Your Wearable Knows You’re Getting Sick Before You Do

By the time you feel the scratchy throat, the battle has usually been under way for a day or two. Your immune system does not wait for your permission to start fighting an infection — it mobilizes the moment it detects an invader, and that mobilization is metabolically expensive. The heart beats a little faster at rest. Breathing quickens slightly. Skin temperature edges up as the body prepares its fever response. The balance of the nervous system tilts toward alarm, and heart rate variability sinks. None of this is something you consciously feel — but a wearable that measures you every night can see it, and the difference between your usual Tuesday and this particular Tuesday is exactly the kind of pattern a continuous sensor is good at catching.

Over the past few years this went from a plausible idea to a demonstrated one. Researchers showed that smartwatch data could flag infections before symptoms appeared, that aggregated fitness-tracker data could sharpen flu surveillance across whole regions, and that under controlled conditions — volunteers deliberately exposed to respiratory viruses — wearable sensors could tell who was getting sick before those people knew it themselves. This article walks through what the research actually shows, how to read your own overnight numbers, what a deviation does and does not mean, and what to do (and not do) when your data says something is brewing. One framing rule runs through all of it, and it is worth stating before the first study: wearables screen, they do not diagnose.

The Signature of an Oncoming Infection

When your immune system detects a virus, it releases signalling molecules that raise your metabolic rate, recruit defensive cells, and begin nudging your core temperature upward. That response has a readable signature. Resting heart rate rises, because a body fighting an infection needs more oxygen delivered per minute even while lying still. Respiratory rate rises for the same reason. Skin temperature climbs as the fever machinery warms up. And heart rate variability — the beat-to-beat variation that reflects a calm, recovered nervous system — typically falls, because the autonomic balance shifts toward the “fight” setting.

Night-time is when this signature is easiest to read. Overnight, the confounders of daily life — stairs, meetings, coffee, arguments — are stripped away, and what remains is close to a controlled measurement of your physiology at rest. That is why most modern devices build their baselines from sleep data, and why the morning summary is the number worth watching rather than any reading taken mid-afternoon.

The sensor side has matured, too. Resting heart rate has been standard on wrist devices for a decade, but overnight respiratory rate and skin-temperature deviation are now common as well, and HRV is measured by most recent smartwatches — including the ones many readers already own. In other words, the four channels the illness signature runs through are already on millions of wrists, recorded every night, whether or not anyone looks at them.

Crucially, the immune signature often appears before symptoms. Congestion, sore throat, and cough are largely produced by the immune response itself, and they lag its beginning by a day or more. Researchers who analysed wearable data from people who later tested positive for COVID-19 found exactly this pattern: respiratory rate and resting heart rate ran measurably above each person’s baseline during illness, and in a meaningful share of cases the deviation was already visible in the days before symptom onset.

Natarajan, A., Su, H.W. & Heneghan, C. (2020). “Assessment of physiological signs associated with COVID-19 measured using wearable devices.” npj Digital Medicine, 3, 156.

The Studies That Put It to the Test

The best-known demonstration came from Stanford during the pandemic. Researchers collected historical smartwatch data from people who went on to test positive for COVID-19 and asked a simple question: did the physiology deviate before the person felt sick? In a majority of the infections that left a detectable trace in the data at all, the answer was yes — resting heart rate and related signals drifted away from the individual’s own baseline before symptom onset, sometimes by several days. The comparison against each person’s own history is the key: the algorithm was not asking whether your heart rate was high for a human, but whether it was high for you. The same work went a step further and prototyped a real-time alerting approach on the same principle — watch the personal baseline, flag the sustained deviation — which is essentially what consumer platforms have been building toward since.

Mishra, T. et al. (2020). “Pre-symptomatic detection of COVID-19 from smartwatch data.” Nature Biomedical Engineering, 4(12), 1208–1220.

