Your night-time cough may warn of a flu wave a week early, UKHSA study finds
What happened
On 3 September 2026, the UK Health Security Agency published the results of a study it carried out with Sleep Cycle, a Swedish company that makes a popular sleep-tracking app. The question they set out to answer was unusual. Could coughing picked up by a phone at night tell public health teams that a wave of respiratory illness was building? The answer appears to be yes, at least some of the time. The researchers compared three years of anonymised night-time cough data from the app with the UKHSA's existing surveillance systems in England, covering January 2023 to January 2026. They looked at total coughs, coughs per user, and coughs per hour of sleep. They found that rises in night-time coughing often appeared around a week before rises in flu and COVID-19 activity showed up in the usual data. The cough signal also tracked closely with the level of respiratory illness reported through NHS 111, and behaved consistently across different regions. Professor Steven Riley, UKHSA's Chief Data Officer, said that "passive nocturnal cough monitoring can complement other surveillance systems". The findings have been posted as a preprint on medRxiv and have not yet completed peer review.
Why this matters
Most of the ways we currently spot a wave of illness depend on people doing something. Ringing 111. Booking a GP appointment. Turning up at A and E. Getting a test. That creates a delay. By the time enough people have felt ill enough to seek help, and that help has been recorded and counted, the wave is already well under way. A week does not sound like much. In an outbreak it is a great deal. A week is enough time to move vaccine supplies, to warn hospitals to open extra capacity, to bring in extra staff, or to tell the public that now is the moment to get a jab. There is a second reason this is interesting. It costs almost nothing. The data already exists, because people were using the app to track their sleep anyway. No one had to be recruited, tested, or asked to fill in a survey. Dr Emil Carlsson, a research scientist on the study, described the idea as turning consumer-generated health data into something "epidemiologically meaningful". That is the real claim here: everyday gadgets producing genuinely useful public health information.
What the evidence actually says
This is genuinely promising work. It is also early, and it is worth being clear about the limits. First, the paper is a preprint. That means it has been made public before other scientists have formally checked it. Preprints are normal and useful in fast-moving public health work, but conclusions can change during peer review. Second, this is not a diagnostic tool. The study is about patterns across whole populations, not about individuals. Your app noticing you coughed more last night does not mean you have flu. It might mean the window was open, or you have hay fever, or you had a glass of wine. Third, the people who use a sleep-tracking app are not a cross-section of the country. They are more likely to be younger, better off, and comfortable with technology. Older people, who are most at risk from flu, may be under-represented. Researchers can adjust for some of this, but not all of it. Fourth, the phrase used in the findings is that cough rises were "often" seen about a week ahead. Not always. A signal that works most of the time is useful for planning, but it is not a guarantee. On privacy, the study describes a sensible design. Cough events are identified by a machine learning model that runs on the phone itself, analysing short overlapping audio clips during sleep sessions the user has started. According to the study, no raw audio is sent to the company's servers. Only counts leave the device, and the data used by UKHSA was aggregated and anonymised. That design matters. It is the difference between a phone counting coughs and a phone recording your bedroom.
Practical advice
There is nothing here that you need to do. But there are a few sensible things to keep in mind. If you use a sleep app or a smartwatch, treat its health readings as interesting rather than authoritative. They can spot patterns. They cannot diagnose you. If you are worried about a cough, judge it the ordinary way. See a pharmacist or GP if a cough lasts more than three weeks, if you are coughing up blood, if you are short of breath, if you have chest pain, or if you have unexplained weight loss. Do not wait for an app to tell you something is wrong, and do not be reassured by an app that says nothing is. If you care about your data, take two minutes to look at the privacy settings in any health app you use. Look for whether processing happens on your device or in the cloud, and whether data is shared in an identifiable form. Good apps make this easy to find. And if a flu wave is forecast, the useful response is the boring one. Get vaccinated if you are eligible, wash your hands, and stay home when you are ill so you do not pass it on.
What to know
UKHSA and Sleep Cycle published a study on 3 September 2026 suggesting that night-time cough data from a sleep app can signal rising flu and COVID-19 activity roughly a week before existing surveillance picks it up. The study compared three years of anonymised app data with UKHSA surveillance in England between January 2023 and January 2026, and found the cough signal tracked closely with NHS 111 reports. UKHSA's Chief Data Officer describes it as something that could complement existing systems, not replace them. The work is a preprint and has not yet been peer reviewed. App users are not a representative sample of the population, and the early-warning effect was described as often, not always. Cough detection ran on users' own devices, with no raw audio sent to servers, which is a meaningful privacy protection. For individuals, none of this changes what to do about your own cough. That still comes down to how long it lasts and what else you are feeling. Sources: GOV.UK / UK Health Security Agency, "AI sleep data could provide one-week early warning of flu and COVID-19", 3 September 2026, https://www.gov.uk/government/news/ai-sleep-data-could-provide-one-week-early-warning-of-flu-and-covid-19 | Gavi VaccinesWork, "Could your night-time cough help predict the next wave of flu or COVID-19?", 2026, https://www.gavi.org/vaccineswork/could-your-night-time-cough-help-predict-next-wave-flu-or-covid-19 | UKAuthority, "UKHSA explores can sleep data spot respiratory disease trends", 2026, https://www.ukauthority.com/articles/ukhsa-explores-can-sleep-data-spot-respiratory-disease-trends This article is for general information and does not replace advice from a doctor, pharmacist or other qualified healthcare professional. Health and sleep apps are not medical devices for diagnosis. If you are worried about a cough or any other symptom, speak to a healthcare professional.
Related articles
Three quiet changes to how the UK regulates medicines and devices
On 1 September 2026, the Medicines and Healthcare products Regulatory Agency announced that the government has tabled three amendments to the Health Bill curren…
Four hours of training and an AI: heart scans by non-experts show promise
Researchers at the Mayo Clinic have tested a way of using artificial intelligence to help people with no ultrasound experience take useful pictures of the heart…