The inflatable arm cuff has measured blood pressure the same way since the 1900s. Squeeze, listen, release, record. It works, but it captures one number at one moment — and hypertension doesn’t behave in single moments.
A new generation of wearables is skipping the cuff entirely. Optical sensors, accelerometers, and machine learning models now estimate blood pressure from the shape and timing of a pulse, turning a once-a-day clinic reading into a continuous stream of data.
Nearly half of U.S. adults — 48.1%, or roughly 120 million people — live with high blood pressure, according to the CDC. Nearly one in four has it under control. That gap between prevalence and control is exactly what continuous, passive monitoring is built to close.
Why the Cuff Was Never the Endgame
Cuff-based readings are accurate but episodic. A patient who’s calm and seated in a doctor’s office can register normal pressure while spiking every night during sleep apnea episodes — a pattern no clinic visit will ever catch.
Ambulatory monitoring, tracking pressure over 24 hours instead of a single sitting, has stronger predictive value for cardiovascular events than one-off clinic numbers. That’s not new information. What’s new is the hardware finally making it practical for daily use instead of a research protocol.
The Sensor Fusion Doing the Actual Work
Cuffless devices don’t inflate anything. They read pulse wave velocity — how fast a pressure wave travels through the arteries — and correlate it against blood pressure using models trained on paired sensor and cuff data.
Pulse transit time is the workhorse metric here. A 2019 review in Biomedical Engineering Letters lays out why the technique holds promise for measuring blood pressure unobtrusively, potentially improving diagnosis and monitoring of hypertension and related cardiovascular disease. Combined with photoplethysmography — the same optical technique already running in most smartwatches — it gives a workable proxy for systolic and diastolic pressure without a single squeeze of the arm.
Three mechanisms carry the load:
- Pulse wave analysis — sensors track how pulse shape and timing shift with arterial pressure
- Continuous streaming — data flows constantly instead of in single snapshots
- Personalized calibration — models adjust to individual physiology as more readings accumulate
None of this replaces a cardiologist. It replaces the gap between visits.
Where This Is Actually Deployed
Some of the clearest examples of AI-driven vital-sign sensing are moving from consumer wearables into clinical and field settings. Medical device maker Sempulse builds its cuffless approach around photoplethysmography placed at the ear, a location chosen because it holds up during motion and long-term wear — a harder problem than a stationary wrist reading. The company’s Halo device pairs pulse transit time with pulse arrival time to estimate blood pressure while a person is actively moving, extending measurement into sports and field conditions where cuff-based devices are impractical.
Consumer hardware is chasing the same problem from a different angle. Samsung’s Galaxy Ring 2 added continuous blood pressure tracking alongside sleep and menstrual cycle prediction — one thread in the broader shift of how AI made wearables essential in 2026, where sensor fusion is turning passive jewelry into ambient medical monitoring.
What FDA Clearance Actually Signals
Not every device making blood pressure claims has earned the right to. FDA clearance means a manufacturer demonstrated substantial equivalence to an existing validated device, or supplied direct clinical evidence of accuracy. It’s the line between a wellness gadget and a device a physician can act on.
That distinction matters more as the market gets crowded. Interest in the space has outpaced regulatory clarity — validation protocols built for cuff-based devices weren’t designed for continuous, AI-derived estimates, and agencies are still adapting them.
What’s Still Broken
Accuracy holds up reasonably well in controlled trials and degrades under real-world conditions: movement artifacts, temperature swings, and skin tone variation all introduce noise that training data doesn’t always account for.
| Challenge | Why It Persists |
|---|---|
| Calibration drift | Most devices still need periodic cuff-based recalibration |
| Cost | Sensor fusion and on-device processing add expense over basic cuffs |
| Generalization | Models trained on narrow populations underperform on others |
| Regulatory lag | Validation frameworks built for cuffs don’t map cleanly onto continuous AI estimates |
Where It’s Headed
Multi-modal sensing — stacking PPG with ECG and bioimpedance — is the near-term fix for calibration drift. Further out, the more interesting shift is predictive rather than descriptive: systems that flag a hypertensive episode building before it happens, rather than reporting one after the fact.
Pair that with electronic health record integration and the device stops being a gadget and starts being a data feed a physician actually uses to adjust medication timing.
Hypertension management has run on a snapshot for over a century. The technology closing that gap isn’t dramatic — it’s a sensor at the wrist or the ear, quietly doing the math a cuff never could.
Related: This AI Toilet Is Turning Your Bathroom Into a Health Sensor
