Apple Watch Series 12 and iOS 27: What's New in Apple Health, and the New RMSSD Type in HealthKit

Apple announced Apple Watch Series 12 on September 9, 2026, alongside a redesigned Health app and a set of Apple Intelligence health features. The watch reaches stores on September 18, and watchOS 27 shipped on September 14.

Alongside the launch, iOS 27 adds a new HealthKit quantity type for RMSSD, the heart rate variability metric most of the wearable industry standardized on years ago and that HealthKit had not exposed until now. Apple did not mention it in the announcement.

What follows is the full announcement, the change to the API, and the consequences for existing health apps. On sensor accuracy itself, our analysis of Apple Watch heart rate accuracy draws on 82 studies and more than 430,000 participants.

What's New in Apple Watch Series 12

The Series 12 Heart Sensing System

Apple redesigned both the optical and electrical heart sensors, fitting larger power-efficient green LEDs to the optical sensor.

Heart rate is now sampled every five seconds throughout the day.

Heart rate variability is measured up to 24 times more often than before. Apple published the multiplier without a corresponding sampling interval, leaving the exact cadence unstated.

Two HRV views appear in the Vitals app. Apple describes Recovery HRV as best for identifying daily signals of stress and recovery, and overall HRV as best for broader health including cardiovascular health. Overnight vitals now include Recovery HRV compared against a personal baseline, and a new daytime view switches between overnight and daytime metrics.

Accuracy claims rest on a study Apple ran in July and August 2026 across more than 1,000 participants, reporting the highest accuracy among the devices tested. The pedometer was rebuilt with new machine learning models, validated with video step counting against smartwatches available as of July 2026.

Apple Watch Series 12 Price and Release Date

Series 12 runs on the S11 chip in 42mm and 46mm sizes, across aluminum, titanium, and ceramic finishes, behind Ceramic Shield 2 that Apple states is 60 percent tougher than Ion-X.

Pricing starts at $399. Pre-orders opened September 9 and the watch reaches stores on September 18 in more than 50 countries. Battery life runs up to 24 hours in everyday use, or about 10 hours of outdoor workout tracking, with a 15-minute charge delivering up to 12 hours.

How the Apple Watch Readiness Score Works

Readiness produces a single score from 0 to 10 each day, with one of four recommendations: Recover, Pace Yourself, Ready, or Go For It.

It draws on recent activity, training load, vitals, and sleep score, and updates through the day. Apple developed the scoring algorithm using data from the Apple Heart and Movement Study, working with exercise scientists and physicians on staff.

What's New in the Apple Health App

The redesigned Health app organizes around two new tabs.

Insights surfaces timely health information with a summary that updates through the day, spanning heart, sleep, readiness, fitness, vitals, and cycle tracking.

Longevity analyzes longitudinal data across heart health, sleep, mental wellbeing, movement, metabolic health, hearing, and nutrition.

Health Age sits inside Longevity and compares metrics against the user's chronological age. The algorithm factors in Apple Watch metrics including VO2 max, resting heart rate, sleep, and HRV, and users can add lab values such as A1c and LDL.

Movement evaluations measure flexibility, strength, balance, movement mechanics, and VO2 max using the iPhone camera together with the watch. The experience runs entirely on device, and no video is recorded, stored, or shared.

Labs lets users schedule and purchase panels covering more than 50 biomarkers for $119 at approximately 2,000 Quest Diagnostics locations in the United States.

Apple Health Features: What Ships Now and What Comes Later

The announcement mixes shipping features with ones still months out, and the split matters for anyone planning around it.

Available now: Series 12 hardware and its sensing system from September 18, the Readiness score, and watchOS 27 from September 14.

Later this year: the redesigned Health app, starting in US English and requiring an Apple Intelligence capable device. Health Age, Labs, and movement evaluations arrive with it.

Not new: hypertension notifications, which Apple introduced in September 2025. Sleep apnea notifications and the ECG app predate this launch by longer still.

What's New in watchOS 27

watchOS 27 arrived on September 14 for Series 9 and later, SE 3, and Ultra 2 and later, and requires an iPhone 11 or later running iOS 27.

Cycle Tracking adds perimenopause and menopause support, with users age 40 and over notified when logged cycle patterns suggest perimenopause.

Workout Buddy now runs without the iPhone nearby and is available in English and Spanish.

GymKit was extended so AirPods Pro 3 users can sync heart rate data through iPhone to compatible treadmills and indoor bikes.

RMSSD in HealthKit: What Changed in iOS 27

HealthKit has carried exactly one heart rate variability type since iOS 11 shipped in 2017: the standard deviation of NN intervals, or SDNN. Every iOS health app reading HRV has been reading SDNN, whether or not that was the right metric for the job.

iOS 27 adds a second one. The identifier for RMSSD, the root mean square of successive differences, is now part of HealthKit across iOS, iPadOS, watchOS, macOS, visionOS, and Mac Catalyst, all at version 27.0.

It is the only addition among HealthKit's 122 quantity types in this release. Apple's documentation lists it at version 27.0 on every platform, and the page itself remains a bare declaration with no description, which is how a newly added symbol appears before the documentation catches up.

