An innovative tool using artificial intelligence is making waves in the medical field, particularly for the potential it holds in screening for high blood pressure and diabetes through non-invasive means. This breakthrough technology, which utilizes high-speed video to monitor blood flow changes in the skin of the face and hands, offers a novel alternative that eliminates the need for blood pressure cuffs, blood tests, or wearable devices, according to new research findings.

In a study conducted in a hospital in Japan, researchers discovered that the system was capable of accurately identifying a significant number of cases of hypertension and diabetes. These preliminary findings were shared at the American Heart Association’s Scientific Sessions in Chicago, but they are pending further validation through peer-reviewed publication. Lead researcher Ryoko Uchida, a project researcher in the advanced cardiology department at the University of Tokyo, noted that this contact-free method could empower individuals to monitor their health from home, without undergoing traditional testing procedures.

The system relies on high-speed video technology operating at 150 frames per second to analyze pulse wave and blood flow characteristics. In this study, 215 participants—comprised of diagnosed patients and healthy individuals—underwent video recording while seated in a controlled setting. Both five- and thirty-second clips were captured for data extraction. Advanced machine learning algorithms were then applied to assess variations in pulse wave arrival across various regions on the face and hands, with a specific focus on 22 facial areas and eight hand zones.

Simultaneously, conventional blood pressure readings were taken using a continuous monitor, while blood glucose levels were evaluated through an A1C test, which indicates average blood sugar levels over the prior two to three months. When the video-based measurements were compared against the blood pressure monitor’s readings, the AI system showed a remarkable 94% accuracy rate in detecting stage 1 hypertension among a subgroup of 77 individuals. This condition is characterized by systolic numbers ranging between 130-139 mmHg or diastolic numbers between 80-89 mmHg, per guidelines from the American Heart Association and the American College of Cardiology.

For diabetes detection, the video and algorithm proved to be 75% accurate in identifying individuals with A1C scores at or above 6.5, which is the threshold for a diabetes diagnosis. This subgroup included those diagnosed with Type 1 and Type 2 diabetes alongside other individuals with high A1C levels. While these early findings are promising, Uchida also highlighted the necessity for additional research to verify the tool’s applicability across diverse populations and environments.

Concerns have been raised regarding the practicality of this technology. Dr. Geoff Rubin, a professor and chair of medical imaging at the University of Arizona College of Medicine, pointed out that adherence to protocols for the AI testing may prove to be as resource-intensive as more conventional methods. “If individuals are already scheduled to remain still in a controlled environment, why wouldn’t they just opt for a blood pressure cuff or a simple blood draw, which would provide definitive results in a similar timeframe?” he pondered.

Rubin, while expressing skepticism, remains interested in the innovation’s possibilities. He acknowledged that while remote sensing and wearable technologies hold transformative potential for health care, obstacles such as user engagement need to be addressed. Despite the intriguing insights that this non-contact approach may deliver in the future, critical questions regarding its practicality and effectiveness remain unanswered.