AI Detects Hormone Disorder From Hand Photos | Kobe University Study
Artificial intelligence just got a new superpower. It can now spot a rare hormone disorder using only a photo of your hand. Researchers at Kobe University in Japan developed this groundbreaking tool. The condition is called acromegaly. It happens when the body makes too much growth hormone. This usually starts in middle age.
What Is Acromegaly?
Acromegaly causes hands and feet to grow larger. Facial features slowly change too. Bones and internal organs grow abnormally. The disease progresses very gradually. Therefore, diagnosis often takes up to ten years. Without treatment, it can shorten life expectancy by about a decade.
Privacy Comes First
Earlier AI systems used facial photos for detection. However, this raised serious privacy concerns. Patients felt uncomfortable sharing images of their faces.The Kobe team found a clever solution. They focused only on hands instead. “We decided to focus on the hands, a body part we routinely examine alongside the face,” explained graduate student Yuka Ohmachi. The researchers took extra privacy steps. They only photographed the back of hands and clenched fists. They deliberately excluded palm images because palms contain unique line patterns that could identify individuals. This approach encouraged widespread participation. A total of 725 patients joined the study. They came from 15 medical centers across Japan. Together, they contributed more than 11,000 images.
Impressive Results
The findings were published February 27 in the Journal of Clinical Endocrinology & Metabolism. The AI model achieved high sensitivity and specificity. Here’s the impressive part. When tested on the same photos, the AI outperformed experienced endocrinologists. “Frankly, I was surprised that the diagnostic accuracy reached such a high level,” Ohmachi admitted. “Achieving this level of performance without facial features makes this approach much more practical for disease screening”.
What’s Next?
The team plans to expand their AI system. It could identify other conditions that change hand appearance. These include rheumatoid arthritis, anemia, and finger clubbing .Doctors won’t rely solely on this tool. They’ll still consider medical history and lab tests. The AI serves as support, not replacement.Study lead Hidenori Fukuoka sees bigger potential. “It could support non-specialist physicians in regional healthcare settings,” he said. “This would help reduce healthcare disparities”. For patients with acromegaly, earlier diagnosis means earlier treatment. And that could add years to their lives.
