What question did this study set out to answer?

To explore the efficacy of multi-night photoplethysmography (PPG) for screening narcolepsy in patients.

May 10, 2026

0697 Narcolepsy Risk Screening at Scale with Multi-Night PPG

Key Points

To explore the efficacy of multi-night photoplethysmography (PPG) for screening narcolepsy in patients.
Analyzed data from 19,074 patients monitored over 3 to 30 nights using EnsoSleep PPG.
Defined narcolepsy presence through mean sleep latency (MSL) and sleep-onset REM periods (SOREMP).
Utilized machine learning algorithms to interpret over 1,000,000 diagnostic polysomnograms.
0.6% of patients met the ePPG diagnostic criteria for narcolepsy (N=105).
93.4% of patients exhibited no SOREMPs; 21.2% had MSL ≤ 8 minutes with < 2 SOREMPs.
Prevalence aligns with expectations for a clinic population, underlining potential for large-scale screening.

Abstract

Abstract Introduction Narcolepsy is a devastating, often unrecognized CNS hypersomnia involving excessive daytime sleepiness, sleep disturbance, sleep paralysis, and hypnagogic and hypnopompic hallucinations. Cataplexy may or may not be present. Clinical hyperfocus on sleep-disordered breathing as the cause of excessive daytime sleepiness risks missing narcolepsy diagnoses. We aimed to address this problem by using multi-night data from photoplethysmography (PPG) based home sleep apnea tests (HSAT) to screen for narcolepsy. Methods EnsoSleep PPG (ePPG) is an FDA-cleared, interoperable, clinically validated software as a medical device (SaMD) HSAT. ePPG studies were scored via a machine learning (ML) based module trained by applying ML and statistical signal processing methods, including multiple deep neural networks, to a database of over 1,000,000 diagnostic polysomnograms with concurrently recorded PPG. Included studies had: ≥ 4 hours of total sleep time, test duration ≤ 10 hours, and ≥ 3 nights total monitoring per patient (Min 3, Max 30). Narcolepsy presence was defined as a mean sleep latency (MSL) ≤ 8 minutes averaged over the total number of HSAT nights and ≥ 2 sleep-onset REM periods (SOREMPS) over the total nights monitored. Results The study included N=19,074 patients with mean age (49.3 years, SD=13.6), mean BMI (31.3, SD=6.6) and sex 65% male and 35% female. Data came from 90 individual sleep centers. Means were: AHI 15.9 events/hour (SD=15.3), sleep duration 6.5 hours (SD=0.8), sleep latency 17.7 minutes (SD=14.1), and REM latency 99.2 minutes (SD=51.2). SOREMP frequency was: none (93.4%), one (5.4%), two (1.0%), three (0.1%) and four (0.1%). 4,050 (21.2%) patients had a MSL ≤ 8 minutes and 2 SOREMPs, and 121 (0.6%) patients had ≥ 2 SOREMPs and MSL 8 minutes. ePPG narcolepsy diagnostic criteria was met for 0.6% (N=105) of patients and not met for 99.4% (N=18,969). Conclusion Our multi-night based narcolepsy prevalence was similar to that expected for a sleep clinic population and indicates an opportunity to screen for narcolepsy at scale from HSATs. Multi-night ePPG testing could offer similar insights to MSLTs without the corresponding operational overhead or burden. Further SaMD development of multi-night HSAT in narcolepsy patients offers the possibility of an at home narcolepsy test (HNT). Support (if any)

Mark Helpful

Bookmark

Relay