newsroom.heart.org
A new AI-powered tool has shown promise for fast, non-invasive screening of high blood pressure and diabetes, potentially paving the way for at-home, no-contact monitoring. Presented as a preliminary study at the American Heart Association’s Scientific Sessions 2024, this system uses high-speed video analysis of blood flow patterns in facial and hand skin to identify hypertension and diabetes with notable accuracy. This innovation holds particular promise for individuals who may avoid traditional blood pressure measurements and blood tests.
In the study, researchers at the University of Tokyo combined a high-speed camera with an AI algorithm to analyze subtle changes in blood flow across the face and palm. This setup, capturing up to 150 images per second, enables the detection of pulse waves through wavelengths that reveal underlying vascular changes. The technology demonstrated a 94% accuracy in detecting stage 1 hypertension, a 75% accuracy in identifying diabetes, and high accuracy for general blood pressure readings. Such results suggest a strong potential for video imaging as an alternative to conventional diagnostic tools.
This tool could be particularly valuable for individuals seeking easier and faster ways to monitor health metrics outside clinical settings. By simply analyzing blood flow changes, this video-based approach could help detect early signs of hypertension and diabetes, diseases that are key risk factors for cardiovascular complications. Lead researcher Ryoko Uchida highlighted the convenience and potential accessibility of the tool, envisioning future devices integrated into smartphones or mirrors for quick assessments. However, further development and real-world validation are necessary to improve accuracy and applicability.
Despite its promise, the study’s authors caution that the system remains in early development and was tested in a controlled, indoor setting with limited population diversity. Broader testing, particularly with individuals from varied demographic groups and under different lighting conditions, will be crucial for validating its reliability. Additionally, to accurately detect high blood pressure, the algorithm will need modifications to account for irregular heartbeats. Uchida hopes that, with improved accuracy, the tool could eventually gain approval from regulatory bodies like the FDA for widespread use.
This AI-powered approach could revolutionize screening methods for high blood pressure and diabetes by eliminating the need for invasive procedures and expensive equipment. However, as Dr. Eugene Yang of the University of Washington notes, the technology requires validation through rigorous clinical testing before it can be reliably integrated into clinical or consumer health settings. Until then, healthcare providers are encouraged to rely on validated devices for essential health metrics.
This development underscores the growing potential of AI in healthcare, offering a glimpse into a future where patients can monitor critical health markers with ease, potentially leading to earlier diagnosis and intervention.