. Medical VOC analysis is concerned with biomarkers, measurable biological signals that track health or disease. A 2013 review in Clinical Microbiology Reviews catalogued the state of that transition across infectious disease applications, from bacterial culture headspace to breath testing (Sethi et al., 2013).

The Laboratory Era

Early medical VOC work focused on breath. The logic was elegant, but breath analysis proved vulnerable to confounders: diet, medications, oral hygiene, respiratory infection, even circadian rhythm. The signal-to-noise ratio was often poor.

Attention then turned to other biological matrices. The observation that followed seems obvious in retrospect: the most complex microbial ecosystem in the human body is the gut, and the most direct way to sample its volatile output is through fecal analysis. Rather than measuring the microbiome indirectly through blood or breath, where its signal is diluted by countless other processes, researchers could measure it at the source.

The Chemistry of Microbial Volatility

The gut microbiota survives largely by fermenting the complex carbohydrates humans cannot digest. The primary end products of that fermentation are the short-chain fatty acids acetate, propionate and butyrate. Measurements of colonic contents reported by Cummings and colleagues put these three at roughly a 60:20:20 molar ratio, with acetate the most abundant (Cummings et al., Gut, 1987). Together they account for the large majority of the short-chain fatty acid pool in the colon.

Short-chain fatty acids are not the only output. Amino acid fermentation generates indole, skatole (3-methylindole) and phenolic derivatives. Some bacteria produce branched-chain fatty acids from leucine and valine. Some produce hydrogen sulfide. Some produce methane. The complete volatile output, the volatilome, is a biochemical fingerprint of both community composition and functional capacity.

This distinction matters. VOC analysis is not detecting odor in a casual sense. It is detecting the quantitative, reproducible metabolic output of enormous numbers of bacterial cells. The volatilome reflects not just what bacteria are present but what they are actively doing. A 2024 review in Nature Reviews Immunology sets out how these same short-chain fatty acids link diet to microbiome composition and to immune function (Mann et al., 2024).

Biomarker Compounds and Pathways

Research has associated particular VOC classes with particular microbiota configurations. In inflammatory bowel disease, for example, the proportion of sulfur-containing compounds tends to rise, which is biochemically coherent: inflamed tissue with altered oxygen conditions favors sulfate-reducing bacteria that generate hydrogen sulfide and related compounds.

Healthy microbiota, by contrast, typically show a profile weighted toward short-chain fatty acid production. Butyrate producers such as members of the genera Faecalibacterium and Roseburia behave as keystone organisms, and when butyrate production falls, the volatile profile shifts accordingly. Elevated branched-chain fatty acids and volatile phenolics point instead toward amino acid fermentation, which can suggest protein reaching the colon undigested or a shift away from carbohydrate-focused fermentation.

The table below summarizes the main compound classes reported in fecal headspace studies, the substrate they come from, and the organisms most often associated with them.

VOC class	Representative compounds	Substrate	Commonly associated organisms	What a rise typically indicates
Short-chain fatty acids	Acetate, propionate, butyrate	Fermentable fiber, resistant starch	Faecalibacterium, Roseburia, Eubacterium rectale, Bacteroides	Active saccharolytic fermentation
Branched-chain fatty acids	Iso-butyrate, iso-valerate	Leucine, valine, isoleucine	Amino acid fermenting Clostridium clusters	Protein fermentation, low fiber availability
Sulfur compounds	Hydrogen sulfide, dimethyl sulfide, dimethyl disulfide	Dietary sulfate, cysteine, methionine	Desulfovibrio and other sulfate reducers	Sulfur metabolism, mucus degradation, inflammation
Indoles	Indole, skatole (3-methylindole)	Tryptophan	Escherichia coli, Clostridium, Peptostreptococcus, Bacteroides	Proteolytic activity, higher protein intake
Phenolics	Phenol, p-cresol	Tyrosine, phenylalanine	Clostridium, Bacteroides	Aromatic amino acid fermentation
Aldehydes and ketones	Acetaldehyde, acetone, 2-butanone	Carbohydrate intermediates	Mixed fermentative community	Active carbohydrate turnover
Methane	Methane	Hydrogen and formate from bacterial fermentation	Methanobrevibacter smithii (archaea)	Methanogenesis, often with slower transit

Studies That Established the Foundation

Before fecal VOC analysis could move beyond curiosity, researchers had to demonstrate that VOC profiles are reproducible, that they differ meaningfully between health and disease, and that those differences are not artifacts of the analytical method.

