Innovative AI-powered app aims to prevent eye rubbing in keratoconus patients
A new artificial intelligence (AI)-powered smartwatch application can objectively detect and quantify eye rubbing, a behavior that contributes to the progression of conditions like keratoconus and corneal ectasia after refractive surgery, according to a study.
The application, deployed on a Samsung Galaxy Watch 4, leverages motion data from sensors, including a gyroscope, accelerometer, and linear acceleration, to identify eye-rubbing behavior. Researchers trained the AI model using advanced deep-learning algorithms—long short-term memory and gated recurrent unit—alongside machine learning methods.
The model demonstrated impressive accuracy, achieving 94%. It effectively recognized and counted eye-rubbing episodes, with the gated recurrent unit model and XGBoost algorithm performing particularly well.
This innovative tool detects eye rubbing and provides real-time alerts to help patients reduce the habit.
Reference
Drira I, Louja A, Sliman L, et al. Eye-Rubbing Detection Tool Using Artificial Intelligence on a Smartwatch in the Management of Keratoconus. Transl Vis Sci Technol. 2024;13(12):16. doi: 10.1167/tvst.13.12.16. PMID: 39666356.
