AI Could Help NHS Clinicians Diagnose Childhood Sleep Apnoea, Study Finds

Key Takeaways

• AI autoscoring identified severe cases with 100% accuracy
• Processing time under five minutes versus four hours manually
• Reduces specialist workload and speeds family results
• Multi‑site UK trial approved, expanding to US
• Potential to prioritize surgeries for high‑risk children

Summary

Seluna’s AI‑driven autoscoring software accurately identified paediatric sleep apnoea in a 500‑patient NHS trial, achieving 100 % sensitivity for severe cases and processing each study in under five minutes. The platform matched human inter‑scorer variability while dramatically cutting analysis time from hours to minutes. Pediatric sleep apnoea, affecting up to 4 % of children, remains under‑diagnosed due to labor‑intensive manual scoring. Seluna now plans a UK‑wide multi‑site study and a parallel US optimisation trial to broaden deployment.

Pulse Analysis

Childhood obstructive sleep apnoea affects roughly 4 % of the global paediatric population, yet many cases go undiagnosed because traditional polysomnography requires labor‑intensive manual scoring. A single overnight study can generate dozens of physiological channels, and clinicians often spend up to four hours per patient interpreting the data. This bottleneck strains already thin specialist teams and delays treatment decisions that can influence cognitive development and behaviour. As health systems grapple with rising demand for sleep studies, the industry is actively seeking automated solutions that preserve diagnostic accuracy while accelerating turnaround.

Seluna’s cloud‑based autoscoring platform addresses that gap by applying a cascade of machine‑learning models to raw sleep signals. In a retrospective trial of 500 NHS Greater Glasgow and Clyde paediatric studies, the system flagged severe apnoea with perfect (100 %) sensitivity, while achieving 86 % and 92 % accuracy for mild and moderate cases respectively—performance that sits within the range of human inter‑scorer variability. Each report was generated in under five minutes, a stark contrast to the multi‑hour manual process. The speed and consistency promise to free physiologists for higher‑value tasks and deliver families quicker clinical feedback.

The positive results have catalysed a multi‑site UK study involving six leading children’s hospitals and an upcoming optimisation trial in the United States. If the technology scales, NHS trusts could standardise paediatric sleep diagnostics, reduce waiting lists, and better prioritise adenotonsillectomy referrals. For AI vendors, the case illustrates a viable pathway from niche research tools to reimbursable clinical software, subject to NHS procurement and FDA clearance. Ultimately, widespread adoption may reshape paediatric respiratory care, turning data overload from a liability into a diagnostic asset.