AI Detects Rare Hormone Disorder from Hand Images

Researchers trained a neural network on over 11,000 images of the back of the hand and clenched fists from 725 participants to diagnose acromegaly, achieving a positive predictive value of 0.88 and a negative predictive value of 0.93, and even outperforming endocrinology specialists in the same task. The approach offers privacy advantages over face-based methods and could speed screening, but it won’t replace clinicians; further validation in larger, diverse populations and exploration of other hand-related conditions are planned.