What happened
Samsung Electronics makes the Sleep Apnea Feature, software for a Galaxy Watch and phone that reads blood oxygen during sleep. It looks for repeated drops in oxygen over two nights and includes machine-learned algorithms. An accredited sleep lab ran a multi-center study in an enriched adult population, with polysomnography as the reference. FDA reviewers report that sensitivity was 82.7%, or 167 of 202 people with moderate to severe obstructive sleep apnea. Specificity was 87.7%, or 235 of 268, which missed the acceptance criterion. After review of the false positives, FDA accepted a modified specificity of 91.1%. Of 33 false positives, 10 previously undiagnosed people had mild apnea, and FDA wrote that the result benefited them. FDA granted the De Novo request on 6 February 2024. The feature assesses risk and makes no diagnosis. FDA reports no follow-up of whether participants sought care.
Sources
-
Regulatory document, by authors with no tie to the system.
U.S. Food and Drug Administration, Center for Devices and Radiological Health. De Novo classification request for Sleep Apnea Feature (Samsung) (DEN230041): decision summary. Granted 2024-02-06.
The Samsung Sleep Apnea Features' sensitivity was 82.7% (167 out of 202 subjects) with a 95% confidence interval of [76.7%, 87.6%], which passes the predetermined acceptance criteria.
Details
AI system
- Workflow
-
The watch derived blood oxygen from its PPG sensors and scored each of two nights within 10 days. Nights with under three hours of sleep, or with oxygen readings for under 70% of the time, gave an insufficient-data notice.
Follow-up
- Conditions for replication
-
The user is 22 or older and has no diagnosis of sleep apnea. Users with symptoms see a clinician whatever the result.
Entry details
- Sector
-
Healthcare.
- Purpose
-
Samsung built the Sleep Apnea Feature to alert adults without a diagnosis to signs of moderate to severe obstructive sleep apnea from their watch's oxygen readings.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Samsung's Sleep Apnea Feature caught most moderate to severe cases (AIBD-0770). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0770.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)