Google AI Spots High Blood Pressure Using Watch Sensors

A team of researchers from Google and various medical institutions has demonstrated that ordinary smartwatch signals can be used to flag potential high blood pressure without requiring calibration against an arm cuff. The findings were published in a study in Nature on November 5. While many people might use a google sports band or similar wearables primarily for fitness, this technology points toward passive health screening capabilities during everyday wear, although the initial system missed about 33 percent of hypertension cases.

The system, designated as AI-PPG-ACC-HTN, combines optical pulse-wave measurements with accelerometer signals. According to Google Research, the model aggregates these signals across multiple days and factors in the user's height to calculate a probability of hypertension. It does not provide exact systolic or diastolic blood pressure readings, and no personal cuff measurement is needed for calibration, though cuff-based measurements were used as a clinical reference during testing.

Google researchers utilized ordinary smartwatch signals to flag potential high blood pressure without arm cuff calibration.

The prospective, multicenter study followed 196 participants over seven days of continuous smartwatch monitoring and compared the outcomes against twenty-four-hour ambulatory blood pressure monitoring. Participants had no known history of cardiovascular disease and were not taking blood pressure medication, which limits how broadly the findings apply to individuals already receiving treatment for the condition. The AI successfully identified 65.8 percent of participants with hypertension, while correctly screening 90 percent of those without the condition as negative.

The study supplement notes that the cohort was intentionally risk-enriched. Within this selected group, 80.6 percent of positive smartwatch screens matched hypertension on the reference test. The waveform model was initially pretrained using nearly half a million hours of unlabeled sensor data from over two thousand consenting participants. Researchers from Google Research, Beth Israel Deaconess Medical Center, and Harvard Medical School collaborated on the project.

Why it matters

This development suggests a potential role for smartwatch screening in prompting formal medical assessments and confirmatory diagnoses using established cuff-based methods. By having a wearable device prompt a clinical check, individuals may catch undetected health issues earlier during routine daily activities.

However, because the current system missed a notable portion of hypertension cases and was tested on a specific, risk-enriched group without existing cardiovascular medication, further clinical validation is necessary before widespread deployment.