Sleep Apnea Feature

Samsung's Sleep Apnea Feature for Galaxy smartwatches employs PPG-based SpO₂ and pulse monitoring over two nights, utilizing an embedded AI pipeline to estimate an apnea-hypopnea index (eAHI) for screening adults for moderate-to-severe obstructive sleep apnea. This over-the-counter technology aims to broaden accessibility to initial sleep apnea risk detection, alerting users with significantly elevated eAHI readings to consult healthcare professionals.

Samsung's Sleep Apnea Feature, integrated into Galaxy smartwatches, is designed as an over-the-counter screening tool for adults aged 22 and older to detect signs of moderate-to-severe obstructive sleep apnea. The system leverages photoplethysmography (PPG) sensors to continuously monitor the user's pulse and blood oxygen saturation (SpO₂) levels over two separate nights of sleep. This collected physiological data is then processed by an embedded artificial intelligence (AI) pipeline specifically developed to analyze these signals. The AI algorithm preprocesses the PPG and SpO₂ data to estimate an apnea-hypopnea index (eAHI), which quantifies the number of apnea and hypopnea events per hour of sleep. If this estimated AHI is consistently 15 events per hour or higher across both monitored nights, the feature alerts the user, recommending they seek a formal medical evaluation for sleep apnea. The development of this algorithm was informed by data from over 1,000 subjects, with its performance benchmarked against polysomnography results, demonstrating high accuracy with up to 100% sensitivity and 94.1% specificity, even in conditions of low peripheral perfusion, meeting established accuracy standards for such screening tools.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the AI pipeline specifically differentiate between obstructive sleep apnea events and other respiratory disturbances or signal artifacts using only PPG-based SpO₂ and pulse data?; What is the long-term reliability and performance of the PPG sensor and AI algorithm across diverse populations with varying skin tones, wrist sizes, and comorbid conditions not extensively covered in the initial 1,000-subject study?; What is the positive predictive value and negative predictive value of an eAHI above 15 events/hour in a general population, considering the prevalence of moderate-to-severe OSA?; How does the device's performance in detecting sleep apnea compare when users engage in activities or sleep in positions that might compromise PPG signal quality (e.g., wrist flexion, pressure on the watch)?; What are the defined thresholds and mechanisms within the AI for determining 'low perfusion conditions,' and how does the algorithm adjust or maintain accuracy under such circumstances?