Real-World Performance and Safety of Apneal Dx for Sleep Apnea Screening

Overview

Sleep apnea-hypopnea syndrome (SAHS) is a frequent and largely under-diagnosed sleep-related breathing disorder. Adult prevalence is estimated between 9% and 38%. Untreated SAHS increases cardiometabolic and neurocognitive risk. Reference diagnostic methods (in-lab polysomnography and ambulatory ventilatory polygraphy) have limited capacity, which delays care. Apneal Dx is a Class IIa software as a medical device (CE-marked 10 June 2026). It measures cardiorespiratory sleep activity using the sensors of a standard smartphone worn on the chest, with no additional sensor, to screen for and aid in the diagnosis of adult obstructive sleep apnea-hypopnea syndrome at home. This study is a non-interventional, prospective, real-world post-market clinical follow-up (PMCF) of Apneal Dx used within its intended purpose in routine care. It does not add any examination, visit, randomisation or change to patient management. Data are collected from automatic device logs and optional in-app questionnaires completed by patients and healthcare professionals. The study covers France, Spain, Italy, the United Kingdom and Germany. There is no investigational site and no principal investigator.

OBJECTIVES This PMCF study addresses residual evidence gaps after CE marking of Apneal Dx, under MDR Articles 61(11), 83, 84 and 86 and Annex XIV Part B, in support of a strategy based on Article 61(10). Primary objectives: (A) Demonstrate end-to-end technical performance in the unsupervised home environment and detect safety-relevant failure modes. (B) Confirm concordance between the Apneal Dx AHI severity category and the healthcare professional's final clinical diagnosis obtained through the routine care pathway. (C) Quantify real-world false positive and false negative rates at AHI thresholds of 5, 15 and 30 events per hour. Secondary objectives: (D) Care-pathway time indicators (time to consultation, diagnosis and treatment). (E) How device outputs are used in clinical decisions. (F) Diagnostic performance against polysomnography (PSG) and ventilatory polygraphy (PG), when those tests have already been performed in routine care. (G) Real-world screening utility, conditions of use and usability. METHODS Design: observational, prospective, single-cohort, longitudinal real-world data collection. Non-comparative; no randomisation; no blinding; no imposed follow-up visit. Target enrolment: 400 evaluable patients (estimated). Individual participation lasts up to 24 months (questionnaires at 6, 12 and 24 months after the first recording). Study period: 04 September 2026 to 10 June 2031 (57 months, aligned with the CE certificate cycle). No intervention is assigned by the protocol. Apneal Dx is used under normal conditions of use and within its intended purpose. A PSG or PG result is collected only if the test has already been performed in the usual care pathway; the study never prescribes, prompts, schedules or funds a confirmatory test. Data sources: (1) automatic technical metadata from the device (session duration, signal quality, processing events, error codes, software version); (2) healthcare professional questionnaire (H01-H14) on the HCP web platform; (3) patient questionnaire (P01-P12) in the mobile application. No medical records, no SNDS, no biospecimens, no linkage with an external database. Setting: home and routine clinical practice. Decentralised collection with no investigational site, no participating institution and no principal investigator. The sponsor headquarters (Mitral SAS, Charenton-le-Pont, France) is declared as the single administrative location. Participants may reside in France, Spain, Italy, the United Kingdom or Germany. Ethics: no ethics committee is territorially or institutionally competent in the absence of a site. Human subjects protection review is therefore recorded as not required. Participation follows a dedicated in-app information notice, distinct from the terms of use, and a voluntary in-app agreement. A permanent right to object applies. Analysis: descriptive (point estimates and 95% Wilson confidence intervals, distributions, weighted Cohen kappa for concordance, sensitivity and specificity by AHI threshold, median delays). No superiority or non-inferiority hypothesis test. No multiplicity adjustment. Missing data are not imputed.