Monitor this situation.
Unsubscribe anytime.
[SITUATION] · [ACTIVE] · [HEALTH]
3 clusters · 8 sources · 7 days · First seen · Last updated
AI development for smartwatch health monitoring
Overview
Researchers are developing artificial intelligence systems to enhance the medical utility of consumer smartwatches. In the United States, Google researchers introduced an AI system, AI-PPG-ACC-HTN, designed to detect hypertension using photoplethysmography and accelerometry sensors without requiring traditional cuff calibration. In a study of 196 participants, the system demonstrated a sensitivity of 65.8 percent and a specificity of 90 percent for primary screening.
Parallel research presented at the American Heart Association’s Scientific Sessions suggests that AI analyzing ECG data from smartwatches may identify left ventricular systolic dysfunction (LVSD). In the WATCH-LV study, Medical AI and Incheon Sejong Hospital presented findings using 30-second ECG recordings to detect LVSD, defined by an ejection fraction of 40 percent or less.
In a study of over 2,200 adults in South Korea, the AI model showed high accuracy, achieving an area under the ROC curve of 0.907 for the Apple Watch and 0.895 for the Galaxy Watch. While researchers previously noted that 9 percent of participants received different results depending on the smartwatch brand, the WATCH-LV study indicated comparable overall accuracy across devices, despite differing calibration thresholds.
To mitigate the clinical burden of false positives, researchers found that requiring positive results on at least two of three consecutive days was effective. While this decision rule significantly reduced false alarms in a group of 108 outpatient patients, it also resulted in missing certain cases. Further randomized studies are planned to confirm the model's effectiveness for daily use.
Entities
American Heart Association · Apple Watch · Incheon Sejong Hospital · Paolo Di Achille · Nature
Timeline
-
[HEALTH] 2 sourcesMedical AI and Incheon Sejong Hospital present smartwatch heart failure screening results
Medical AI and Incheon Sejong Hospital have developed an AI model that uses smartwatch ECG data to detect heart failure, significantly reducing false alarms through multi-day screening.
-
[HEALTH] 3 sourcesAI analysis of smartwatch ECG data may detect heart dysfunction
AI software analyzing smartwatch ECG data may help identify left ventricular systolic dysfunction, though results can vary between device brands and recording methods.
-
[TECHNOLOGY] 3 sourcesGoogle Research develops AI to detect hypertension via smartwatch
Google Research has unveiled an AI system that detects hypertension using smartwatch pulse and motion sensors without requiring traditional cuff calibration.
Sources
ad-hoc-news.de · cryptoslate.com · dev.to · it-boltwise.de · knowridge.com · newmediawire.com · newsroom.heart.org · tbreak.com
This summary has been updated 1 time: see revision history