Oura, WHOOP & Garmin Apple Health Integration: What Syncs

Connect Oura, WHOOP, or Garmin to Apple Health and you would reasonably assume your data is now your data — available to any app you point at it.

It isn’t. Every one of these devices holds back its single most valuable number.

Oura’s list of what it shares with Apple Health does not include the Readiness Score.¹ WHOOP’s does not include Recovery, Strain, or Sleep Performance — and WHOOP states outright that it “does not export HRV data to Apple Health at this time.”² Garmin’s list contains no Body Battery, no Training Readiness, no Training Status, and no HRV Status.³

What crosses the bridge is the raw material: heart rate, steps, calories, sleep, and — from Oura and WHOOP — respiratory rate. What stays behind is the interpretation you actually paid for.

Question	Short answer
Does my Readiness/Recovery/Body Battery score reach Apple Health?	No. None of the three proprietary scores appears on its vendor’s share list.
Does WHOOP send HRV to Apple Health?	No — WHOOP explicitly says it does not, citing an SDNN vs rMSSD mismatch.
Does Oura send HRV?	No. HRV is absent from Oura’s export list, though Oura’s own API exposes it.
Does Garmin sync both ways?	No. Garmin pushes to Apple Health only; it never pulls.
Can a third-party app get the full picture?	Through the vendor’s developer API, yes. Through Apple Health alone, no.
Which device shares the most with Apple Health?	Oura — it is the only one of the three that documents sending sleep stages.

What each device actually writes to Apple Health

All three integrations are opt-in, and all three are narrower than most people assume. Here is the complete picture, taken from each vendor’s own support documentation rather than from reviews.

Data type	Oura¹	WHOOP²	Garmin³
Heart rate	Yes (1-minute intervals)	Yes (during sleep and activities)	Yes (all-day; timed activities send high/low only)
Resting heart rate	No	Yes	No
HRV	No	No — explicitly excluded	No
Sleep	Yes, including stages	Yes (stage detail not documented)	Yes (“Sleep Analysis”; stage detail not documented)
Respiratory rate	Yes (nighttime average)	Yes	No
Blood oxygen (SpO2)	No	Yes	No
Active energy	Yes	Yes	Yes
Steps	Yes	Yes (if enabled)	Yes
Workouts	Yes	Yes (auto-detected)	Yes
Workout GPS routes	Yes	Imports only	No
VO2 max	No (imports it, never exports)	No	No
Two-way sync	Yes	Yes	No — push only

Three details in that table are worth pulling out, because they surprise people who have been trusting these integrations for years.

Garmin is one-way and needs the app open. Garmin’s support documentation states plainly that Connect “pushes data to Apple Health (limited by settings/device support) but doesn’t pull data from an Apple watch or Apple Health,” and that the Connect app must be in the foreground for the transfer to happen.³ Garmin data also does not close your Apple Activity rings, and GPS tracks never make the crossing — a workout arrives in Apple Health without its route.

Oura reads a metric it refuses to write. Oura imports Cardio Fitness (VO2 max) from Apple Health but does not export its own.¹ The traffic is deliberately asymmetric.

WHOOP sends you some of the ingredients but not the recipe. WHOOP’s page groups Resting Heart Rate, SpO2, and Respiratory Rate under the heading “Recovery Metrics.”² Those are three of the inputs to the Recovery score — but not the one it leans on hardest. WHOOP’s own documentation says it integrates HRV into the Recovery Score,⁴ and HRV is precisely the field it will not write to Apple Health. So you receive some of what goes into the calculation, none of what comes out, and not the input that matters most.

This is the gap that matters when you want a training app to do something useful with your data — and it is worth being precise about, including on our own behalf. SensAI reads Apple Health, which means it inherits exactly the limits in the table above: from an Apple Watch it sees HRV; from Oura, WHOOP, or Garmin it sees sleep, heart rate, and workouts, but never their proprietary scores and never their HRV. The design response is not to pretend otherwise. It is to interpret the signals that do cross rather than wait for a score that never arrives.

Why WHOOP won’t send HRV at all

WHOOP’s refusal is the most interesting entry on the whole list, because the company explains it. Apple Health stores HRV as SDNN — the standard deviation of normal-to-normal beat intervals. WHOOP calculates rMSSD, the root mean square of successive differences. WHOOP’s position is that writing one into a field defined as the other would be misleading, so it writes nothing.²

That is defensible, and it points at something every multi-device user eventually runs into: these numbers were never supposed to match.

