Google Research Tests Smartwatch Hypertension Screening

Google Research researchers Paolo Di Achille and Ming-Zher Poh reported a smartwatch-based AI system for passively detecting signs of hypertension in a study published in Nature on November 5, 2026. Called AI-PPG-ACC-HTN, it uses consumer smartwatch sensors without cuff calibration and was validated against 24-hour ambulatory blood pressure monitoring (ABPM). The findings informed the blood pressure trends feature in the Google Health app. The researchers say it is exploratory feasibility research, not a replacement for clinical cuffs or a definitive diagnosis.

The system looks for a persistent hypertension pattern across multiple days rather than estimating exact systolic and diastolic readings at a particular moment. It combines photoplethysmography (PPG), which measures wrist blood-volume changes using reflected light, with accelerometry (ACC), which records movement and subtle wrist vibrations caused by blood leaving the heart. The sensor waveforms are converted into embeddings (compact numerical representations), combined across days, and used with a person’s height to estimate hypertension probability.

Researchers first trained a 1-D deep convolutional neural network with self-supervised contrastive learning on 492,347 hours of unlabeled sensor data from 2,109 consenting participants. It learned to place overlapping 15-second PPG and ACC waveforms from the same person close together in a latent space, identifying physiological features without manual feature engineering or medical labels. In the second stage, the system averaged each person’s waveform features across multiple days to reduce temporary noise, combined them with height as a proxy for arterial path length, and applied regularized logistic regression.

A prospective clinical study at five US sites tested the system with 196 consenting participants who had no known cardiovascular disease and were not taking blood pressure medication. For initial screening, the smartwatch had 65.8% sensitivity, 90.0% specificity and 80.6% positive predictive value (PPV); office screening had 55.3% sensitivity and 77.8% PPV. For confirmatory testing among 48 participants with elevated initial office readings, the smartwatch achieved 78.4% sensitivity, 90.9% specificity and 96.7% PPV. Return office measurements recorded 67.6% sensitivity, 63.6% specificity and 86.2% PPV, while multi-day home monitoring recorded 89.2%, 81.8% and 94.3%, respectively.

The researchers observed that waveforms associated with higher predicted risk had slightly narrower pulse widths, slower systolic upstrokes and lower d/a ratios. Accelerometer signals also showed deeper K waves and wider IJK complexes in those users. The authors say the exploratory work suggests a possible role for passive opportunistic screening that prompts people at risk to seek formal clinical assessment; it does not establish a standalone diagnosis or an existing or planned commercial product capability. The collaboration included Google Research, Beth Israel Deaconess Medical Center and Harvard Medical School.

Google Research researchers Paolo Di Achille and Ming-Zher Poh reported a smartwatch-based AI system for passively detecting signs of hypertension in a study published in Nature on November 5, 2026. Called AI-PPG-ACC-HTN, it uses consumer smartwatch sensors without cuff calibration and was validated against 24-hour ambulatory blood pressure monitoring (ABPM). The findings informed the blood pressure trends feature in the Google Health app. The researchers say it is exploratory feasibility research, not a replacement for clinical cuffs or a definitive diagnosis.

The system looks for a persistent hypertension pattern across multiple days rather than estimating exact systolic and diastolic readings at a particular moment. It combines photoplethysmography (PPG), which measures wrist blood-volume changes using reflected light, with accelerometry (ACC), which records movement and subtle wrist vibrations caused by blood leaving the heart. The sensor waveforms are converted into embeddings (compact numerical representations), combined across days, and used with a person’s height to estimate hypertension probability.

Researchers first trained a 1-D deep convolutional neural network with self-supervised contrastive learning on 492,347 hours of unlabeled sensor data from 2,109 consenting participants. It learned to place overlapping 15-second PPG and ACC waveforms from the same person close together in a latent space, identifying physiological features without manual feature engineering or medical labels. In the second stage, the system averaged each person’s waveform features across multiple days to reduce temporary noise, combined them with height as a proxy for arterial path length, and applied regularized logistic regression.

A prospective clinical study at five US sites tested the system with 196 consenting participants who had no known cardiovascular disease and were not taking blood pressure medication. For initial screening, the smartwatch had 65.8% sensitivity, 90.0% specificity and 80.6% positive predictive value (PPV); office screening had 55.3% sensitivity and 77.8% PPV. For confirmatory testing among 48 participants with elevated initial office readings, the smartwatch achieved 78.4% sensitivity, 90.9% specificity and 96.7% PPV. Return office measurements recorded 67.6% sensitivity, 63.6% specificity and 86.2% PPV, while multi-day home monitoring recorded 89.2%, 81.8% and 94.3%, respectively.

The researchers observed that waveforms associated with higher predicted risk had slightly narrower pulse widths, slower systolic upstrokes and lower d/a ratios. Accelerometer signals also showed deeper K waves and wider IJK complexes in those users. The authors say the exploratory work suggests a possible role for passive opportunistic screening that prompts people at risk to seek formal clinical assessment; it does not establish a standalone diagnosis or an existing or planned commercial product capability. The collaboration included Google Research, Beth Israel Deaconess Medical Center and Harvard Medical School.