Work on fecal volatile organic compounds as early diagnostic biomarkers found that VOC patterns could distinguish preterm infants at risk of necrotizing enterocolitis and late-onset sepsis, with differences reported in the days preceding clinical onset (Berkhout et al., Expert Review of Gastroenterology & Hepatology, 2018). That review is important because it deals with a population where the alternative is invasive and where timing genuinely matters. Separately, a 2023 assessment in Microorganisms showed that GC-MS analysis of fecal samples could discriminate between patient populations with distinct gastrointestinal disease states and healthy controls (Dalis et al., 2023).

Reported Diagnostic Performance

The most useful synthesis to date is a 2024 systematic review and meta-analysis in the Journal of Crohn's and Colitis. Krishnamoorthy and colleagues reviewed 16 studies and meta-analyzed 10, covering 696 inflammatory bowel disease cases against 605 controls. The pooled sensitivity was 87 percent (95 percent CI 0.79 to 0.92) and pooled specificity 83 percent (95 percent CI 0.73 to 0.90), with an area under the curve of 0.92. That is a genuinely strong result for a non-invasive test, and it is also a pooled figure across heterogeneous methods, which is exactly why individual studies vary so much.

Individual studies illustrate that spread. Arasaradnam and colleagues studied 62 participants (48 with inflammatory bowel disease, 14 healthy controls) using an electronic nose and field asymmetric ion mobility spectrometry on urine, and reported 75 percent accuracy separating inflammatory bowel disease from controls, with sensitivity and specificity of 67 percent each when trying to separate ulcerative colitis from Crohn's disease (Inflammatory Bowel Diseases, 2013). In colorectal cancer, Bond and colleagues analyzed fecal headspace from 137 participants (60 with no neoplasia, 56 with adenomatous polyps, 21 with adenocarcinoma). Propan-2-ol alone gave an area under the ROC curve of 0.76; combined with 3-methylbutanoic acid the AUROC rose to 0.82, with sensitivity 87.9 percent and specificity 84.6 percent (Alimentary Pharmacology & Therapeutics, 2019). An earlier study by the same group showed that urinary VOC analysis could also discriminate colorectal cancer from controls, establishing that more than one sample matrix carries the signal (Arasaradnam et al., PLoS ONE, 2014).

Study	Matrix and method	Population	Reported performance
Krishnamoorthy et al., 2024 (meta-analysis)	Fecal, breath and urinary VOC, mixed platforms	696 IBD cases, 605 controls, 10 studies	Pooled sensitivity 87%, specificity 83%, AUC 0.92
Bond et al., 2019	Fecal headspace GC-MS	137 participants (21 adenocarcinoma)	AUROC 0.82, sensitivity 87.9%, specificity 84.6% for propan-2-ol plus 3-methylbutanoic acid
Arasaradnam et al., 2013	Urinary VOC, eNose and FAIMS	48 IBD, 14 controls	75% accuracy IBD vs control; 67% sensitivity and 67% specificity for UC vs CD
Zheng et al., 2024	Fecal shotgun metagenomics plus targeted ddPCR	5,979 fecal samples, 8 validation populations	AUC above 0.90 in the discovery cohort for both UC and CD models
Wirbel et al., 2019	Fecal shotgun metagenomics	969 metagenomes, 8 cohorts	Cross-study transfer AUROC around 0.8; improved by leave-one-study-out training

Linking Diet, Metabolism and Measurable Output

A 2024 study in Scientific Reports used automated headspace solid-phase microextraction with GC-MS to track how food components are metabolized by gut microbiota in an in vitro fermentation model, observing the relationship between microbial metabolism and VOC production in real time (Dell'Olio et al., 2024). Conducted in vitro rather than in patients, it nonetheless established proof of concept for connecting specific dietary substrates to measurable volatile outputs.