Device	HRV metric	Measurement window
Apple Watch	SDNN⁵	Opportunistic background samples while still — roughly every 4 hours by default, every 15 minutes with AFib History enabled
WHOOP	rMSSD⁴	Recorded during sleep, weighted to the deepest period
Garmin	RMSSD⁶	Continuous across the entire sleep period, averaged from 5-minute windows
Oura	Not named in current documentation⁷	5-minute samples across the night, averaged

Four devices, four different answers, and none of them wrong. SDNN captures total variance including slow respiratory and circadian components; rMSSD is dominated by fast, vagally mediated beat-to-beat change. They are different statistics computed from the same underlying heartbeats, and they do not converge.

Then the windows diverge on top of that. WHOOP samples the deepest sleep, when parasympathetic tone peaks — a window that will tend to produce the highest of the four readings. Apple samples whenever you happen to be sitting still, which could be a stressful Tuesday afternoon.

Garmin says the quiet part directly in its own technical documentation: “Because HRV is a statistical measure, differences in the timing and duration of the measurement affect the results. This makes it a challenge to make apples-to-apple comparisons between HRV measurements produced with alternative protocols and different devices.”⁶

So the practical rule is the one experienced users learn the hard way: never compare your HRV across devices, and never compare it to anyone else’s. Compare it only to your own trailing baseline on one device. We go deeper on that in our guide to what the HRV validation studies actually show.

Sleep stages: only one of the three documents sending them

Apple Health has been able to store real sleep stages since iOS 16 and watchOS 9, when Apple introduced four distinct values — asleepCore (corresponding to AASM stages N1 and N2), asleepDeep (N3), asleepREM, and asleepUnspecified — alongside the older inBed and awake.⁸

Having the container does not mean the vendors fill it.

Oura documents sending stages. Its export list itemizes Sleep as Duration, Start Time, End Time, and Sleep Stages.¹ Oura also warns that the hypnogram may render differently in Apple Health than in the Oura app, because Apple applies its own rounding and display intervals.
Apple Watch writes core, deep, and REM sleep plus awake natively, and has since watchOS 9.⁸
WHOOP’s documentation says only “Sleep tracking.” It does not state whether stages are included.
Garmin’s documentation says only “Sleep Analysis.” Same silence.

We are not going to tell you what WHOOP and Garmin do at the stage level, because neither company documents it and we could not confirm it from a primary source. If a review states it confidently, ask where they got it.

What is worth knowing is that the stage data is the least trustworthy part of the payload regardless of who sends it. In a validation study of six devices against polysomnography, researchers found two-state sleep/wake agreement of 86–89% — but once they asked for multi-state sleep stages, agreement fell to 53% for Apple Watch, 50% for Garmin, 61% for Oura, and 60% for WHOOP.⁹ That study ran on the hardware of its day — Apple Watch Series 6, Garmin Forerunner 245, Oura Generation 2, WHOOP 3.0 — so read the ranking rather than the exact percentages.

That pattern repeats everywhere it has been tested. Across 62 adults, a 2025 study in Sleep Advances from the group of Johan Verbraecken, MD, PhD, at the Multidisciplinary Sleep Disorders Centre, Antwerp University Hospital, found six wrist wearables each detected over 90% of sleep epochs — but achieved specificity of only 29.4% to 52.2%, with Cohen’s κ between 0.21 and 0.53.¹⁰ High sensitivity, poor specificity: the devices are good at noticing you are asleep and bad at noticing you are awake.

An earlier comparison of seven consumer sleep trackers — a different device set, including two Garmin watches — found the same asymmetry against polysomnography: epoch-by-epoch sensitivity of 0.93 or better, specificity of just 0.18 to 0.54. It also found performance degraded further on nights with disrupted sleep.¹¹ Precisely the nights you most want read correctly.

There is a genuine bright spot. Work published in Sensors by the group of Charles A. Czeisler, MD, PhD, of the Division of Sleep Medicine at Harvard Medical School and the Division of Sleep and Circadian Disorders at Brigham and Women’s Hospital, tested three commercial devices against polysomnography in 35 healthy adults and found sleep-versus-wake sensitivity of at least 95% for all of them. At the stage level, Oura showed no statistically significant difference from polysomnography for wake, light, deep, or REM — while Fitbit and Apple Watch showed systematic discrepancies.¹² For a fuller treatment, see our breakdown of deep sleep accuracy across wearables.

The data that does cross is not uniformly good either

It would be convenient if everything that made it through the pipe were solid. It isn’t, and the differences are large enough to change decisions.

