Google Research’s Nature paper is refreshingly practical: using the PPG and accelerometer in an ordinary smartwatch, it passively screens for signs of hypertension without cuff calibration. The system is called AI-PPG-ACC-HTN.
In a prospective, multicenter study of 196 people, compared against the gold standard of 24-hour ambulatory blood pressure monitoring, initial screening had a sensitivity of about 65.8%, specificity of about 90%, and positive predictive value of about 80.6%. In comparison, same-day in-office screening had a sensitivity of about 55.3%. In the confirmation setting, specificity was about 90.9% and PPV about 96.7%.
The paper also clearly states its limits: this is an opportunistic screening study, not a replacement for cuff-based diagnosis or a ready-to-use commercial product. Related findings have already been incorporated into the blood pressure trends feature in the Google Health app. As AI moves into healthcare, work that can be benchmarked against the gold standard and is candid about its limits is more credible than empty claims of disruption.
#AI #Google #Health
In a prospective, multicenter study of 196 people, compared against the gold standard of 24-hour ambulatory blood pressure monitoring, initial screening had a sensitivity of about 65.8%, specificity of about 90%, and positive predictive value of about 80.6%. In comparison, same-day in-office screening had a sensitivity of about 55.3%. In the confirmation setting, specificity was about 90.9% and PPV about 96.7%.
The paper also clearly states its limits: this is an opportunistic screening study, not a replacement for cuff-based diagnosis or a ready-to-use commercial product. Related findings have already been incorporated into the blood pressure trends feature in the Google Health app. As AI moves into healthcare, work that can be benchmarked against the gold standard and is candid about its limits is more credible than empty claims of disruption.
#AI #Google #Health