Research on ulcerative colitis reported distinct fecal VOC profiles that discriminated patients with an ileoanal pouch from those with an intact colon, including lower proportions of sulfide and branched-chain fatty acids in the pouch (Yao et al., Molecular Nutrition & Food Research, 2025). The implication is important: altering intestinal anatomy produces measurable changes in the volatilome that reflect altered microbial composition and function. Animal work points the same way, with a murine short bowel syndrome model showing coordinated shifts in volatile organic compounds and bile acids as the microbial community changed (Wolfschluckner et al., Nutrients, 2023).

The link between the volatile fraction and the wider metabolome is also now well documented. In 786 individuals from the TwinsUK cohort, Zierer and colleagues measured 1,116 fecal metabolites and found host genetics explained only about 17.9 percent of variation while gut microbial composition explained on average 67.7 percent (standard deviation 18.8 percent). That is the theoretical basis for treating metabolite measurement as a functional readout of the microbiome (Zierer et al., Nature Genetics, 2018).

The Miniaturization Challenge

Translating this research into practice required solving an engineering problem: replicating laboratory-grade sensitivity and specificity in a device small, inexpensive and simple enough for home use. Laboratory GC-MS instruments are large, power-hungry, expensive, and require skilled operators. Those constraints are acceptable in a research laboratory and incompatible with at-home monitoring.

Three technical problems had to be addressed. Separation, where microfluidic gas chromatography columns trade some efficiency for a dramatic reduction in size. Detection, where alternatives such as flame ionization detection offer simpler, more power-efficient operation than mass spectrometry at the cost of specificity. And sampling, where laboratory workflows use solid-phase microextraction to preconcentrate volatiles, a step home devices must perform automatically and reliably.

Sensor Arrays and Electronic Noses

An alternative to miniaturizing GC-MS emerged: sensor arrays that detect VOC signatures without fully identifying every compound. These systems, often called electronic noses, use an array of chemical sensors with different selectivity patterns. Exposed to a complex mixture, each sensor responds differently, producing a characteristic pattern that can be associated with health or disease states.

A 2024 review in the Journal of Cancer Research and Clinical Oncology surveys electronic nose applications across gastrointestinal disease, covering colorectal cancer, inflammatory bowel disease and hepatic conditions, and notes that reported discrimination is frequently in the 0.8 to 0.9 AUROC range while cross-study reproducibility remains the field's weak point (Ma et al., 2024). The practical advantages are real: relatively inexpensive, portable, non-invasive, suitable for high-throughput analysis, and applicable across diverse patient demographics.

Miniaturization has progressed substantially on the hardware side, with capacitive micromachined ultrasonic transducer arrays and comparable microfabricated sensors supporting portable VOC detection with wireless transmission to a smartphone. The underlying principle across all of these systems is pattern recognition rather than absolute identification.

Machine Learning and the Interpretation Problem

Miniaturization solved one problem and created another. A sensor array generates readings from dozens of channels over time. Which patterns are meaningful, and which are noise?

Reviews of machine learning applications in microbiome analysis describe models that integrate metagenomics, metabolomics and clinical data, with reported classification performance varying substantially by dataset and validation design (Kumar et al., Frontiers in Microbiology, 2024). Deep learning approaches are well suited to VOC data specifically, because the relationships between sensor readings and health states are likely non-linear and context-dependent (Przymus et al., Frontiers in Microbiology, 2025). Research on multimodal data integration suggests that combining taxonomic, genetic and metabolomic data outperforms any single data type, which implies that the strongest at-home systems would pair VOC analysis with complementary measurements (Li et al., Microbial Cell Factories, 2022).

The caveat is external validation, and it is not a small one. In the largest cross-cohort test of microbiome classifiers to date, Wirbel and colleagues assembled 969 fecal metagenomes across eight colorectal cancer cohorts and showed that a model trained on one study and applied to another performs materially worse than its own cross-validation accuracy suggests, and that training on all other studies simultaneously recovers much of that loss (Nature Medicine, 2019). Any headline accuracy figure that has not survived that test should be read as provisional.

A remaining challenge is interpretability. A model that performs well but cannot explain its reasoning is a research curiosity rather than a clinical tool, which is why explainable AI methods matter here.

