Google Research introduces AI system to detect hypertension using smartwatches

According to Google Research:

Background

High blood pressure is a widespread health issue that often goes unnoticed because it rarely shows obvious symptoms. Traditional wearable blood pressure trackers usually require regular arm cuff calibration to measure relative changes, which limits their usefulness as independent screening tools.

Technology

Google Research scientists created a multimodal AI system named AI-PPG-ACC-HTN. It analyzes sensor data from consumer smartwatches over multiple days, using a deep convolutional neural network trained through self-supervised contrastive learning. By combining these passive physiological signatures with the user's height, the system calculates a continuous hypertension probability score.

Impact

Validated against 24-hour ambulatory blood pressure monitoring in a clinical study, the smartwatch system showed high specificity and predictive value. While not meant to replace standard medical devices, this technology could offer a scalable screening tool to catch the condition early and encourage people to seek professional medical care.