技術：高血圧を検知できるスマートウォッチ

Smartwatches may be able to detect hypertension (high blood pressure) with the help of an artificial intelligence (AI) system that interprets sensor readings, a paper in Nature reports. Tests of the system, trialled using a consumer smartwatch, show that it can detect hypertension at similar sensitivities to traditional office-based blood pressure screening methods.

Hypertension is a major risk factor for cardiovascular disease, but over half of affected adults are unaware of their condition. Wearable technology has the potential to detect signals of hypertension, although the sensors are typically not good enough for the direct measurement of blood pressure and readings need to be calibrated against traditional cuff blood pressure measurements. Recent research has suggested that deep learning could help, but previous assessments have not used gold-standard blood pressure monitoring methods as reference points and have not compared measurements with established hypertension screening methods.

Ming-Zher Poh and colleagues show that applying an AI algorithm to signals recorded using a Google smartwatch can detect hypertension with accuracy levels comparable to traditional office-based cuffed blood pressure devices. The smartwatch performs photoplethysmography (an optical technique that measures changes in blood volume in microvascular tissue beneath the skin) and uses an accelerometer to measure wrist motion. Combining these measurements with participant height and running them through an algorithm produces an estimated probability of hypertension without cuff calibration. This estimate requires multiple readings.

The system was trained using nearly 500,000 hours of smartwatch data collected from 2,109 people, with additional testing and training in several cohorts. The approach was then validated in 196 participants by comparing the smartwatch-estimated presence of hypertension and standard office blood pressure measurements, evaluated against reference 24-hour ambulatory blood pressure monitoring, the gold-standard clinical monitoring method. Detection of hypertension was comparable between the smartwatch system and standard office measurements. The smartwatch correctly identified slightly more cases of hypertension than office-based cuffed monitoring, with sensitivities of 65.8% and 55.3%, respectively, but the differences were not statistically significant. Both methods had equal specificities (successfully avoiding false positives 90% of the time). The authors note that their system would not replace traditional cuff-based blood pressure monitoring but could act as a tool to identify individuals with signs of hypertension, which could facilitate earlier diagnosis and intervention and may reduce the occurrence of cardiovascular events.

• Article
• Published: 05 November 2026

Di Achille, P., Cai, L., Wu, J. et al. Opportunistically Detecting Signs of Hypertension on a Consumer Smartwatch. Nature (2026). https://doi.org/10.1038/s41586-026-11336-3