Clinical Validation Standards

Before any diagnostic moves into practice it must meet validation standards. Sensitivity measures the share of true cases correctly identified. Specificity measures the share of healthy individuals correctly identified. Both are summarized by the area under the receiver operating characteristic curve, where 0.5 is chance and 1.0 is perfect. Most established clinical tests sit somewhere in the 0.75 to 0.95 range, and the pooled 0.92 reported for VOC analysis in inflammatory bowel disease sits inside that band while resting on relatively small individual studies.

Validation proceeds in stages: proof-of-concept work establishing that an effect exists, analytical validation confirming the measurement itself is reproducible across operators and instruments, and clinical validation using larger prospective multi-center studies with blinded outcome assessment. For at-home devices, regulators also assess usability, safety and manufacturing consistency.

The 2024 Nature Medicine work by Zheng and colleagues shows what a mature version of this looks like in an adjacent modality. Working from 5,979 fecal metagenomes across multiple geographies and ethnicities, they selected ten bacterial species for an ulcerative colitis model and nine for a Crohn's disease model, achieved areas under the curve above 0.90 in the discovery cohort, held performance across trans-ethnic validation cohorts from eight populations, and then converted the signature into a droplet digital PCR assay practical for clinical laboratories. That is the evidence trajectory fecal VOC analysis has not yet completed. For a broader view of how this research area fits together, see our overview of gut microbiome science and VOC analysis.

Honest Limitations

It is worth being direct about where this technology stands. Fecal VOC analysis has demonstrated genuine promise in research settings. Multiple peer-reviewed studies show that VOC profiles distinguish disease states from healthy controls. But most VOC-based approaches remain research tools rather than clinical tests. No major diagnostic company currently offers a clinical VOC test for microbiota assessment, and no major clinical guideline recommends VOC analysis as a standard diagnostic approach.

Technical Constraints

Miniaturized sensors have lower sensitivity and selectivity than laboratory instruments, so an at-home device may miss low-abundance compounds visible on a research GC-MS. Standardization is a serious gap: different groups use different collection methods, analytical techniques, preprocessing steps and analysis approaches, which makes cross-study comparison difficult and is the reason the Krishnamoorthy meta-analysis had to pool across quite different platforms. Environmental factors matter more than early researchers anticipated, with storage conditions, temperature, humidity and even container material influencing profiles.

Biological Complexity

The volatilome is informative about metabolism but is not a complete picture of composition or function. The same VOC profile could in principle arise from different microbiota compositions producing similar metabolic output, a version of what systems biologists call the metabolic equivalence problem. The volatilome is also shaped by diet, medications, stress, transit time and inflammation. Those are genuine biological information rather than pure noise, but they mean interpretation requires clinical context.

What the Field Still Needs

• Larger prospective multi-center clinical validation studies with blinded outcome assessment and adequate statistical power.
• Consensus standards for sample collection, storage, analytical technique and data preprocessing.
• Longitudinal data on how VOC profiles change over time in healthy individuals and in response to interventions.
• Mechanistic validation connecting specific bacteria to specific volatile outputs under defined conditions.
• Health equity research, since most published work has been conducted in relatively affluent populations in industrialized nations.

Where the Technology Is Heading

The most interesting direction is not diagnosis but longitudinal monitoring. Diagnostic tests answer what is wrong right now. Monitoring answers how a system is changing. Today you might obtain one or two metagenomic analyses a year; between those snapshots, changes in your microbiota are invisible.

Hypothetical scenario. Consider a hypothetical case built only from the mechanisms described above. A person increases fermentable fiber intake substantially over four weeks. Their stool form does not change. On a laboratory GC-MS panel of the kind used by Garner and colleagues, the proportion of short-chain fatty acid derived volatiles rises while branched-chain fatty acids and indolic compounds fall, consistent with a shift from proteolytic toward saccharolytic fermentation. Nothing about this scenario reflects a real person or a real SNIFR result. It illustrates the direction of change the published chemistry would predict, which is precisely the kind of change a single annual sequencing snapshot would miss.

An at-home VOC monitoring approach could offer feedback on how the microbiota responds to fiber intake, stress, medications or fermented foods. Combined with occasional sequencing and dietary tracking, this kind of multi-omic integration would also help address the metabolic equivalence problem by pairing functional data with compositional data.

