What changes in the algorithm
Samsung has announced that the new version of the algorithm for detecting nighttime apnea on Galaxy devices has received approval from South Korea’s Ministry of Food and Drug Safety (MFDS), as reported by IT之家. Compared with the previous version, the system does not merely identify signals compatible with moderate or severe forms of apnea: according to the source’s account, it uses artificial intelligence components to count episodes of breathing interruption for each hour of sleep and divides them into three categories, namely mild, occasional and frequent.
The change therefore concerns the level of detail in the report provided to the user, which moves from a binary indication of the presence of suspicious signals to a quantitative estimate of event frequency. The three-band classification is presented as a way to frame the recurrence of breathing interruptions during the night, but the material does not specify the numerical thresholds that separate one band from another.
Devices and rollout timeline
The rollout will begin in November but will initially remain limited to the South Korean market: according to IT之家, the first devices to receive the update will be the Galaxy Watch Ultra2, Galaxy Watch9 and the Galaxy Ring smart ring in the Korean version. At a later stage, the update will also reach Galaxy Watch7 and Galaxy Watch8, provided they run Wear OS 7.0 or later versions of the system.
The sequence indicates a multi-stage release strategy that prioritizes the most recent products and the market in which regulatory authorization was obtained. The text does not clarify whether and when approval will be extended to other countries, nor whether the operating system requirement entirely precludes access for earlier models or only affects their timing. These aspects remain to be verified.
Declared limitations and conditions of use
Samsung states, according to the source, that the function’s results should be understood as a support element in consultation with a healthcare professional and cannot replace a medical diagnosis. The function is therefore not presented as a standalone clinical tool, but as a system for collecting information useful for a discussion with a doctor.
To obtain the result, an operational condition must also be met: recording two nights of sleep within a ten-day period, with sessions of at least five hours each. The detail is relevant because it suggests that the system requires multiple observations to generate an estimate. The material does not indicate how the data collected are handled, nor does it specify the algorithm’s accuracy or the possible presence of false positives and false negatives—elements that a complete assessment of the function should consider.
