Wearable Health Tech: From Step Counts to Clinical Signals

A wrist device that once counted steps now takes an ECG, estimates blood oxygen, flags irregular heart rhythms, tracks skin temperature, detects sleep apnoea and — on some models — measures blood pressure trends. Hundreds of millions of people wear one. The interesting questions are which of those measurements you can trust, which have been validated to a clinical standard, and where the data goes once it leaves your wrist.

How the sensors work

• Optical heart rate (PPG). Green LEDs shine into the skin; a photodiode measures how much light returns, which varies with blood volume as the heart beats. Cheap, continuous, and the basis of most wearable metrics — heart rate, heart-rate variability, and (with red and infrared LEDs) blood-oxygen estimates. Accuracy suffers with motion, poor fit, cold skin and, historically, darker skin tones, which absorb more green light; newer devices compensate but the effect hasn't disappeared entirely.
• Electrical (ECG). Electrodes on the back of the watch and the crown or bezel measure the heart's electrical signal when you touch both. A single-lead ECG can identify atrial fibrillation; it can't diagnose a heart attack.
• Accelerometers and gyroscopes for steps, activity classification, fall detection and sleep staging (which is inferred from movement and heart rate, not brain waves — sleep-stage accuracy is modest).
• Skin temperature, used for cycle tracking and illness detection; electrodermal activity for stress; bioimpedance for body composition; and in rings and patches, additional form factors for the same sensors.
• Continuous glucose monitors — a separate category of medical wearable using a subcutaneous filament — have moved from diabetes management into consumer "metabolic health" products, with over-the-counter versions now available.

Consumer feature vs medical device

This is the line that matters. A wellness feature — step counts, "readiness scores", sleep stages — is unregulated and its accuracy is whatever the manufacturer chooses. A medical-device feature has been submitted to a regulator with clinical evidence. The ECG and irregular-rhythm notifications on Apple, Samsung, Google and Withings devices have FDA clearance and CE marking; sleep-apnoea detection and hypertension notifications on the latest devices have followed. Blood-oxygen readings mostly have not been cleared for medical use, and the disclaimer text says so. When a wearable says something matters medically, check whether that specific feature is cleared — the device being expensive is not evidence.

What the evidence shows

Large studies have established that wearable atrial-fibrillation detection has high positive predictive value in older populations and can lead to earlier diagnosis and stroke prevention — the strongest clinical result in the field. Activity tracking modestly increases physical activity, especially with goal-setting and feedback. Wearable data helped identify infection onset before symptoms during the pandemic. Remote patient monitoring with clinical-grade wearables reduces hospital readmissions for heart failure and COPD when integrated into a care pathway.

The cautions are equally established. Screening a young, healthy population for arrhythmia produces false positives and anxiety without much benefit. Metrics like heart-rate variability and sleep scores vary between devices and are not comparable. And "more data" doesn't help a clinician unless it arrives in a form that fits their workflow — a PDF of six months of heart-rate graphs does not.

Where the data goes

Health data from consumer wearables is, in most jurisdictions, not covered by medical-privacy law like HIPAA, which applies to healthcare providers rather than device makers. It's governed by consumer-privacy law and the company's terms — and those terms have historically permitted sharing with partners, use for advertising, and sale on acquisition. Regulators are catching up: the US FTC has taken enforcement actions over health-data sharing, several states have passed consumer health-data laws, and GDPR treats health data as a special category. Practical advice: read the data-sharing settings; use on-device processing and local storage where offered; be cautious with third-party apps that request full health-data access; and know that insurers and employers offering wearables in wellness programmes may see aggregate or individual data depending on the arrangement.

Where it's heading

Three developments are shaping the next generation. Non-invasive measurement of things that currently need blood — glucose above all — remains the holy grail, still unsolved despite years of announcements. Continuous, passive clinical monitoring is moving from research into products: cuffless blood pressure, respiratory monitoring, and early-warning models that combine signals — increasingly built into garments as well as wrist devices. And AI health assistants that interpret your data conversationally are being built into the major platforms, which raises the medical-device question again: at what point does an interpretation become a diagnosis? That question, more than any sensor, will define the field's next few years.