An artificial intelligence algorithm can now detect high blood pressure and diabetes from a simple 5-second facial video with remarkable accuracy, according to new research presented at ESC Congress 2026. The breakthrough could transform how millions of undiagnosed patients are identified, moving screening out of clinics and into everyday environments like pharmacies, workplaces, or even homes.

How Does AI Read Your Face to Detect Disease?

Researchers at the University of Tokyo and Institute of Science Tokyo developed a machine-learning algorithm that analyzes facial and palm videos using a spectroscopic camera. The system doesn't just look at your appearance; it extracts detailed physiological data from your skin and blood vessels.

The algorithm measures three key indicators:

• Pulse-wave dynamics: The algorithm detects how stiff your arteries are by analyzing the subtle patterns of blood flow in your face and palms.
• Skin blood-flow patterns: The system tracks how blood moves through tiny vessels in your skin, which can reveal metabolic dysfunction associated with diabetes.
• Spectral characteristics of skin coloring: The algorithm analyzes the color and light absorption properties of your skin, which reflect underlying physiological conditions.

The study involved 215 participants, including both people with diagnosed conditions and healthy volunteers. Each person underwent a brief, high-speed video recording while the algorithm extracted and analyzed the physiological data.

What Are the Accuracy Numbers?

The results are striking. The algorithm detected hypertension with 95% accuracy from a 30-second video and maintained 90.3% accuracy using just a 5-second recording. For diabetes detection, the algorithm achieved 88.2% accuracy from 30 seconds and 81.2% accuracy from 5 seconds.

The sensitivity metrics are equally impressive. When screening for normal blood pressure, the algorithm correctly identified it 100% of the time. For detecting actual hypertension, it caught 89.2% of cases. These numbers suggest the tool could reliably identify people who need further medical evaluation.

The algorithm could also estimate blood pressure without a traditional cuff. The mean error for systolic blood pressure was within acceptable clinical limits, though researchers noted that future work will focus on reducing variability through larger, more diverse datasets.

Why Does This Matter for Global Health?

The scale of undiagnosed disease is enormous. Approximately 1.4 billion adults aged 30 to 79 years have hypertension worldwide, and 589 million people live with diabetes, yet many cases remain undetected. Current screening methods rely on dedicated clinic visits or wearable devices, which limits how many people can be reached.

Hypertension and diabetes are among the leading modifiable risk factors for cardiovascular disease. Early detection enables treatment and lifestyle changes to begin sooner, potentially preventing heart attacks, strokes, and other serious complications.

"Our machine-learning algorithm accurately detected hypertension and diabetes from facial spectroscopic video recordings as short as 5 seconds. We intend to validate these findings in larger cohorts across more diverse populations to support real-world application. If validated, this contactless approach could allow people to be screened in everyday settings without cuffs, blood sampling or a dedicated clinic visit, helping to identify at-risk individuals who would otherwise remain undiagnosed and therefore untreated," explained Ryoko Uchida.

Ryoko Uchida, Researcher at University of Tokyo and Institute of Science Tokyo

How to Deploy This Technology in Real-World Settings

Because the approach is quick, easy, and contactless, it could be deployed far beyond hospitals. Here are the key deployment scenarios researchers envision:

• Pharmacy screening stations: Patients could receive a quick facial scan while picking up medications, identifying undiagnosed conditions before they worsen.
• Workplace health programs: Employers could offer contactless screening during annual health fairs, reaching employees who might not visit clinics regularly.
• Community health clinics: Mobile health units and resource-limited settings could use the technology without expensive equipment, making screening accessible globally.
• Specialized imaging equipment: The current technology requires a spectroscopic camera, a specialized device that captures high-speed video and analyzes light absorption through skin.

Associate Professor Nico Bruining, Programme Co-Chair of the ESC Digital and AI Summit, emphasized the potential impact: "It is remarkable that AI-supported technologies are enabling the development of such powerful tools for early disease prevention. Because this approach is quick, easy and contactless, it could be used in many settings beyond hospitals, giving it the potential to reach far more people than traditional screening methods".

Nico Bruining, Programme Co-Chair of the ESC Digital and AI Summit

What Happens Next?

The research team plans to validate these findings in larger cohorts across more diverse populations before real-world deployment. Current work focuses on reducing variability in blood pressure estimates through expanded datasets and feature optimization.

The latest advances in AI and cardiovascular care will be discussed at the ESC Digital and AI Summit in Basel, Switzerland, on November 12 and 13, 2026. The event will bring together clinicians, researchers, innovators, and industry experts who are shaping the future of cardiovascular health.

This breakthrough represents a shift in how medicine thinks about screening. Rather than waiting for patients to schedule appointments, AI-powered tools could identify at-risk individuals in their daily environments, democratizing access to early detection and potentially saving millions of lives.