Spit into a tube. That, in the not too distant future, might be all it takes for a police officer at the roadside to know whether you have been awake for the past 24 hours. Not a breathalyzer for alcohol, but something stranger: a chemical readout of how long it has been since you last slept. Researchers in Zurich have just shown the idea is more than wishful thinking.
Sleep loss is a peculiar kind of hazard. It dulls reaction times and wrecks coordination in ways that look uncomfortably like drunkenness, yet there has never been an objective way to measure it. A drunk driver leaves a number on a breathalyzer. A drowsy one leaves only an excuse.
That gap matters because drowsy driving is implicated in tens of thousands of crashes in the United States every year. Some states have written laws to deter it: New Jersey’s Maggie’s Law lets prosecutors treat 24 hours of continuous wakefulness behind the wheel as reckless driving. The trouble is enforcement. Without a test, the system leans entirely on what tired drivers are willing to admit, which is to say, not much.
So Thomas Kraemer and his colleagues at the University of Zurich went looking for sleeplessness in saliva. “Until now, sleep deprivation has been impossible to measure biochemically,” says Kraemer, the study’s corresponding author, “and yet it is one of the greatest burdens of our time.”
A fingerprint left in the spit
The team recruited 20 healthy young men who normally sleep somewhere between seven and nine hours a night. Each man went through three scenarios, in random order, roughly a week apart: a full night with no sleep at all, four consecutive nights trimmed to six hours, and a well-rested control of around eight hours. Saliva was collected repeatedly, before and after, then run through liquid chromatography and mass spectrometry to read off its chemical contents. The choice of saliva was deliberate. Blood draws are invasive and, for a roadside officer, legally awkward; spit is quick, cheap and already used worldwide for drug testing.
What came back was a haystack. Some 6,000 molecular features turned up across the 440 samples. Buried in there, though, was a signal.
After a single sleepless night, the saliva carried a distinct chemical signature that simply was not present in the rested samples. The researchers trained a machine-learning model to spot it, then whittled the panel down to just 12 molecules, few enough to imagine in a portable test one day. The model was strict in a way that matters for the law: when it called a sample sleep-deprived, it was right almost every time (the team tuned it that way, on the principle that wrongly accusing a rested person is worse than missing a tired one). It did let some genuinely sleep-deprived samples slip past, catching a bit over three-quarters of them. And it managed all this without a reference reading from the same person, which is the quietly important part. In the real world you rarely have a well-rested sample from the same driver sitting in a freezer for comparison.
The restricted-sleep condition, oddly, produced nothing usable. Four nights at six hours, an accumulated deficit equal to one missed night, left no exploitable fingerprint. Chronic grogginess, it seems, hides better than the acute kind.
There were limits, and the team is candid about them. The fingerprint shone brightest in the morning and around midday, then faded toward late evening. The reason is a tug-of-war: as the night wears on, the rising pressure to sleep and the body’s circadian clock briefly pull against each other, blurring the signal just when a tired driver might most need catching. A test taken at eight in the morning would be on far surer ground than one at eleven at night. And a handful of men, even after a full eight hours of recovery sleep, did not return to a properly rested chemical profile, a hint that one night in bed may not wipe the slate clean for everyone.
From 20 men to the wider world
The obvious caveat is the cohort. Twenty young men, all healthy, all on regular day-night schedules, is a narrow slice of the driving public, and a deliberate one: young men are the group most likely to crash from drowsiness, which is partly why the researchers started there. Even so, women, shift workers, the elderly, anyone on medication or stimulants, none of them were tested, and any of them might carry a different metabolic story. The authors are upfront that these models are investigational, built to probe what is possible rather than to start convicting people tomorrow.
Still, the direction of travel is clear. Kraemer’s group is now scaling up to an international study of more than 1,000 samples, drawn this time from shift workers, women and frequent drivers, the very groups the first study left out. If the sleepiness fingerprint survives that broader scrutiny, the question stops being whether your spit can betray a sleepless night and becomes what we ought to do once it can.
Source: Journal of Proteome Research, DOI: 10.1021/acs.jproteome.5c01064
Frequently Asked Questions
Can a saliva test really tell if someone is sleep deprived?
In this study, a machine-learning model reading 12 molecules in saliva flagged sleep-deprived samples with high reliability, and when it did call a sample sleep-deprived it was correct about 94 percent of the time, all without needing a comparison sample from the same person. It is an early proof of concept rather than a roadside-ready device, but it is the first direct biochemical signal of acute sleep loss found under realistic conditions.
Why didn’t restricted sleep show up the same way?
Participants who slept just six hours for four nights racked up a sleep deficit equal to one fully missed night, yet their saliva showed no usable chemical change against the rested state. Acute, total sleep loss appears to leave a sharper metabolic mark than chronic, moderate shortfall, which may simply hide better in the body’s chemistry.
Could this be used to prosecute drowsy drivers?
Not yet. The researchers stress the models are investigational and were tested only on 20 healthy young men on regular schedules. Laws like New Jersey’s Maggie’s Law already treat 24 hours of wakefulness behind the wheel as reckless, so a reliable test would have real legal teeth, but far larger and more diverse trials are needed first.
What happens next with the research?
The team is launching an international study of more than 1,000 samples that includes women, shift workers and frequent drivers, all groups missing from the first round. The aim is to see whether the so-called sleepiness fingerprint holds up across a far broader and messier population before anyone considers practical use.
ScienceBlog.com has no paywalls, no sponsored content, and no agenda beyond getting the science right. Every story here is written to inform, not to impress an advertiser or push a point of view.
Good science journalism takes time — reading the papers, checking the claims, finding researchers who can put findings in context. We do that work because we think it matters.
If you find this site useful, consider supporting it with a donation. Even a few dollars a month helps keep the coverage independent and free for everyone.