EnsoSleep

EnsoSleep is an advanced software-only medical device designed to automatically analyze physiological signals from polysomnography studies to score sleep stages, detect sleep-disordered breathing, arousals, and other sleep-related events in adult and pediatric patients aged 13 and older. Its methodology relies on sophisticated algorithms applied to diverse inputs like EEG, ECG, and airflow, with clinical validation demonstrating performance comparable or superior to manual scoring consensus and predicate devices for assessing sleep quality and aiding in sleep disorder diagnosis.

EnsoSleep is a software-only medical device intended to support physicians in evaluating sleep quality and diagnosing sleep disorders, including respiratory-related conditions, in both adult and pediatric patients aged 13 and older. The system functions by analyzing previously recorded physiological signals from polysomnography (PSG) studies, which can be conducted either in a clinical laboratory or, for adult patients, at home. Utilizing automated algorithms, EnsoSleep processes a comprehensive suite of input signals—such as electroencephalogram (EEG), electrocardiogram (ECG), airflow, and oximetry—to automatically score sleep study results. This automated scoring includes the staging of sleep (Wake, N1, N2, N3, REM), the detection of arousals, leg movements, and a variety of sleep-disordered breathing events like obstructive apneas, central apneas, hypopneas, Cheyne-Stokes respiration, and periodic breathing. Additionally, the software identifies certain cardiac events (bradycardia, tachycardia, noting these are not for primary cardiac diagnosis), derives respiratory rate, and calculates key indices such as the Apnea-Hypopnea Index (AHI) and Total Sleep Time (TST), all of which are presented for review and verification by a qualified clinician. The clinical efficacy and performance of EnsoSleep were substantiated through studies involving retrospective PSG data from an adult cohort (N=100), a pediatric cohort (N=47), and a distinct adult respiratory rate sample (N=100). The device's automated scoring capabilities were rigorously compared against a consensus reference, established by the 2/3 Majority Scoring method from multiple independent, registered sleep technologists, as well as against a predicate device. This validation comprehensively assessed four primary endpoints: the accuracy of sleep staging, the agreement in diagnosing sleep apnea, the precision in detecting specific sleep events (including sleep-disordered breathing, apneas, and arousals), and the reliability of physiological signal analysis (such as respiratory rate and TST). The findings consistently showed that EnsoSleep's performance, quantified by metrics like positive agreement (PA), negative agreement (NA), and overall agreement (OA), met or surpassed the pre-established acceptance criteria across all evaluated endpoints and types of events. Notably, the device demonstrated statistically significant enhancements in several sleep staging categories for both adult and pediatric groups, and it exhibited accuracy comparable to or better than the predicate device in identifying sleep-disordered breathing events and in diagnosing the severity of sleep apnea.

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 EnsoSleep algorithm's performance in pediatric patients (13-17 years) compare across different AASM pediatric scoring rule variations, particularly for hypopneas and respiratory arousals?; What is the specific impact of using home sleep testing (HST) data versus in-lab polysomnography (PSG) data on the accuracy of EnsoSleep's automated scoring for adult obstructive sleep apnea and related indices?; Given the expansion to include automated Cheyne-Stokes Respiration (CSR) and Periodic Breathing (PB) detection, what are the sensitivity and specificity of these specific event detections in populations with known heart failure or neurological conditions prone to these breathing patterns?; How does the performance of PPG-derived Total Sleep Time (TST) vary in comparison to EEG-derived TST across different patient demographics (e.g., varying skin pigmentation, BMI) or in the presence of comorbid conditions affecting peripheral perfusion?; For the newly added automated cardiac event detection (bradycardia, tachycardia), what were the defined thresholds, and how does their detection agreement correlate with clinically significant cardiac events observed during PSG, even if not intended for primary cardiac diagnosis?