Researchers at Kobe University in Japan have created an artificial intelligence (AI) system that can detect acromegaly, a rare hormonal disorder, using only photos of the back of a person’s hand and a clenched fist. By avoiding facial images, the system protects patient privacy while maintaining a high level of diagnostic accuracy.
Acromegaly is a rare condition that typically develops during middle age due to excessive production of growth hormone. It causes enlarged hands and feet, noticeable facial changes, and abnormal growth of bones and internal organs. Since the disease progresses gradually over many years, early diagnosis is often challenging.
Delayed Diagnosis of Acromegaly Increases Health Risks
If left untreated, acromegaly can lead to severe health complications and shorten life expectancy by around ten years. According to Kobe University endocrinologist Hidenori Fukuoka, the disease often goes undiagnosed for up to a decade because of its slow progression and rarity. Although researchers have explored AI-based diagnostic tools using photographs, he noted that medical professionals have not yet widely adopted them in clinical settings.
After reviewing earlier research, the team found that most AI systems depend on facial images to detect diseases, raising privacy concerns. To address this issue, they developed their approach using hand photographs instead.
To protect patient privacy, the researchers chose to analyze hand images, as the hands are routinely examined alongside the face during medical evaluations and are often affected by acromegaly-related changes. Yuka Ohmachi, a postgraduate student at Kobe University, explained that focusing on the hands provided a safer alternative for diagnosis.
To further improve data security, the team used only images of the back of the hand and clenched fists, excluding palm photos because unique palm patterns could potentially identify individuals. Researchers trained and evaluated the AI model using more than 11,000 images collected from 725 patients across 15 medical institutions in Japan.
AI System Surpasses Specialists in Acromegaly Detection Accuracy
The findings, published in the Journal of Clinical Endocrinology & Metabolism, showed that the AI system achieved excellent sensitivity and specificity when detecting acromegaly. In comparisons with human experts, the technology performed better than experienced endocrinologists who analyzed the same hand images.
Yuka Ohmachi said she was impressed by the system’s high diagnostic accuracy using only photos of the back of the hand and a closed fist. She highlighted that achieving strong results without relying on facial images makes the technology more practical and suitable for large-scale disease screening.
The researchers plan to expand the AI system to identify other conditions that produce visible changes in the hands, including rheumatoid arthritis, anemia, and digital clubbing. Ohmachi believes this achievement could open new possibilities for the use of AI in healthcare.
The team also stressed that the tool will assist doctors rather than replace them. Medical diagnosis still requires a combination of patient history, laboratory tests, and physical examinations.
According to the researchers, the technology could help support medical professionals, minimize diagnostic mistakes, and enable earlier treatment. Fukuoka added that further development of the system could create a healthcare screening platform to identify possible hand-related disorders, connect patients with specialists, and help reduce gaps in medical access, especially in areas with fewer experts.
Read the original article on: o globo
Read more:Scientists May Have Discovered a Method to Prolong Fertility in Mice