The Sound of Sleep: Machine-learning snore analysis could transform sleep apnoea treatment

A pioneering analysis has shown that machine-learning technology could transform how obstructive sleep apnoea (OSA) is identified and managed – helping the NHS reduce waiting times and costs through home-based analysis using only a smartphone. The research, led by PFL Healthcare in collaboration with the University of Sheffield, 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. While oral devices can help between 50% and 70% 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% sensitivity and 74% 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 indicates 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. The post The Sound of Sleep: Machine-learning snore analysis could transform sleep apnoea treatment appeared first on Digital Health Technology News.

Source: https://www.healthtechdigital.com/the-sound-of-sleep-machine-learning-snore-analysis-could-transform-sleep-apnoea-treatment/