Sleep app interprets snoring patterns to provide predictive insights into treatment suitability
A pioneering study, led by PFL Healthcare in collaboration with the University of Sheffield, has demonstrated that machine-learning technology could revolutionise the management of obstructive sleep apnoea (OSA).
Researcher hope the technology will enable the NHS to cut waiting times and costs through home-based analysis conducted using only a smartphone.
The research developed a machine-learning system that analyses snoring recordings to predict whether a patient is likely to benefit from oral device (OD) therapy.
Using PFL Healthcare’s SoundSleep app, the system interprets snoring patterns collected over multiple nights and provides predictive insights into treatment suitability, without the need for invasive, hospital-based sleep studies.
An estimated 1.5 million people in the UK may be living with undiagnosed OSA, researchers found. While oral devices can help between 50 per cent and 70 per cent of patients, identifying who will respond best has historically required resource-intensive, multiple night studies, costing the NHS an average of £217 per session and adding to already long waiting lists.
In this study, the machine-learning model analysed snoring data from 934 participants. With at least seven nights of recordings, it predicted OD effectiveness with 80 per cent sensitivity and 74 per cent specificity, demonstrating how multi-night analysis offers a more representative picture of sleep health than traditional single-night testing.
Supporting research published alongside this work examined night-to-night variability in Apnoea–Hypopnoea Index (AHI) and its impact on diagnostic accuracy. The findings reinforce the value of multi-night, machine-learning-based assessment: by understanding a person’s “signature” patterns in sleep and breathing, clinicians can gain deeper insights to guide personalised treatment.
Sam Johnson, Head of Research at PFL Healthcare, said: “Accurately identifying patients who will respond to oral device therapy has always been a bottleneck in OSA care. This research shows how machine-learning analysis can make it much simpler, affordable and scalable.”
Together, these studies suggest that machine-learning-enabled, home-based screening could streamline treatment pathways – helping clinicians identify suitable OD candidates faster, reduce unnecessary referrals, and improve patient outcomes.
Sleep health technology firm Resmed announced the UK launch of its latest innovative CPAP (continuous positive airway pressure) mask to treat sleep apnoea.
