DeepRESP

DeepRESP is a cloud-based software application that analyzes sleep study recordings to aid in the diagnosis of sleep disorders in adults 22 years and older. It automatically scores sleep stages, respiratory events, and arousals from EEG and non-EEG signals, with all output subject to medical professional verification. The device can be used in hospitals, patient homes, or ambulatory settings.

DeepRESP is an API-based cloud platform using artificial intelligence and rule-based algorithms for automated sleep study analysis, whereas the predicate Nox Sleep System is a PC-based application with included hardware components. Both devices perform automatic scoring of sleep stages and respiratory events ready for physician review, analyze similar physiological signals (EEG, EMG, EOG, respiratory effort, SpO2), and require manual verification of results. DeepRESP additionally provides automated sleep staging without EEG, a capability the primary predicate lacks.

IEC 62304:2006/A1:2015 (Medical device software lifecycle), ISO 14971:2019 (Risk management), ANSI/AAMI SW96:2023 (Software validation), and FDA Guidance for Software Validation and Content of Premarket Submissions for Software in Medical Devices.

DeepRESP demonstrates non-inferiority or superiority to the Nox Sleep System predicate across severity classification, respiratory event scoring, and sleep state estimation in both Type I (with EEG) and Type III (without EEG) recordings. Clinical validation studies using 2,224 Type I and 3,488 Type III sleep recordings showed DeepRESP achieved comparable or better performance metrics (PPA, NPA, OPA) against manual scoring compared to the predicate device. Both devices serve the same intended purpose of categorizing sleep-related events to aid diagnosis, use similar signal inputs, and require physician verification of automated results, establishing substantial equivalence despite DeepRESP's cloud-based architecture and enhanced EEG-absent scoring capability.