HRV quality varies enormously by device. Over 536 nights of ECG-referenced data, a team writing in Physiological Reports from Joshua A. Hagen, PhD’s Human Performance Collaborative at The Ohio State University found Oura’s fourth-generation ring reached a concordance correlation coefficient of 0.99 for nocturnal HRV, with a mean absolute percentage error of 5.96%.¹³ That is close to laboratory quality from a consumer ring.

Apple Watch did not fare as well in a comparable test. A University College Dublin team under Cailbhe Doherty, PhD, of the School of Public Health, Physiotherapy and Sports Science, took 316 measurements from Apple Watch Series 9 and Ultra 2 against a Polar H10 chest strap across 39 adults over two weeks. The watches underestimated HRV by a mean of 8.31 ms, with a 28.9% mean absolute percentage error, failing the study’s pre-specified ±10 ms equivalence margin. Resting heart rate from the same devices was excellent — a mean difference of −0.08 bpm.¹⁴

Read those two findings together and the lesson is precise: the same device can be trustworthy on one metric and unreliable on another. “Is my wearable accurate?” is the wrong question. “Which of its numbers is accurate?” is the right one.

Even step count — the simplest metric any of these devices produces — has been shakier than it looks. In a Framingham Heart Study analysis of 523 participants across 3,223 person-days, an Apple Watch and a research-grade accelerometer produced an intraclass correlation of just 0.56, a mean difference of 499 steps per day, and agreement within 15% on only 29.7% of days.¹⁵ Worth reading with a caveat the authors state themselves: that study used the original Apple Watch Series 0, and they note the limitations “may be improved in newer devices.” Take it as evidence that agreement between any two step counters is worse than intuition suggests, not as a verdict on current hardware.

And derived metrics degrade fastest of all. Comparing three wearables against a standardized sleep diary across four nights, researchers found concordance correlation coefficients of 0.76 to 0.85 for total sleep time — but only 0.05 to 0.34 for sleep efficiency, and effectively zero for sleep interruptions.¹⁶ That is worth sitting with, because sleep efficiency and interruptions are exactly the ingredients your Readiness and Recovery scores are built from. The totals hold up; the derivatives do not.

Everything in this section is why SensAI treats a wearable reading as evidence rather than as truth: a number gets weighted by how well that specific device measures that specific thing, and a single anomalous night never overrides a stable trend.

Apple Health is a lossy lane — the developer API is the real one

Here is the part that reframes the whole problem. The scores that never reach Apple Health are not secret. They are simply routed somewhere else.

WHOOP’s Developer Platform exposes a Recovery object containing recovery_score, resting_heart_rate, hrv_rmssd_milli, spo2_percentage, and skin_temp_celsius — the exact HRV field WHOOP declines to write into HealthKit, available to any authorized app via OAuth.¹⁷
Oura’s API v2 exposes daily readiness with its score, temperature deviation, and nine named contributors, plus the full 30-second hypnogram and overnight HRV series.¹⁸
Garmin’s Health API exists too, but it is approval-gated behind a commercial developer program rather than self-serve. Its public feature list names steps, intensity minutes, sleep, calories, heart rate, stress, pulse ox, body composition, respiration, blood pressure, beat-to-beat intervals — and Body Battery, the score Apple Health never receives.¹⁹

So “does this app work with my Oura Ring?” has two different answers depending on how the app connects. An app that reads only Apple Health gets a partial, score-free feed. An app that integrates the vendor API directly can see what the vendor sees.

When you are evaluating any training app, that is the question to ask — and it is a better filter than a feature list. Our roundup of HRV apps that work with Oura and WHOOP breaks down which apps take which route.

For the record, SensAI takes the HealthKit route, not the vendor-API route. That is a real constraint and we would rather state it than have you discover it: if your Readiness Score is the number you want acted on, the vendor’s own app is where it lives.

What this means for how you actually train

Strip away the plumbing and three practical conclusions survive.

1. Your scores don’t travel, so don’t build a system that assumes they do. If your plan depends on Body Battery or Recovery reaching a third-party app through Apple Health, it will silently not work. Either use the vendor’s own app for that number, or use an app with a direct API integration.

2. Cross-device comparison is a trap. Two devices reporting “HRV” are reporting different statistics over different windows. A drop on one and a rise on the other is not a contradiction to be resolved — it is two instruments answering two questions. We unpack how each vendor builds its composite in how Body Battery, Recovery, and Readiness are actually calculated.

3. The raw signals are the durable asset. Five things cross from all three devices: heart rate, sleep, steps, workouts, and active energy. Those are also the measurements the validation literature supports best — sleep/wake detection and heart rate are strong across every device tested, while the derived scores built on top of them are not. A system built on the raw signals will outlive any particular vendor’s scoring algorithm, and any particular vendor.

