FDA Clears HoneyNaps' SOMNUM V3.0 Sleep Study Software With Apnea Subtype Scoring
The 510(k) clearance updates the Korean company's polysomnography software, which now labels apnea events as obstructive, central or mixed for physician review.
The FDA cleared this device through the 510(k) pathway, which finds it substantially equivalent to a device already on the market. Clearance is not the same as approval.
Key points
- The FDA cleared SOMNUM under 510(k) K253390, with the company's earlier SOMNUM V1.1.2 as the predicate device.
- The software scores sleep stages, arousals, breathing events and leg movements in recorded adult sleep studies, and physicians can review and edit the results.
- Validation used 100 full-night recordings from four U.S. sleep centers, according to the FDA summary.
HoneyNaps, a Seoul-based developer of sleep medicine software, said on July 16 that the U.S. Food and Drug Administration has cleared SOMNUM V3.0, an updated version of its AI software for analyzing overnight sleep studies, under 510(k) number K253390, according to the company's release on PR Newswire.
The FDA's decision letter and 510(k) summary, dated June 27, 2026, found the software substantially equivalent to the company's earlier SOMNUM V1.1.2 (K223922). It is regulated as a Class II device under the electroencephalograph regulation.
What the update adds
SOMNUM analyzes previously recorded polysomnography data, the multi-channel recordings collected during in-lab sleep studies. According to the FDA summary, it automatically scores sleep stages, arousals, apneas and hypopneas, and periodic leg movements, and physicians can review, edit or delete the scored events before generating a report. Use is restricted to data from adult patients, under a physician's supervision.
The headline change in version 3.0, according to HoneyNaps, is that the software classifies apnea events as obstructive, central or mixed sleep apnea rather than assessing sleep-disordered breathing mainly through composite indices. The company said the event-level classification is "intended to support more detailed clinical scoring."
The FDA summary describes the respiratory event detection as AI/ML-based and the subclassification of apneas into obstructive, central and mixed as rule-based.
The validation study
According to the summary, HoneyNaps evaluated version 3.0 in a retrospective study of 100 full-night recordings from the National Sleep Research Resource STAGES dataset, collected at four U.S. sleep centers. The cohort had 50 men and 50 women. Three certified sleep technologists scored the recordings independently to set the reference standard, with a board-certified sleep physician adjudicating disagreements.
Against that reference, the summary reports overall percent agreement of 97.42% for obstructive apnea, 99.83% for central apnea, 99.94% for mixed apnea and 98.58% for hypopnea. Positive percent agreement, the share of reference-scored events the software also found, was lower: 86.26% for obstructive apnea, 82.19% for central apnea, 75.67% for mixed apnea and 81.71% for hypopnea.
The summary notes that the reported metrics are based on epochs where the technologists agreed and "may not fully reflect device performance in the most clinically ambiguous cases."
Next versions
HoneyNaps said it is developing additional measures of sleep apnea severity, including hypoxic burden, arousal burden and ventilatory burden, and plans to seek further 510(k) clearances for future SOMNUM versions that include them. "We remain focused on advancing AI-driven sleep diagnostics through continued innovation in automated analysis and next-generation digital biomarkers," Sean Ha, president of HoneyNaps USA, said in the release.
Sources
- HoneyNaps Receives U.S. FDA 510(k) Clearance for SOMNUM V3.0 for AI-Based Sleep-Disordered Breathing Analysis, HoneyNaps via PR Newswire, July 16, 2026
- 510(k) decision letter and summary, K253390 (SOMNUM), U.S. Food and Drug Administration, June 27, 2026