Sleep-tracker app could give early notice of flu outbreaks
A sleep-tracking app that listens for coughing could give a week's warning of flu and Covid outbreaks, scientists have found.
Researchers said AI could effectively turn millions of smartphones into an early-warning system for respiratory infections.
The technology analyses sounds recorded while users sleep , detecting coughs without requiring them to report symptoms or seek medical help.
The Sleep Cycle app analyses sounds recorded while users sleep, detecting coughs without requiring them to report symptoms - sleepcycle
A study by the UK Health Security Agency ( UKHSA ) found increases in night-time coughing tended to precede rises in flu and Covid activity by around one week.
The UKHSA said the findings were "highly encouraging" and said the agency was now exploring how this surveillance approach could complement existing systems in the future.
Public health officials closely monitor respiratory infections each winter to assess pressures likely to be placed on hospitals as flu can spread rapidly through the population, with increases in infections followed by rising hospital admissions among elderly and vulnerable patients.
Scientists said the approach could give health services more time to prepare for surges and help officials understand whether infections are accelerating or declining.
Current disease surveillance relies heavily on information such as NHS 111 calls, laboratory tests and hospital admissions.
But such systems inevitably lag behind infections because people may wait before seeking help, while tests take time to process and report.
Researchers examined whether data already being collected by the Sleep Cycle smartphone app could provide an earlier indication of illness spreading through communities.
How it works
The app uses AI to analyse sounds during sleep and provide users with information about their sleeping patterns.
Researchers compared three years of anonymised data from England, collected between January 2023 and January 2026, with official measures of respiratory disease.
The information included the total number of coughs detected, coughs per user and coughs for each hour of sleep.
These figures were compared with NHS 111 calls about acute respiratory infections and PCR test positivity for flu and Covid.
Researchers also combined the data with hospital admission rates for flu, Covid and respiratory syncytial virus (RSV).
The app researchers analysed the data of sounds during sleep with hospital admission rates for flu, Covid and RSV - MP / Alamy Stock Photo
They found the coughing detected by smartphones closely tracked levels of respiratory illness recorded by NHS 111.
Crucially, increases in coughing also tended to emerge about a week before subsequent increases in flu and Covid activity.
Scientists said this raised the possibility of using phones as an additional early warning system alongside existing surveillance.
Prof Steven Riley, the chief data officer at UKHSA, said the technology could provide an "earlier, richer and more resilient" picture of respiratory disease.
He added that passive cough monitoring could complement existing systems without being affected by healthcare-seeking behaviour or laboratory and reporting delays.
Researchers said smartphone monitoring could also provide useful information when conventional testing is limited.
During the Covid pandemic , extensive community testing provided officials with detailed information about infections.
However, widespread testing has since been scaled back, making other surveillance systems increasingly important for tracking the virus.
'Meaningful changes in respiratory illness'
Dr Emil Carlsson, a research scientist and co-lead author, said the study showed that coughs recorded passively could capture "meaningful changes in community respiratory illness".
He said consumer-generated health information could be converted into useful epidemiological signals while maintaining privacy protections.
Sleep Cycle is available on both iPhones and Android devices, with free and paid versions.
The company does not publish a current figure for the number of UK users, although independent app data have previously put the number of active users at around 200,000.
The researchers stressed that smartphone data should be used to supplement rather than replace established disease surveillance.
The technology detects coughing rather than its cause, meaning it cannot distinguish coughs caused by flu or Covid from those caused by other infections or non-infectious conditions.
There are also potential limitations because people who use sleep-tracking apps may not be representative of the population as a whole.
However, traditional surveillance can also be distorted by differences in whether people seek healthcare and the availability of NHS services.
Scientists said combining several sources could therefore provide a more complete picture than relying on any single measure.
The research used aggregated and anonymised information rather than recordings or health information identifying individual users.
Sleep Cycle said its cough signal was generated using "privacy-preserved" data collected automatically during normal sleep.
Dr Mikael Kågebäck, the company's chief technology officer and acting chief executive, said the findings demonstrated the potential of a "completely new category of health data".
He said passively generated smartphone information could provide population-level intelligence on disease trends without requiring people to take any action.
The researchers said further prospective studies were needed to establish how reliably the approximately one-week warning could predict future outbreaks in practice.
The findings have been published on the medRxiv website as a preprint and have not yet undergone peer review.