Therapeutic monitoring is a related frontier. If someone begins a dietary change or a probiotic, symptom improvement is a crude and slow readout. Objective, quantitative feedback on whether the volatilome is shifting could in principle allow much faster adjustment. This is the design intent behind SNIFR's work on passive at-home monitoring and, longer term, on flare-up prediction as an early warning system. It is a development goal rather than a demonstrated capability, and it has not been clinically validated.

On the technical side, improvements in materials science and nanotechnology should increase sensor sensitivity and selectivity while lowering cost, and machine learning models should become more interpretable and more robust to real-world variability. The question is no longer only whether the technology can work in principle. It is whether the field can mature it, and validate it, quickly enough to matter.

Frequently Asked Questions

What is VOC analysis in gut microbiome testing?

VOC analysis measures the volatile organic compounds that gut bacteria release as they metabolize food, giving a functional readout of what the microbiome is actively doing. Garner and colleagues identified 297 distinct volatiles in human fecal headspace in 2007, 44 of them shared by most subjects. Unlike DNA sequencing, it captures metabolic output rather than membership.

How accurate is volatile organic compound analysis for health?

A 2024 meta-analysis in the Journal of Crohn's and Colitis pooled 10 studies covering 696 inflammatory bowel disease cases and 605 controls, reporting sensitivity 87 percent, specificity 83 percent and an area under the curve of 0.92. Individual studies range far more widely. No fecal VOC test is yet an approved clinical diagnostic.

Can you really learn about gut health from stool gas?

Yes, because stool gas is the direct chemical output of bacterial fermentation in the colon. When bacteria ferment fiber and amino acids they release short-chain fatty acids, sulfur compounds, indoles and other volatiles whose relative proportions shift with diet and microbial composition. Reading those proportions is the basis of fecal VOC research.

Why did VOC research move from breath testing to stool testing?

Breath samples mix volatile compounds from the whole body, which weakens the signal from the gut specifically. Researchers turned to fecal analysis because it samples the microbiome's metabolic output at the source, with far less physiological interference. Recent work suggests breath and gut volatile profiles are related, so both sampling routes remain active research areas.

What are the limitations of at-home gut microbiome monitoring?

The main limitations are standardization, sensor sensitivity, and biological context. Miniaturized sensors detect patterns rather than identifying every compound, sample handling and storage alter volatile profiles, and the same VOC signature can arise from different microbial communities. These are solvable engineering and research problems, but they are why the field remains pre-clinical.

Do VOC accuracy figures hold up in new populations?

Often they do not. Wirbel and colleagues assembled 969 fecal metagenomes across eight cohorts and showed that models trained on one study lose accuracy when transferred to another, recovering much of it only when trained across studies. The same caution applies to VOC classifiers, so single-cohort figures should be treated as provisional until externally validated.

What is the difference between diagnosis and monitoring in microbiome science?

Diagnosis asks what is wrong at a single moment; monitoring asks how a system is changing over time. Most microbiome science to date has been diagnostic and snapshot-based. Continuous at-home monitoring is attractive precisely because it can show a personal trajectory, which is where SNIFR's design work is focused.

