Detecting signs of hypertension using wearable sensors (research.google)

Google Research has unveiled a groundbreaking AI system, AI-PPG-ACC-HTN, that uses standard smartwatch sensors to passively detect hypertension, a condition affecting over one billion people worldwide. This innovative approach addresses a significant gap in hypertension awareness, particularly among younger adults, who often miss early diagnoses due to the lack of obvious symptoms. By analyzing a vast dataset from over 2,100 participants without the need for traditional cuff calibration, the system demonstrates accuracy comparable to conventional methods in both initial and confirmatory screening scenarios. The AI system employs a two-stage machine learning approach, incorporating a 1-D deep convolutional neural network trained on an extensive unlabeled data set to identify underlying hypertensive states from passive physiological signals. It aggregates data over several days to minimize transient noise effects and combines this information with individual height to assess hypertension probability. With a sensitivity of 65.8% and specificity of 90.0% in initial screenings, AI-PPG-ACC-HTN not only offers competitive performance against standard office measurements but also serves as a scalable, passive tool for early detection. While not designed to replace traditional blood pressure monitoring, this research could dramatically improve hypertension screening and accessibility, potentially leading to timely interventions and better cardiovascular health outcomes.