CPAP Data Statistics: What Users Import

Original Data, Counted the Privacy-First Way

This page publishes statistics from CPAP Clarity's own anonymous usage aggregates: which machines people import data from, which oximeters and wearables they pair with them, and what the combined picture looks like across thousands of nights.

One thing makes these numbers unusual, and it is worth stating up front. CPAP Clarity processes all health data in your browser, so our servers never see a single AHI, leak rate, or oxygen reading. What we can count is categorical and behavioral: which device model an import came from, whether it succeeded, which plain-language finding a night surfaced. That constraint is the privacy promise working as designed, and everything below lives inside it. Counts are as of July 11, 2026, from the site's launch on April 1, 2026.

The Machines People Actually Import

Across 6,467 import attempts, 4,431 succeeded (a 69% success rate; most of the rest are folders that turn out not to be CPAP data at all). Among imports where the machine model could be identified, the leaders:

Machine	Identified imports
ResMed AirSense 11 AutoSet	1,706
ResMed AirSense 10 AutoSet	539
ResMed AirCurve 11 VAuto	111
Löwenstein Prisma Smart	106
ResMed AirCurve 10 VAuto	61
BMC G3 A20	59
React Health Luna G3	52
ResMed AirCurve 11 ASV	37
ResMed AirSense 11 Elite	36

The headline: the AirSense 11 now outnumbers the AirSense 10 by more than 3 to 1 among data-curious users (deciding between them? See our AirSense 11 vs 10 comparison). The AirSense 10 was the most common machine in the world for most of a decade, but among people importing data in 2026, the transition to the 11 is decisively underway. A further 1,341 ResMed imports arrived without the settings folder that identifies the exact model, so the true ResMed totals run higher than the table shows.

Bilevel machines (the AirCurve line) account for a meaningful minority, and the long tail includes machines few tools can read at all: Löwenstein's Prisma Smart, BMC's G3 family, and React Health's Luna G3 together outnumber Philips DreamStation imports several times over in our data.

The Oximeters Riding Alongside

About 570 overnight oximeter imports identified their device. The Wellue/Viatom family dominates, and inside it the race is a dead heat: the original O2Ring (190 imports) and the O2Ring S (188) are statistically tied, with the Checkme O2 Max (128 including combined imports) third. Fingertip spot-check devices barely register, which makes sense: an overnight recording ring produces a file worth importing, and a spot-checker does not. If you are considering adding one, our pulse oximeter guide covers what the overnight numbers mean.

395,000 Wearable Nights Next to CPAP Data

Users have imported roughly 395,000 nights of wearable sleep data to sit alongside their CPAP nights:

Wearable	Nights imported	Share
Apple Watch	338,472	86%
Oura Ring	18,331	5%
Fitbit	13,750	3%
Samsung Galaxy Watch	12,272	3%
Garmin	9,515	2%
RingConn	2,860	1%

A caveat that matters: these are nights, not people. An Apple Health export contains every night the watch ever recorded, often years per person, so Apple Watch's share reflects both its popularity and the depth of its export. By import events, Apple Watch still leads, with Oura second and Fitbit third.

What Joined Nights Actually Show

This is the statistic no one else has. When a user has two or more sources imported for the same night (CPAP plus an oximeter or a wearable), CPAP Clarity composes one plain-language read of that night. Users have rendered 1,923 of these joined-night reads. How they broke down:

• 31% read clean across every joined source. The machine, the oxygen recording, and the wearable all had an unremarkable night.
• 20% were corroborations: the sources told one consistent story about the night, whatever that story was.
• 14% were coincidence checks that came back clear: no scored breathing event lined up in time with an oxygen dip.
• 18% had nothing specific to say beyond the per-source numbers.
• The rest, about 16% of reads, surfaced a specific finding, and not all findings are concerns. The most common was disrupted REM sleep (about 5% of all reads). Next came the single most interesting category: nights where the CPAP's event counts looked controlled while the oximeter still recorded oxygen dips (about 3%). The remaining 9% spread across a longer tail, some of it reassuring (low event counts paired with strong recovery signals) and some of it worth watching (persistent mask leak across sources, fragmented sleep despite controlled events, and correlations between breathing events and heart-rate or oxygen patterns).

