AI-Powered Hand Photo Analysis Detects Rare Hormone Disorder

A simple photograph of the back of a person’s hand, combined with artificial intelligence, is showing promise in the early detection of acromegaly, a rare and potentially life-threatening hormone disorder. Researchers at Kobe University in Japan have developed an AI model that can diagnose the condition with high accuracy, potentially reducing diagnosis times from years to a matter of moments.

Understanding Acromegaly

Acromegaly occurs when the pituitary gland produces too much growth hormone. This overproduction typically begins in middle age and can lead to a variety of health problems, including enlarged hands and feet, changes in facial features, and abnormal growth of bones and internal organs. If left untreated, acromegaly can significantly reduce life expectancy, by as much as a decade ¹.

How the AI Model Works

The Kobe University team trained their AI model on over 11,000 images of the backs of hands and clenched fists. The dataset included images from 725 participants, roughly half of whom had been diagnosed with acromegaly ¹, ². Crucially, the model does not require images of the palm, addressing privacy concerns and making it more practical for widespread use ⁴.

Accuracy and Performance

The AI model demonstrated a positive predictive value of 0.88, meaning that in 88% of cases where the model predicted acromegaly, the diagnosis was correct. It also achieved a negative predictive value of 0.93, indicating that when the model ruled out acromegaly, it was accurate 93% of the time ¹, ². This level of accuracy surpasses that of endocrinologists when presented with the same images ².

Implications for Healthcare

The development of this AI-powered diagnostic tool has the potential to significantly improve healthcare in several ways:

• Faster Diagnosis: Acromegaly often goes undiagnosed for up to a decade due to its gradual progression and rarity ¹, ⁴. This AI could drastically reduce that timeframe.
• Increased Accessibility: The simplicity of the test – requiring only a photo of the hand – makes it potentially accessible in areas with limited access to specialized medical care.
• Privacy Protection: By focusing on the back of the hand and a clenched fist, the model avoids the privacy concerns associated with facial recognition technology ⁴.

Future Research

Researchers at Kobe University plan to expand their study to include larger and more diverse populations. They are also exploring whether similar AI-based approaches could be used to detect other diseases that manifest in the hands, such as rheumatoid arthritis or anemia ¹.