Apple has not said how the new type relates to the Recovery HRV and overall HRV split in the Vitals app. The division lines up with conventional usage of RMSSD and SDNN, though Apple has published no mapping, which leaves the correspondence unconfirmed until either the documentation or the data on device settles it.

RMSSD vs SDNN: What the Two Metrics Measure

Both metrics describe variation in the time between heartbeats, and both are reported in milliseconds. They measure different things.

SDNN is the standard deviation of the intervals across a recording window. It captures total variability from every source, including slow fluctuations driven by circadian rhythm, temperature, and physical activity. Its interpretation depends heavily on window length, which is why SDNN from a 24 hour recording and SDNN from a five minute reading are not comparable numbers.

RMSSD takes the difference between each pair of consecutive beats, squares those differences, averages them, and takes the square root. Working on beat-to-beat change rather than the spread of the whole recording, it primarily reflects parasympathetic activity, the branch of the autonomic nervous system associated with rest and recovery.

It is also more stable over short recordings and less distorted by breathing patterns. That is why RMSSD became the default for overnight recovery measurement across the wearable industry and in most of the sports science literature.

Sensor quality sits underneath both figures, and contact pressure alone can change an HRV reading, as our analysis of what SDNN reveals about wearable accuracy documents.

The two values are not interchangeable, and a single night can move them in opposite directions. Apps that have displayed HealthKit HRV alongside a figure from another device have in most cases been placing SDNN next to RMSSD. Our breakdown of health scores built on wearable data covers how context such as age and fitness level shapes any HRV reading.

What the HealthKit Change Means for Health Apps

The release changes four things for existing iOS health apps.

Two HRV types instead of one. Existing permission flows do not include RMSSD, so reading it requires a new authorization request. For apps built on native access to wearable data that amounts to a permissions migration.

Historical data is not backfilled either, so RMSSD series begin when a user updates and wears a supporting device. The pattern is familiar from other providers, covered in why mobile health apps struggle with wearable integrations.

Sample volume rises sharply. Heart rate every five seconds amounts to 17,280 samples per user per day where every reading is persisted. At that density, raw sample arrays become an expensive way to read the store, and statistics collection queries and anchored object queries carry the load instead, with background delivery running closer to hourly than continuous.

Baselines shift mid-cohort. A user moving from an older watch to Series 12 changes sampling density underneath any baseline already built from their history. A recovery figure calibrated on a handful of overnight readings behaves differently once it receives an order of magnitude more of them, and rolling baselines computed across that transition can move without any physiological change behind them.

Apple's derived scores stay inside Apple. Readiness, Health Age, and the Longevity analyses are computed by Apple and surfaced in Apple's own app. No HealthKit type exposes a Readiness score to third-party apps, and Apple has not published the algorithm. The inputs are readable; the output is not.

How to Normalize HRV Data Across Apple Watch and Other Wearables

Readiness reaches only the people wearing an Apple Watch. Garmin, Oura, Whoop, and Polar each publish their own recovery or readiness figure, computed differently and expressed on a different scale.

That leaves a 0 to 10 Readiness value and another vendor's recovery percentage with no common basis for comparison or averaging. It ranks among the top challenges when working with wearables in healthcare.

Products spanning several devices generally end up computing a score of their own, from data they normalize themselves, which means owning both the ingestion layer and the scoring logic rather than inheriting whatever each vendor provides.

Open Wearables: An Open-Source Layer for Apple Health and Wearable Data

Open Wearables is the open-source health intelligence platform Momentum maintains. It ingests data from a growing set of wearable sources into one schema and runs self-hosted, keeping health data on the operator's own infrastructure and inside their compliance boundary. The reasoning behind it is set out in our post introducing Open Wearables.

Its unified data model has carried separate fields for SDNN and RMSSD since before Apple exposed RMSSD in HealthKit, because most non-Apple providers already reported it. Apps reading RMSSD through Open Wearables receive the new Apple data in the field they were already querying.

The Resilience Score shows what a documented score looks like. It runs 0 to 100 on a weekly rolling window, updated daily, measuring the stability of overnight HRV through the coefficient of variation of nightly RMSSD averages across a seven-night lookback.

It requires at least five nights of valid data, filters samples to detected sleep stages, and returns null rather than a guess below that threshold. Each response carries the metric type used and a night-by-night breakdown.

Lower variation indicates steadier recovery, while higher variation can reflect stress, illness, or overtraining. The methodology is published and the implementation is open source, leaving the calculation open to inspection or modification. Sleep Score and Resilience Score are the two scores currently live.

For Series 12 data heading into a team's own backend, Open Wearables ships a native Swift SDK for HealthKit background delivery, with Flutter and React Native wrappers over the same native layer, covering the plumbing described in our guide to accessing Apple Health data. Heart Monitor, an independent adopter, runs the platform in production at around 90,000 monthly active users.

Momentum's wearables integration team works with health products on this problem, and our analysis of what you can and cannot do with HealthKit data covers the platform constraints behind the build-or-adopt decision.