Artificial Intelligence Can Diagnose Diseases Through Facial, Eye, and Hand Details!
Two separate scientific studies, one conducted in Japan and the other in Scotland, have revealed the potential for artificial intelligence systems to detect hidden health risks by analyzing ordinary images of the face, palm, and eyes. The science news website The Brighter Side of News reported that a Japanese research team managed to detect high blood pressure and diabetes from facial videos lasting no more than five seconds, while a Scottish project is training algorithms on approximately one million images of the retina to estimate the risk of dementia.
Researchers from the University of Tokyo and the Tokyo Institute of Technology explained that they tested the hypothesis that subtle changes visible in facial and palm videos could reveal cardiovascular and metabolic diseases. This was part of a single-center pilot study involving 215 diagnosed patients and healthy volunteers. Each participant had a short, high-speed video recorded using a spectral camera, and they also underwent traditional blood pressure and diabetes tests.
The system analyzed pulse wave patterns, blood flow in the skin, and spectral characteristics of skin tone. The results on high blood pressure detection will be presented at the European Society of Cardiology Congress in 2026. Given the remarkable accuracy achieved by these indicators, the key findings can be summarized as follows:
• High blood pressure was detected with 95% accuracy based on 30-second videos of the face and palm, with a sensitivity of 89.2% for hypertension cases and 100% for normal blood pressure cases.
• An accuracy of 90.3% was achieved even with recordings lasting no more than five seconds.
• Diabetes was detected with 88.2% accuracy from a 30-second facial video, and with 81.2% accuracy from a five-second recording.
The site quoted researcher Ryoko Uchida, who presented the results, stating that the team “aimed to develop an AI algorithm that enables contactless screening in everyday environments to detect common diseases early and on a large scale.” She added that the system also succeeded in estimating systolic blood pressure using facial video alone, with a mean absolute error of 8.6% and a mean bias of -2.6 mmHg, which falls within the acceptable limit set by the American Association for the Advancement of Medical Instrumentation of five mmHg. However, the variability in results reached twelve mmHg, exceeding the accepted limit of eight mmHg, meaning the system is still far from replacing traditional blood pressure measurement, especially given global estimates indicating that 1.4 billion adults suffer from high blood pressure and 589 million from diabetes.
In Scotland, the NeurEYE project team, a collaboration between the University of Edinburgh and Glasgow Caledonian University, is pursuing a different approach based on routine eye examinations. Researchers collected approximately one million retinal images from optometrists across Scotland and linked them with anonymized information including demographic data, medical history, and previous conditions, aiming to train algorithms on patterns associated with dementia risk.
Bhagwan Deleon, Professor of Clinical Ophthalmology at the University of Edinburgh and co-supervisor of the project, stated that “the eye can tell us more than we thought,” adding that the blood vessels and neural pathways in the retina and brain “are closely linked.” This gives the retina, unlike the brain, the potential for direct examination using equipment already available in optometry clinics.
Miguel Bernabeu, a professor of computational medicine at the University of Edinburgh and another co-supervisor of the project, emphasized that developing fair and unbiased algorithms requires training them on data that represent the entire at-risk population. The project is funded by the NEURii consortium, which includes Eisai, Gates Ventures, and the University of Edinburgh, while approval for the use of health data was obtained from the Public Benefit and Privacy Panel for Health and Social Care, part of NHS Scotland.
The two projects reflect a growing trend toward integrating artificial intelligence into routine health screening before symptoms appear, although the researchers themselves noted that both systems still require broader and more diverse validation studies before they can be adopted as reliable clinical tools.
