UCSF and Cardiogram study: Apple Watch can detect hypertension with 82% accuracy, sleep apnea with 90% accuracy via Cardiogram's deep neural network DeepHeart
A new study out from health startup Cardiogram and the University of California San Francisco (UCSF) suggests wearables like the Apple Watch …
Context & Ripple Effects
This extends the same research line as the earlier UCSF-Cardiogram atrial fibrillation study from May 2017, which showed the Apple Watch paired with the DeepHeart neural network detecting afib at 97% accuracy. The new results move beyond arrhythmia into two chronic conditions — hypertension and sleep apnea — detected purely in software on hardware consumers already own.
The timing matters: within weeks of this study, Apple announced its heart research partnership with Stanford and was reported to be working with telemedicine firm American Well on condition-detection tests, suggesting the company was already building the clinical-validation apparatus these findings point toward.
First-order effects
- Cardiogram's DeepHeart app can now screen for hypertension (82% accuracy) and sleep apnea (90%) on existing Apple Watches, giving the startup clinically-backed claims that differentiate it from generic fitness apps.
Second-order effects
- Apple gains evidence that its heart-rate sensor supports medical-grade screening without new hardware, strengthening the case for the Stanford abnormal-rhythm study it launched later that month and for partner work with players like American Well.
Third-order effects
- If academic studies keep validating software-only detection on consumer wearables, the smartwatch category shifts from fitness accessory to first-line screening platform, with startups like Cardiogram supplying the models and health systems supplying validation.
The trend: Consumer wearables are being repositioned as diagnostic screening tools through academic-AI partnerships, with each validated condition expanding the watch's clinical scope without new sensors.