People can be identified by their breathing patterns with 97% accuracy

Humans have unique breathing patterns that can be used to identify and distinguish individuals, a new study has found.

In the work, published Thursday (June 12) in the journal Current Biology, researchers could use an algorithm to identify individuals based on these distinct "respiratory fingerprints" nearly 97% of the time. The study authors also think the breathing profiles could reveal potential clues about each individual's mental and physical health.

Although it seems straightforward, breathing is a complex process governed by several brain regions. This led to the researchers' hypothesis: "Given that we all have unique brains, we hypothesize that this uniqueness will also be reflected in our breath," first study author Timna Soroka, a doctoral student at the Weizmann Institute of Science in Israel, told Live Science in an email.

To test their hypothesis, Soroka and colleagues recruited 97 volunteers and fitted each person with a wearable device that measures inhalation and exhalation through tubes positioned beneath each nostril. The device logged each participant's breathing for 24 hours while they went about their regular activities, including sleep.

The researchers then used a software program to identify dozens of distinct features across the breathing logs. "These can be very seemingly obscure temporal features, like the duration of the pause before your inhale, or the duration of the pause after your inhale," said study co-author Noam Sobel, a professor of neurobiology at the Weizmann Institute of Science.

Splitting each 24-hour sample into five minute segments, the researchers tracked how these features varied across each recording cycle for each participant. Next, they used machine learning to analyze the resulting trait "maps," revealing that the pattern of variability was unique for each person.

Related: Why can't you suffocate by holding your breath?

They then ran an additional experiment on 42 of the participants, who each underwent an extra day of testing at a random point within a two-year period — and this suggested that the pattern of variability for each individual was strikingly similar across the two testing days, while remaining distinct from anyone else. In other words, each individual had a unique breathing "signature," akin to a fingerprint.

The signatures were distinct enough for the researchers to train a machine learning algorithm to identify individual participants with 96.8% accuracy, based on their breathing patterns alone.

The team had not expected the results to be so statistically robust, Sobel said. "This made us go through [the findings] many, many times, because it was almost too good to be true, in terms of the power. So that was surprising."

The researchers also asked the participants to fill out a questionnaire about their physical and mental health. When they combined the results of those questionnaires with the respiratory analysis, they found correlations between people's breathing profiles and various characteristics, such as a person's self-reported feelings of anxiousness or their body mass index.

For example, individuals who reported higher feelings of anxiety had greater variability in the pauses between their breaths than people who reported lower levels, said Soroka. This raises the possibility that breathing patterns could potentially be used to predict certain aspects of mental and physical health, the paper stated.

Renato Zenobi, a professor of analytical chemistry at ETH Zurich, has investigated the use of chemicals — namely metabolites — in human breath as disease indicators. Zenobi, who was not involved with the new study, said its results might help to strengthen breath tests as a more convenient diagnostic approach.

"Some diseases you diagnose with questionnaires," Zenobi said. "But if one would have something metabolic to measure, plus the breathing pattern, it could be stronger, more robust."

Notably, though, for any test to be useful for diagnosis, the measurements it's taking must be confirmed through controlled studies to be a clear and consistent indicator of a given disease. More research would be needed to validate breathing patterns as a reliable diagnostic tool. Zenobi questioned the logistics of how breath monitoring would be incorporated into clinical practice.

He did find the study "innovative," but also questioned the practicality of the breath-tracking device, given it requires hours of wear to get clear results. This could make it cumbersome to use and decrease the results' reliability, as it required participants to take the device home, he said.

The researchers say they are now working to make their device more comfortable to wear. As they take the work forward, Sobel said they are also contemplating some big, overarching questions, including whether the breathing patterns reflect or are a driver of a person's brain state.