Huxley SANSA Home Sleep Apnea Test (1000-00)

The Huxley SANSA Home Sleep Apnea Test is a chest-worn wearable device employing PPG, ECG, and accelerometer sensors to record various biophysical parameters for adults suspected of sleep apnea. It utilizes a cloud-based AI/ML algorithm to analyze this data, providing outputs like the Apnea-Hypopnea Index and Total Sleep Time to aid healthcare professionals in evaluating sleep-related breathing disorders.

The Huxley SANSA Home Sleep Apnea Test (HSAT) is a prescription-only wearable device designed for adults suspected of having sleep apnea. It aids in evaluating sleep-related breathing disorders by recording, analyzing, and storing biophysical parameters. The system utilizes a single patch worn on the chest, incorporating a reflective photoplethysmography (PPG) sensor, a single-lead electrocardiogram (ECG) sensor, and a 3-axis accelerometer. It collects data on oximetry, heart rate, chest movement, snoring, body position, respiratory effort, actigraphy, and sleep staging (sleep/wake), with the ECG channel serving as a reference. The device is intended for use in both clinical and home settings under the direction of a healthcare professional, with each study lasting approximately 10 hours. Data gathered by the SANSA HSAT is transferred via USB to a cloud-based software platform where an algorithm, combining signal processing and AI/Machine Learning components, computes time-series data and summary metrics, including the sansa-Apnea Hypopnea Index (sAHI) and Total Sleep Time (TST). Clinical performance data indicated SpO2 measurement accuracy with an overall average root mean square error (ARMS) of 2.4% (ranging from 1.7% to 3.7% across different Fitzpatrick skin tone subgroups) against arterial blood gas samples. In a comparison study with polysomnography (PSG) involving 533 participants, the device demonstrated a sensitivity of 88.2% and a specificity of 87.3% for diagnosing moderate to severe sleep-disordered breathing. Subgroup analyses showed varied performance across race, Fitzpatrick skin tone, and BMI. The AI-based sleep/wake classification algorithm showed a sensitivity (sleep) of 95% and specificity (sleep) of 63%.

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/ML algorithm differentiate between various types of respiratory events (e.g., obstructive versus central apneas/hypopneas) using the collected sensor data, particularly without direct airflow measurement?; What is the impact of the reported 63% specificity for sleep staging on the accuracy of the total sleep time (TST) and subsequently the sansa-Apnea Hypopnea Index (sAHI) calculation, especially in patients with highly fragmented sleep?; Given the variability in SpO2 accuracy across Fitzpatrick skin tones (ARMS 3.7 for types 3-4 vs. 1.7 for types 5-6), what specific algorithm adjustments or sensor characteristics contribute to this performance, and are there further plans to optimize accuracy consistently across all skin tones?; The ECG channel is designated for reference only; what specific data or patterns from this reference ECG contribute to the AI/ML algorithm's assessment of sleep-related breathing disorders or other reported parameters, if it's not used for cardiac diagnostic purposes?; Considering the diagnostic sensitivity for moderate to severe sleep-disordered breathing was 66.7% in individuals with Fitzpatrick skin tones 5-6, what are the potential clinical implications for this subgroup, and how does this compare to other home sleep apnea tests using similar sensor technologies?