One-stop strabismus digital diagnosis via AI-integrated skin-like and wearable "Eyelectronics".

Abstract

Strabismus, affecting ~4% of children, impairs vision and psychosocial health. However, clinical diagnosis requires multiple instruments and stepwise examinations of ocular alignment, extraocular muscle function, and deviation angle. It is limited by low diagnostic objectivity, poor pediatric compliance, and high cost. Here, we propose a strategy for one-stop strabismus digital diagnosis via artificial intelligence (AI)-integrated, skin-like, and wearable "Eyelectronics." The ultralightweight, imperceptible eye-wearable system features an ultrathin (~60 micrometers in thickness), breathable, and multidirectional (0°/45°/90°) strain-sensing array conformally adapted to the sensitive eyelid. It enables wireless, mild-restricted measurement of eyelid deformation during eye movements. Through biomechanical modeling validated by ocular magnetic resonance imaging simulations, we establish a prior correlation between eyelid deformation and eye movements. The Eyelectronics, powered by our physiology knowledge-driven end-to-end AI algorithm, achieves simultaneous measurement of strabismus angle and identification of paretic muscle. It delivers a 96.6% four-direction classification accuracy and a 1.2° measurement accuracy in ocular motility examinations. Clinical benchmarking against the clinical standard (Hess screen test) confirmed diagnostic agreement (intraclass correlation = 0.978). This system bridges quantitative biomechanical sensing with digital diagnosis, promoting a paradigm for future strabismus treatment.