Apple Watch has evolved from a fitness tracker into a genuine medical monitoring device, earning multiple FDA clearances along the way. But most users activate maybe 30% of the health capabilities built into the device. This guide covers every major health feature, the evidence behind it, and how to interpret the data meaningfully.
ECG and Atrial Fibrillation Detection
The ECG app, first introduced on Apple Watch Series 4 in 2018, was the company's first FDA-cleared medical device feature. It records a single-lead electrocardiogram (Lead I) by measuring the electrical potential between your left arm (wrist touching the watch) and your right arm (finger touching the Digital Crown).
The 30-second recording is classified by an algorithm into four categories: Sinus Rhythm (normal), Atrial Fibrillation (irregular, potentially dangerous rhythm), Low or High Heart Rate, or Inconclusive. Apple received FDA clearance based on a validation study showing 99.6% sensitivity and 99.4% specificity for AFib detection.
More significant was the Apple Heart Study, which enrolled 419,297 participants — the largest arrhythmia study ever conducted. The study found that the Irregular Rhythm Notification (the passive, background AFib screening that runs via optical sensors) had a positive predictive value of 84% for confirmed AFib. Importantly, it identified AFib in 0.52% of participants, many of whom had no prior diagnosis.
What ECG Cannot Detect
The watch cannot diagnose heart attacks, coronary artery disease, heart failure, or most other cardiac conditions. Single-lead ECG misses many rhythm abnormalities that a 12-lead clinical ECG would catch. It's a screening tool, not a diagnostic one. But for its intended purpose — AFib detection — the evidence is strong.
Blood Oxygen (SpO2) Monitoring
Apple Watch Series 6 and later can measure blood oxygen saturation using red and infrared LEDs combined with a photodiode on the back crystal. Measurements can be taken on-demand (15 seconds) or automatically during sleep.
The system measures SpO2 in the 70–100% range with ±2% accuracy per Apple's specifications. In practice, several studies have found performance varies by skin tone — darker skin can cause systematic overestimation of SpO2. Apple has faced FDA pressure on this point, and the company is calibrating across a broader range of skin tones in newer validation studies.
Clinical Uses
- Sleep apnea screening: Consistent SpO2 dips below 90% during sleep warrant discussion with a physician about overnight pulse oximetry testing
- Altitude adjustment: Tracking acclimatization during high-altitude travel
- COVID-19 recovery monitoring: During the pandemic, SpO2 tracking helped flag deteriorating cases, though clinical pulse oximeters remain more accurate
- COPD and asthma trending: Patients can track correlation between symptoms and saturation
Understanding the clinical framing for consumer SpO2 monitoring matters before acting on readings. Unlike medical pulse oximetry, Apple Watch SpO2 uses reflective photoplethysmography from the wrist, which carries inherently lower accuracy than fingertip pulse oximeters — validation studies report mean absolute errors of 1.5–3% for wrist-based reflective devices versus less than 1% for FDA-cleared fingertip devices. The practical sweet spot is detecting trends and episodic desaturations during sleep as a potential OSA screening signal, not capturing precise absolute spot readings. Data from the Apple Research Study showed that consistent nocturnal SpO2 dips below 90% in Watch wearers correlated with self-reported excessive daytime sleepiness — a finding consistent with obstructive sleep apnea's known physiological signature. The practical guidance: treat Watch SpO2 as a screening tool for patterns over days and weeks, not a medical-grade measurement for acute clinical monitoring.
Skin Temperature Sensing
Introduced with Apple Watch Series 8, the skin temperature system uses dual sensors — one measuring wrist temperature and one ambient temperature — to calculate relative changes from your baseline. The watch does not display an absolute temperature; it shows deviations from your nightly baseline.
This feature powers two capabilities: (1) Cycle Tracking retrospective ovulation confirmation, where the luteal phase temperature rise (typically 0.1–0.2°C) can be detected; and (2) Sleep tracking's illness detection flag, where unusual temperature deviations appear in the Health app.
For menstrual cycle tracking, a 2023 validation study in npj Digital Medicine found wrist temperature could detect ovulation with sensitivity comparable to luteinizing hormone test strips, though with a 1–2 day lag. Apple Watch is not a substitute for LH strips if you're trying to conceive, but it provides useful retrospective cycle phase data.
Heart Rate Variability (HRV)
Apple Watch measures HRV nightly, reporting SDNN (Standard Deviation of Normal-to-Normal intervals) averaged during sleep. Nighttime SDNN is one of the most validated HRV metrics in cardiovascular medicine. Population norms show SDNN declines with age: healthy 25-year-olds average 65–95ms; 55-year-olds average 45–70ms.
The Health app shows your HRV trend over time. Consistent declines can indicate: cumulative training stress, illness onset (1–3 days before symptoms), poor sleep quality, alcohol effects, or advancing cardiovascular disease risk. Many competitive athletes and executive health patients now monitor HRV as their primary readiness biomarker.