Retrospective studies have a weakness, though: you only analyse the people who got sick, after you know they got sick. The strongest evidence comes from controlled viral-challenge research, where healthy volunteers are deliberately exposed to influenza or a common-cold virus under medical supervision and monitored continuously from before the exposure. That design removes the guesswork — the researchers know exactly who was exposed, when, and who went on to develop an infection. In that setting, wearable sensor data distinguished the people who were becoming infected from those who were not before symptoms appeared. The infection was visible in the physiology first, and in how people felt second.

Grzesiak, E. et al. (2021). “Assessment of the Feasibility of Using Noninvasive Wearable Biometric Monitoring Sensors to Detect Influenza and the Common Cold Before Symptom Onset.” JAMA Network Open, 4(9), e2128534.

From One Wrist to Public Health

The same signal scales up. Traditional influenza surveillance depends on people getting sick enough to visit a clinic, and on those visits being counted and reported — a pipeline with a built-in delay. Scripps researchers asked whether ordinary bedside data could shortcut it: they analysed de-identified resting heart rate and sleep data from tens of thousands of fitness-tracker users across the United States and found that adding it to traditional surveillance significantly improved real-time, state-level predictions of influenza-like illness. When a region’s wearable users start sleeping worse and their resting heart rates creep up, flu activity in that region is rising — a population-scale early-warning system assembled from data people were already generating in their sleep.

Radin, J.M., Wineinger, N.E., Topol, E.J. & Steinhubl, S.R. (2020). “Harnessing wearable device data to improve state-level real-time surveillance of influenza-like illness in the USA: a population-based study.” The Lancet Digital Health, 2(2), e85–e93.

There is an honest asymmetry here worth noticing. Aggregated across thousands of people, the noise averages out and the signal is strong. On a single wrist — yours — the same signal is real but much noisier. A public-health model can be wrong about you personally and still be right about your state; you do not have that luxury. Which is why the next two sections — how to interpret a deviation, and what to actually do about one — matter more for daily life than any of the studies above.

A Nudge, Not a Verdict

Everything that an oncoming infection does to your overnight numbers, other things do too. A couple of glasses of wine raise resting heart rate and suppress heart rate variability for a night. So do a hard training session, a hot bedroom, a late heavy meal, a stressful day, and a short night of sleep. An elevated reading on one morning is therefore not a diagnosis of anything — it is a question: what explains this? Reading your own data well is the skill of running that differential honestly, the same habit that makes recovery scores useful instead of tyrannical. The researchers behind the detection studies are explicit about this: their systems flag anomalies, and an anomaly is only a candidate for an explanation, of which infection is one among several.

A worked example: your resting heart rate is five beats above baseline. If yesterday included interval training and a birthday dinner with wine, the explanation is almost certainly not a virus — note it and move on. If instead you trained lightly, drank nothing, slept in a cool room, and the elevation arrives together with suppressed HRV and a faster respiratory rate, the picture changes. Same number, entirely different meaning — context is the difference. And in both directions the system is imperfect: false alarms happen, some infections never produce a clear signal, and an unremarkable morning is not proof of health. The research supports treating these numbers as a screening layer — a prompt to look closer — never as a test result.

How to Read Your Own Early-Warning System

Used with the right expectations, the pre-symptomatic window is genuinely useful: it buys you a day or two in which resting more, drinking more water, and cancelling the hard session can change how the next week goes — and it can spare the people around you an exposure they did not need. A practical protocol:

• Know your baseline first. Trends against your own normal are the entire method — the studies detected deviations from personal baselines, not from population averages. A few weeks of consistent overnight data gives you that reference; without it, no single reading means much.
• Look for a cluster, not a single number. One metric drifting is noise. A resting heart rate several beats above your baseline plus suppressed HRV plus elevated respiratory rate or skin temperature, all on the same morning, is a pattern worth respecting.
• Rule out the boring explanations. Alcohol, a hard workout, heat, stress, a late meal, a short night — check those first. A journal makes this check take ten seconds instead of a guess.
• When a cluster has no explanation, act early. Sleep more, hydrate, eat properly, downshift training to easy movement or rest, and spare your colleagues — work from home if you can. If confirming an infection matters for the people around you, take a test.
• Never let a good score override real symptoms. Feeling terrible is data too — arguably the most important kind. The device screens; your body reports.

