How breath reveals your metabolism

Where your energy comes from

Your body is designed to run on the glucose from your last meal and the fat you have already stored, moving between them continuously. After eating, glucose is plentiful, insulin rises, and the body burns what has just arrived while storing the surplus. As glucose runs down, insulin falls and the body begins breaking stored fat back into fuel.

Which of the two it is drawing on right now is the difference between a diet that is working and one that is not. Weight is a lagging, noisy proxy for it. Nothing you can see or feel reports it directly.

When your body switches fuel

Metabolic reactions produce by-products. Some are volatile — small enough and light enough to cross from blood into the air in your lungs and leave on the next exhale. Concentration in breath then tracks concentration in blood, which makes an exhale a continuous, non-invasive sample of what is happening inside.

More than a thousand such compounds have been catalogued in human breath. Rysflo reads three of them, chosen because each maps cleanly onto a different system.

Are you burning fat?

When you burn fat steadily, it breaks down into fragments faster than your cells can use them up. The liver packs the surplus into ketones. One of those ketones is unstable and falls apart on its own into acetone — a molecule small and light enough to slip out of your blood into your lungs and leave on the next breath.

Reported concentrations sit at roughly 0.5–2 ppm at baseline, 2–5 ppm under caloric restriction, and higher in nutritional ketosis. Across studies the relationship with blood beta-hydroxybutyrate is exponential rather than linear, and the fit explains about 77 per cent of the variance (R² = 0.77, individual studies 0.54 to 0.94).

Why the reading is taken fasted. Breath acetone tracks blood ketones closely while fasting, and stops tracking them once you have eaten — in one crossover study the correlation held at ρ = 0.58 overnight and disappeared after both high- and low-carbohydrate meals. Acetone also forms by irreversible decarboxylation and clears slowly, so a morning reading summarises the overnight window rather than reacting to the last thing you ate. The fasted protocol is the condition in which the marker behaves.

The fit is strong but not exact, and it is exponential, so breath acetone is best read as a directional trend marker: the direction of travel is reliable, an exact rate of fat loss is not.

Did your gut absorb the meal?

Human cells do not produce molecular hydrogen. All breath H₂ is microbial in origin. Carbohydrate that escapes small-intestinal absorption reaches the colon, where anaerobes ferment it and release hydrogen; a fraction — historically estimated near fourteen percent — is absorbed into blood and exhaled.

This is the oldest of the three in clinical use. Hydrogen breath testing has been a standard for carbohydrate malabsorption since 1969, and the 2017 North American Consensus defines a rise of at least 20 ppm above baseline as a malabsorption signal. Lactose malabsorption affects a large share of adults worldwide; fructose malabsorption is also common.

The known limit: a substantial minority of people are methanogen-dominant, converting hydrogen to methane. A hydrogen-only reading under-represents their fermentation. Adding a methane channel is the planned refinement, and until then those readings should be interpreted with that in mind.

Is sugar fermenting in your gut?

Gut commensals including Klebsiella, E. coli and Candida produce small quantities of ethanol continuously. In healthy people first-pass hepatic metabolism clears almost all of it, so peripheral blood and breath concentrations remain very low. Portal-vein ethanol has been shown to run many times higher than peripheral, and to rise stepwise with liver-disease severity.

This is the least settled of the three. The mechanism is robust and non-alcoholic fatty liver disease is highly prevalent, including in lean individuals. But resolving breath ethanol at these concentrations is analytically demanding, and fermented foods, ripe fruit, sugar alcohols and the oral microbiome all interfere. It is an emerging research marker, not a diagnostic one.

Why three, and not one

Each of the three reports on a different system, and each rests on established biochemistry. Reading all three from a single exhale is what we built, and it is the part that has to be proved by our own work rather than anyone else’s. The precedent is continuous glucose monitoring: Zeevi and colleagues, and later the PREDICT study, showed that the same food produces substantially different responses between individuals, and that nutrition guided by continuous measurement outperforms generic dietary guidance in controlled trials.

Those trials establish the principle of personalisation and validate specific read-outs such as continuous glucose. They do not validate breath acetone, hydrogen and ethanol as guiding signals.

What this cannot tell you

Rysflo uses an array of MEMS metal-oxide gas sensors. Each element sits on a microfabricated hotplate that heats a thin metal-oxide film; target gases react at the film surface and change its electrical resistance, which is read as the signal. The same sensing class has been shown to agree closely with reference mass spectrometry for handheld acetone measurement.

Selectivity is the central design problem. A single metal-oxide film responds to many reducing gases, and acetone, hydrogen and ethanol all elicit a response, so the three signals must be separated rather than read off one element. The array uses different oxide chemistries and operating temperatures with pattern recognition across their joint response, and each analyte requires its own validation rather than inheriting the acetone result. These films are also sensitive to humidity, ambient temperature and slow baseline drift, so periodic calibration and standardised sampling — a fasted morning reading, a ventilated room, a full hotplate warm-up — are part of the design rather than optional extras.