Narcolepsy Revolution – Protocol and Methodology A diagnostic accuracy study protocol using the Dreem 3 headband for ambulatory diagnosis of narcolepsy in children and young adults

Abstract Background Narcolepsy is a rare, lifelong neurological disorder that often begins in childhood or adolescence. Diagnosis is frequently delayed because current diagnostic testing relies on specialist in-patient sleep investigations: overnight polysomnography (PSG) followed by a multiple sleep latency test (MSLT), interpreted according to International Classification of Sleep Disorders criteria (ICSD-3-TR). These investigations are expensive, labour intensive, and available in a limited number of centres, contributing to delays and inequity of access. Automated analysis of sleep-stage probabilities (hypnodensity) using neural networks has shown promising diagnostic performance in research cohorts but still requires hospital-based PSG acquisition. The Dreem 3 headband (DH) is a comfortable, dry-montage EEG device designed for home use. Combined with its proprietary machine learning classification of sleep stages, it may offer accurate ambulatory sleep physiology assessments and support clinical decision making. Methods This was a single-centre, prospective, observational study recruiting 60 participants aged 10 to 35 years undergoing investigation for hypersomnolence within GSTT sleep services and scheduled for PSG and MSLT as part of routine care. Exclusion criteria included physician-diagnosed medical or psychiatric disorder that could independently account for excessive daytime sleepiness; and/ or regular use of prescribed or recreational medication known to affect sleep architecture. Participants first wore the DH at home for five weeknights, followed by a continuous 48-hour weekend recording using two devices in rotation. They then underwent routine in-patient PSG and MSLT. PSG and MSLT were interpreted according to ICSD-3 by an experienced sleep physician and a final diagnosis determined by a sleep physiology consultant. The primary outcome is accuracy of ambulatory DH-based assessment of sleep physiology and subsequent diagnosis of sleep disorders. We evaluate proprietary and in-house developed machine learning methods to detect SOREM epochs and predict narcolepsy diagnosis from PSG, PSG+MSLT and DH data. All algorithmic outcomes will be compared to clinical outcomes derived from current clinical standard of care. Discussion This study will provide proof-of-concept evidence for a home-based wearable EEG approach to narcolepsy diagnosis. Patient and public involvement work with young people with confirmed narcolepsy indicates high acceptability of the DH protocol: in a survey of ten young people, eight reported they would be willing to wear a sleep headband nightly at home for five days (two were unsure), and seven reported they would be willing to wear it continuously for 48 hours over a weekend (two were unsure; one said no). These findings informed the decision to restrict continuous wear to the weekend, reflecting feedback that daytime wear during school or work hours would be unacceptable. If validated, this approach could reduce delays to diagnosis, improve equity of access, and support development of a subsequent multicentre study. Trial registration IRAS Project ID: 321547. Registered October 2022. Recruitment was completed on 30 January 2026.