Two honest observations. First, most joined nights are reassuring, and that is the expected result for an engaged group already established on therapy. Second, the minority that did surface something is precisely the argument for looking at more than one signal: about 1 night in 35 in our data (roughly 3%) showed oxygen behavior the CPAP's own counts would not have flagged on their own. What any of that means for an individual night is a conversation for that person and their clinician, not a statistic.

Over 30-night windows, the trend engine's most frequent findings were persistent leak patterns and persistent event-rate patterns, in that order.

What Mask-Finder Answers Suggest

Our Mask Finder quiz has 104 completed runs, a small sample worth reading loosely. Among completions, 55% landed on a nasal pillow recommendation, 23% hybrid, 20% nasal cradle, and 2% full face. That inverts the conventional wisdom that full-face masks dominate, and the likely explanation is selection: people who seek out a mask quiz on a data-analysis site skew toward side sleepers and minimal-contact preferences, not the population of all CPAP users.

Engagement Worth Noting

Two behaviors stood out enough to count. Users have generated about 1,960 PDF therapy reports to bring to appointments, and about 960 AI exports (the copy-paste summary designed for ChatGPT or Claude). The report-for-your-doctor path and the ask-an-AI path are now the site's two biggest post-import actions, which says something about where people take their data next.

Methodology

• Source: CPAP Clarity's first-party anonymous telemetry aggregates (daily rollup tables), counted from public launch on April 1, 2026 through July 11, 2026.
• What the telemetry is: anonymous, categorical events (device model detected, import succeeded or failed, which finding a night surfaced). No account exists, no cookies are set, and consent gates collection.
• What it is not: health values. AHI, leak, oxygen, and every other therapy number are processed in the browser and never transmitted, so no statistic on this page is computed from anyone's health data. That is why these are counts and shares rather than medians of clinical numbers.
• Counting caveats: import counts are events, not deduplicated people (someone importing twice counts twice). Wearable nights reflect export depth as well as popularity. All cohorts here are aggregate; we do not publish any slice small enough to describe an individual.
• Selection caveat: everyone in this data chose to analyze their CPAP data with a browser tool. That skews engaged, technical, and ResMed-heavy relative to all CPAP users. Treat these as statistics about data-curious CPAP users, not the therapy population.

We plan to refresh these numbers periodically as the dataset grows. If you want your nights in the next edition, import your data free, and it never leaves your browser.

Frequently Asked Questions

What is the most popular CPAP machine in 2026?

Among CPAP Clarity users importing data, the ResMed AirSense 11 AutoSet leads by a wide margin, outnumbering the AirSense 10 AutoSet by more than 3 to 1. That reflects data-curious users rather than all CPAP owners, but the direction is clear: the fleet is turning over to the AirSense 11.

How were these statistics collected?

From anonymous, categorical usage telemetry: which device model an import came from, whether it succeeded, which plain-language finding a joined night surfaced. Health values like AHI and oxygen never leave the user's browser, so none of these numbers are computed from anyone's therapy data.

How often do wearables and CPAPs disagree?

In 1,923 joined-night reads, most nights were reassuring: about a third read clean across every source and another fifth showed the sources telling one consistent story. About 1 night in 35 surfaced oxygen behavior that the CPAP's own event counts would not have flagged alone, which is the strongest case in our data for pairing an oximeter with a CPAP.

Which wearable do CPAP users pair most?

Apple Watch, by both nights imported (86% of roughly 395,000 wearable nights) and import events, with Oura Ring second and Fitbit third. Apple Watch's share is inflated by export depth, since one Apple Health export can contain years of nights.

Are these numbers a clinical study?

No. They are behavioral statistics from a self-selected group of people who chose to analyze their own CPAP data. They say nothing about therapy effectiveness for any individual, and nothing here replaces a conversation with a sleep clinician.

Will these statistics be updated?

Yes, periodically as the dataset grows. Each edition states its count date; this one covers April 1 through July 11, 2026.