Crash Detection and Fall Detection
Fall Detection (Series 4+) uses accelerometer and gyroscope to identify hard fall signatures followed by immobility. Crash Detection (Series 8+ with iPhone 14+) employs a high-g accelerometer rated to 256g, microphone, barometer, and GPS to identify motor vehicle crashes. Both will contact emergency services and share location if you're unresponsive.
Apple published validation data showing Crash Detection activates in 96% of severe crash scenarios while maintaining low false positive rates — though athletes performing high-impact sports such as skiing, mountain biking, and contact sports have occasionally triggered false alerts, a manageable tradeoff given the life-safety stakes. Fall Detection in independent clinical evaluations detects hard falls with sensitivity above 90% in older adults, though specificity is lower in highly active users. Both features automatically call emergency services and share your GPS location if you remain unresponsive for approximately 60 seconds. The populations deriving the most life-safety benefit are solo athletes in remote environments, elderly individuals living alone, and frequent long-distance drivers — groups where delayed emergency response is the primary determinant of outcome.
Sleep Staging
watchOS 9 introduced multi-stage sleep tracking: Core (light), Deep (slow-wave), and REM sleep, plus awake periods. The algorithm uses heart rate, heart rate variability, respiratory rate, and movement to classify sleep stages. In validation studies against polysomnography (PSG), consumer wearables including Apple Watch achieve about 70–80% agreement on sleep stage classification — better than nothing, but not clinical-grade. The most reliable metrics are total sleep time and time in bed.
Stage-level accuracy is uneven: REM detection tends to be the most reliable in wearable research (sensitivity ~80%), while N1 (light transitional sleep) is frequently misclassified, and N2/N3 (deep slow-wave) separation is inconsistent across devices. For practical clinical use, treat the staging data as directional rather than definitive. Actionable patterns include: consistently low deep sleep percentages (below ~13% of total sleep) paired with daytime fatigue, REM durations that are markedly shorter than age-expected norms (~20–25% in adults), or highly fragmented sleep with frequent brief awakenings. If you observe persistently abnormal patterns — especially frequent oxygen desaturations alongside disrupted staging — share the exported data with a sleep medicine physician rather than self-diagnosing; a formal PSG remains the gold standard for diagnosing sleep apnea, periodic limb movement disorder, and REM sleep behavior disorder.
Using Your Data in Medical Appointments
The practical value of Apple Watch health data has grown substantially as physicians adapt to reviewing wearable data. Most useful for clinical discussions:
- Resting heart rate trends: Sustained elevation above your baseline warrants investigation
- HRV decline: Relevant to cardiologists and functional medicine physicians
- ECG PDFs: Export via Health app → Browse → Heart → Electrocardiograms → select recording → Export PDF
- Irregular rhythm notifications: A clinical event worth documenting and discussing
- Overnight SpO2 data: Repeated dips below 90% are actionable
To export all health data: Health app → Profile icon → Export All Health Data. This generates a ZIP file with XML data that can be shared with healthcare providers or imported into health platforms like QuanMed AI for pattern analysis.
Series Comparison: Which Watch to Buy for Health Monitoring
| Feature | Series 6/7 | Series 8/9 | Series 10/Ultra |
|---|---|---|---|
| ECG + AFib | Yes | Yes | Yes |
| Blood Oxygen | Yes | Yes | Yes |
| Skin Temperature | No | Yes | Yes |
| Crash Detection | No | Yes (+ iPhone 14) | Yes |
| Sleep Staging | No | Yes | Yes |
| High-G Accelerometer | No | Yes | Yes |
For comprehensive health monitoring, Series 8 or later provides the full feature set. Series 6 and 7 remain excellent for core cardiovascular monitoring but lack temperature sensing and crash detection.
Limitations to Keep in Mind
Apple Watch is a powerful wellness and screening tool, not a medical diagnostic device for most purposes (except AFib detection, which has genuine clinical-grade evidence). The watch works best as a longitudinal data source — trends over weeks and months reveal far more than any single reading. Don't make medical decisions based on isolated readings; look for persistent patterns and use the data as a conversation starter with your healthcare team.
There are specific scenarios where you should not rely on wearable data: if you are experiencing acute chest pain, shortness of breath, syncope, or neurological symptoms, seek emergency care immediately — do not wait for the ECG app or oxygen reading to confirm anything. The watch is also inappropriate for rule-out purposes; a normal ECG strip does not exclude coronary artery disease, structural heart disease, or non-AFib arrhythmias. The critical distinction is between screening (identifying asymptomatic individuals who warrant further investigation) and diagnosis (confirming or excluding a specific condition) — Apple Watch performs the former, not the latter. To get full value from your wearable data, pair it with annual professional health evaluations including standard bloodwork, resting ECG, and physical examination. A clinician who reviews both your wearable trends and objective clinical data will extract far more actionable insight than either source provides in isolation.