That last point is the design principle behind SensAI. Rather than trusting a black-box readiness number, it reads the underlying HealthKit signals your device does share, weighs them against your own history and your training plan, and explains the reasoning. When your overnight heart rate is elevated four nights running, the useful output is not a score out of 100 — it is a specific change to today’s session, and a reason.

If you are still choosing hardware, our wearable comparison and our data-quality criteria for AI coaching cover which device to buy before you worry about where its data goes.

Frequently asked questions

Does the Oura Readiness Score sync to Apple Health?

No. Oura’s documented list of data shared with Apple Health includes sleep with stages, heart rate, respiratory rate, active energy, steps, workouts, weight, and height — but not the Readiness Score, Sleep Score, or Activity Score.¹ Those remain in the Oura app. They are, however, available through Oura’s API v2 to apps that integrate it directly.¹⁸

Why is my WHOOP HRV missing in Apple Health?

Because WHOOP does not send it. WHOOP states that Apple Health uses SDNN while WHOOP calculates rMSSD, and that it “does not export HRV data to Apple Health at this time.”² The number is available through the WHOOP Developer Platform as hrv_rmssd_milli.¹⁷

What data can another app get from my Oura Ring?

That depends on which route the app takes, and there are only two. An app reading Apple Health can access what Oura writes there — sleep with stages, heart rate, respiratory rate, steps, active energy, and workouts. An app integrating Oura’s API v2 directly can additionally access Readiness, Sleep Score, the overnight HRV series, SpO2, and the full hypnogram.¹⁸ Ask which route an app takes before assuming it sees everything. For which specific apps take which route, see our roundup of HRV apps that work with Oura and WHOOP.

Does Garmin sync with Apple Health both ways?

No. Garmin Connect pushes to Apple Health but never pulls from it, and the Connect app has to be open in the foreground for the transfer to run.³ Garmin also does not send GPS routes with workouts, and its data will not close your Apple Activity rings.

Does Body Battery appear in Apple Health?

No. Garmin’s list of what Apple Health can receive covers active energy, body fat percentage, BMI, flights climbed, heart rate, resting energy, sleep analysis, steps, distance, water, weight, and workouts.³ Body Battery, Training Readiness, Training Status, HRV Status, Stress, Pulse Ox, and sleep score are not on it.

Which wearable shares the most data with Apple Health?

Of the three, Oura — it is the only one that documents writing sleep stages, and it sends workout routes, which Garmin does not. WHOOP sends more physiological channels than Garmin (resting heart rate, SpO2, respiratory rate) but withholds HRV. Garmin sends the narrowest set and only in one direction.

Are the numbers that do sync accurate?

It depends entirely on which number. Nocturnal HRV from a recent Oura ring reached a concordance correlation coefficient of 0.99 against ECG,¹³ while Apple Watch HRV underestimated a chest-strap reference by 8.31 ms on average.¹⁴ Sleep/wake detection is strong across devices; sleep staging is weak, at roughly 50–61% agreement with polysomnography.⁹ Treat each metric on its own record.

The bottom line

The integration you switched on does less than you think. Every proprietary score — Oura Readiness, WHOOP Recovery, Garmin Body Battery — stays inside the app that generated it. WHOOP withholds HRV outright, on a technically sound argument about SDNN versus rMSSD. Garmin’s pipe runs one direction only and drops your GPS tracks on the way.

What does cross is the raw physiology: heart rate, sleep, energy, steps, and respiratory rate from everything except Garmin. That is a narrower feed than the marketing implies, but it is also the part the validation literature supports best — and the part that stays yours when you switch devices.

Build on those. Treat vendor scores as one opinion from one algorithm you cannot inspect, use the trend on a single device rather than the number across several, and choose training tools that either integrate the vendor API properly or are honest about reading only what Apple Health carries.