Notice what this protocol does not include: panic, daily self-diagnosis, or cancelling your life over a single odd reading. The wearable’s job is to move your attention a day earlier. Your job is the same as it always was — rest when your body asks for it, and let the trend, not one anxious morning, decide how seriously to take the signal.

Training While Sick: The Neck Check

The classic heuristic for exercising with an illness is the neck check, and it has held up because it is conservative in the right places. If your symptoms are mild and entirely above the neck — a runny nose, sneezing, a slightly scratchy throat — gentle, low-intensity movement is usually fine. Keep it easy, shorten it, and stop if you feel worse. If symptoms are systemic or below the neck — fever, body aches, chest congestion, a deep cough, an upset stomach — the answer is rest. And one rule has no exceptions: a fever means rest, always.

The reason for the hard line is that training hard through a febrile infection places unusual strain on a body already working at capacity, and in rare cases infections can involve the heart muscle itself — a situation exercise can make worse. There is no need for alarm here, only for patience: the fitness you might lose in a week of rest is trivial, and the downside of forcing it is not. When you return, return gradually — easy sessions first, volume before intensity — and let your resting heart rate and HRV settle back to baseline before you go hard again. The recovery of your overnight numbers is a decent proxy for the recovery of everything else, and an elevated resting heart rate that persists after symptoms fade is your body saying it is not done yet.

A reasonable rule of thumb for the comeback: spend roughly as many easy days as you spent properly sick before reintroducing intensity, and treat any relapse of symptoms as an instruction to step back down. Nobody has ever lost a season to a week of patience; plenty of people have lost a month to a workout they forced.

Where Lamplit Fits In

The differential reading this article keeps recommending — is it a virus, or was it the wine? — requires two kinds of data side by side: what your body measured and what your life contained. Lamplit is built for exactly that combination. The journal and mood check-in, free on every plan, are where “scratchy throat” and “slept badly, felt wired” get recorded a day before any fever — often the earliest symptom log you have. Workouts, sleep, and nutrition can be logged manually for free, so the boring explanations are already written down when a strange morning arrives. Weeks later, that same log answers the question every past illness leaves behind: what did the days before it actually look like?

With Lamplit Pro, the other half arrives automatically: your wearable’s resting heart rate, HRV, and sleep sync in through Apple Health on iOS or Health Connect on Android, so the trend that drifted overnight sits directly next to the journal entry that might explain it. That is the combination the research points to — objective signals interpreted through personal context. And because this is health data, privacy is not a footnote: your data is yours, it is not sold, and sharing anything with the community is strictly opt-in.

The Honest Limits

Wearables screen; they do not diagnose. No device can tell you which infection you have, or whether you have one at all — only a test and a clinician can. Detection algorithms are tuned on populations and will fit some physiologies better than others; medications such as beta-blockers change heart-rate signals entirely; and both false alarms and silent misses are documented in every study cited above. If you feel genuinely unwell, if a fever is high or persistent, or if symptoms include chest pain, breathing difficulty, or confusion, seek medical care regardless of what any screen says. The watch is a smoke detector, not a fire brigade.

Two quieter caveats deserve a sentence each. First, most of the headline research was done during a pandemic, when base rates of infection were unusually high — the same alert fires more false positives in an ordinary winter. Second, this is intimate data: before you share readings with any platform, it is fair to ask who can see them and what they are used for. Screening for illness should never cost you your privacy.

The Bottom Line

Infections leave fingerprints in your physiology — higher resting heart rate, faster breathing, warmer skin, lower HRV — often a day or two before symptoms, and research from retrospective smartwatch analyses, population flu surveillance, and controlled viral-challenge experiments all confirms the signal is real. On your own wrist it is probabilistic: a cluster of deviations from your personal baseline, with no boring explanation, is a prompt to rest, hydrate, downshift, and pay attention — not a diagnosis, and never a substitute for a test or a doctor. Screen with the device. Confirm with medicine. And when there is a fever, rest — always.

Start free on Lamplit — journal your symptoms and mood next to your health data, and catch the pattern a day before it catches you.