References

• Garner CE, Smith S, de Lacy Costello B, et al. Volatile organic compounds from feces and their potential for diagnosis of gastrointestinal disease. The FASEB Journal. 2007;21(8):1675-1688. doi:10.1096/fj.06-6927com
• Cummings JH, Pomare EW, Branch WJ, et al. Short chain fatty acids in human large intestine, portal, hepatic and venous blood. Gut. 1987;28(10):1221-1227. doi:10.1136/gut.28.10.1221
• Berkhout DJC, Niemarkt HJ, de Boer NKH, et al. The potential of gut microbiota and fecal volatile organic compounds analysis as early diagnostic biomarker for necrotizing enterocolitis and sepsis in preterm infants. Expert Review of Gastroenterology & Hepatology. 2018;12(5):457-470. doi:10.1080/17474124.2018.1446826
• Dalis C, Mesfin FM, Manohar K, et al. Volatile Organic Compound Assessment as a Screening Tool for Early Detection of Gastrointestinal Diseases. Microorganisms. 2023;11(7):1822. doi:10.3390/microorganisms11071822
• Dell'Olio A, Rubert J, Capozzi V, et al. Non-invasive VOCs detection to monitor the gut microbiota metabolism in-vitro. Scientific Reports. 2024;14:15479. doi:10.1038/s41598-024-66303-7
• Yao CK, Green K, Day AS, et al. Distinct profiles of fecal volatile organic compounds discriminate ulcerative colitis patients with an ileoanal pouch from those with an intact colon. Molecular Nutrition & Food Research. 2025;69(7):e70003. doi:10.1002/mnfr.70003
• Arasaradnam RP, Ouaret N, Thomas MG, et al. A Novel Tool for Noninvasive Diagnosis and Tracking of Patients with Inflammatory Bowel Disease. Inflammatory Bowel Diseases. 2013;19(5):999-1003. doi:10.1097/MIB.0b013e3182802b26
• Arasaradnam RP, McFarlane MJ, Ryan-Fisher C, et al. Detection of Colorectal Cancer (CRC) by Urinary Volatile Organic Compound Analysis. PLoS ONE. 2014;9(9):e108750. doi:10.1371/journal.pone.0108750
• Bond A, Greenwood R, Lewis S, et al. Volatile organic compounds emitted from faeces as a biomarker for colorectal cancer. Alimentary Pharmacology & Therapeutics. 2019;49(8):1005-1012. doi:10.1111/apt.15140
• Krishnamoorthy A, Chandrapalan S, Ahmed M, et al. The diagnostic utility of volatile organic compounds in inflammatory bowel disease: a systematic review and meta-analysis. Journal of Crohn's and Colitis. 2024;18(2):320-330. doi:10.1093/ecco-jcc/jjad132
• Ma T, Chang Z, Zhang N, et al. Application of electronic nose technology in the diagnosis of gastrointestinal diseases: a review. Journal of Cancer Research and Clinical Oncology. 2024;150(8):390. doi:10.1007/s00432-024-05925-w
• Sethi S, Nanda R, Chakraborty T. Clinical Application of Volatile Organic Compound Analysis for Detecting Infectious Diseases. Clinical Microbiology Reviews. 2013;26(3):462-475. doi:10.1128/cmr.00020-13
• Zheng J, Sun Q, Zhang M, et al. Noninvasive, microbiome-based diagnosis of inflammatory bowel disease. Nature Medicine. 2024;30(12):3555-3567. doi:10.1038/s41591-024-03280-4
• Wirbel J, Pyl PT, Kartal E, et al. Meta-analysis of fecal metagenomes reveals global microbial signatures that are specific for colorectal cancer. Nature Medicine. 2019;25(4):679-689. doi:10.1038/s41591-019-0406-6
• Kumar B, Lorusso E, Fosso B, et al. A comprehensive overview of microbiome data in the light of machine learning applications: categorization, accessibility, and future directions. Frontiers in Microbiology. 2024;15:1343572. doi:10.3389/fmicb.2024.1343572
• Przymus P, Rykaczewski K, Martin-Segura A, et al. Deep learning in microbiome analysis: a comprehensive review of neural network models. Frontiers in Microbiology. 2025;15:1516667. doi:10.3389/fmicb.2024.1516667
• Li P, Luo H, Ji B, et al. Machine learning for data integration in human gut microbiome. Microbial Cell Factories. 2022;21(1):241. doi:10.1186/s12934-022-01973-4
• Mann ER, Lam YK, Uhlig HH. Short-chain fatty acids: linking diet, the microbiome and immunity. Nature Reviews Immunology. 2024;24(8):577-595. doi:10.1038/s41577-024-01014-8
• Wolfschluckner V, Obermüller B, Horvath A, et al. Metabolomic Alterations of Volatile Organic Compounds and Bile Acids as Biomarkers of Microbial Shifts in a Murine Model of Short Bowel Syndrome. Nutrients. 2023;15(23):4949. doi:10.3390/nu15234949
• Zierer J, Jackson MA, Kastenmüller G, et al. The fecal metabolome as a functional readout of the gut microbiome. Nature Genetics. 2018;50(6):790-795. doi:10.1038/s41588-018-0135-7

SNIFR is designed to provide insights about gut health patterns, not to diagnose or treat medical conditions. Individual results may vary as gut health is influenced by numerous factors including diet, stress, sleep, and genetics. SNIFR is currently in development, and features described may evolve before commercial release.