References

Oura. “Apple Health Integration.” Oura Support, last updated July 21, 2026. https://support.ouraring.com/hc/en-us/articles/360025438734-Apple-Health-Integration ↩ ↩² ↩³ ↩⁴ ↩⁵
WHOOP. “Apple Health Integration.” WHOOP Support, last updated April 16, 2026. https://support.whoop.com/s/article/Apple-Health-Integration ↩ ↩² ↩³ ↩⁴ ↩⁵
Garmin. “Sharing Your Garmin Connect Data With Apple Health.” Garmin Support Center. https://support.garmin.com/en-US/?faq=lK5FPB9iPF5PXFkIpFlFPA ↩ ↩² ↩³ ↩⁴ ↩⁵
WHOOP. “Heart Rate Variability (HRV) Insights & WHOOP Metrics.” WHOOP Support, May 28, 2025. https://support.whoop.com/s/article/Heart-Rate-Variability-HRV-Insights-WHOOP-Metrics ↩ ↩²
Apple Inc. “Heart Rate, Calorimetry, and Activity on Apple Watch.” Apple health technology whitepaper, November 2024. https://www.apple.com/health/pdf/Heart_Rate_Calorimetry_Activity_on_Apple_Watch_November_2024.pdf ↩
Garmin. “HRV Status.” Garmin Health Science. https://www.garmin.com/en-US/garmin-technology/health-science/hrv-status/ ↩ ↩²
Oura. “Heart Rate Variability.” Oura Support, last updated July 14, 2026. https://support.ouraring.com/hc/en-us/articles/360025441974-Heart-Rate-Variability ↩
Apple Inc. “HKCategoryValueSleepAnalysis.” Apple Developer Documentation. https://developer.apple.com/documentation/healthkit/hkcategoryvaluesleepanalysis ↩ ↩²
Miller DJ, Sargent C, Roach GD. “A Validation of Six Wearable Devices for Estimating Sleep, Heart Rate and Heart Rate Variability in Healthy Adults.” Sensors (Basel), 2022;22(16):6317. https://pubmed.ncbi.nlm.nih.gov/36016077/ ↩ ↩²
Schyvens AM, Peters B, Van Oost NC, Aerts JM, Masci F, Neven A, Dirix H, Wets G, Ross V, Verbraecken J. “A performance validation of six commercial wrist-worn wearable sleep-tracking devices for sleep stage scoring compared to polysomnography.” Sleep Advances, 2025;6(2):zpaf021. https://pubmed.ncbi.nlm.nih.gov/40303381/ ↩
Chinoy ED, Cuellar JA, Huwa KE, Jameson JT, Watson CH, Bessman SC, Hirsch DA, Cooper AD, Drummond SPA, Markwald RR. “Performance of seven consumer sleep-tracking devices compared with polysomnography.” Sleep, 2021;44(5):zsaa291. https://pubmed.ncbi.nlm.nih.gov/33378539/ ↩
Robbins R, Weaver MD, Sullivan JP, Quan SF, Gilmore K, Shaw S, Benz A, Qadri S, Barger LK, Czeisler CA, Duffy JF. “Accuracy of Three Commercial Wearable Devices for Sleep Tracking in Healthy Adults.” Sensors (Basel), 2024;24(20):6532. https://pubmed.ncbi.nlm.nih.gov/39460013/ ↩
Dial MB, Hollander ME, Vatne EA, Emerson AM, Edwards NA, Hagen JA. “Validation of nocturnal resting heart rate and heart rate variability in consumer wearables.” Physiological Reports, 2025;13(16):e70527. https://pubmed.ncbi.nlm.nih.gov/40834291/ ↩ ↩²
O’Grady B, Lambe R, Baldwin M, Acheson T, Doherty C. “The Validity of Apple Watch Series 9 and Ultra 2 for Serial Measurements of Heart Rate Variability and Resting Heart Rate.” Sensors (Basel), 2024;24(19):6220. https://pubmed.ncbi.nlm.nih.gov/39409260/ ↩ ↩²
Spartano NL, Zhang Y, Liu C, Chernofsky A, Lin H, Trinquart L, Borrelli B, Pathiravasan CH, Kheterpal V, Nowak C, Vasan RS, Benjamin EJ, McManus DD, Murabito JM. “Agreement Between Apple Watch and Actical Step Counts in a Community Setting: Cross-Sectional Investigation From the Framingham Heart Study.” JMIR Biomedical Engineering, 2024;9:e54631. https://pubmed.ncbi.nlm.nih.gov/39047284/ ↩
Klier K, Wagner M. “Agreement of Sleep Measures—A Comparison between a Sleep Diary and Three Consumer Wearable Devices.” Sensors (Basel), 2022;22(16):6189. https://pubmed.ncbi.nlm.nih.gov/36015949/ ↩
WHOOP. “Developer Platform — API Reference (v2).” https://developer.whoop.com ↩ ↩²
Oura. “Oura API v2 Documentation.” https://cloud.ouraring.com/v2/docs ↩ ↩² ↩³
Garmin. “Health API — Garmin Connect Developer Program.” https://developer.garmin.com/gc-developer-program/health-api/ ↩