top of page
Search

Hands off the eyes! Eye rubbing, Keratoconus and Machine Learning

Title of Article: Preventing Keratoconus through Eye Rubbing Activity Detection: A Machine Learning Approach.

 

What are the key takeaway points from this article?

Keratoconus (KC) is a vision-impairing disease of the cornea that is demonstrating an increasing prevalence.  Eye rubbing is a known modifiable risk factor, with studies indicating that eye rubbing is associated with KC in 45-83% of patients. Consequently, finding ways to mitigate this behavior could be crucial in preventing both the development and severity of KC. The authors of the present study sought to evaluate the accuracy of a machine-learning system for the detection and prevention of eye rubbing. The authors also conducted a literature review on AI techniques for detecting KC diagnosis and face-touching activity.  

 

An inertial measurement unit (IMU) served as the sensing device, collecting accelerometer and gyroscope data via a wrist-mounted sensor.  The IMU data tracked hand movements while the machine-learning algorithms determined whether these movements involved eye rubbing in real time. Two participants were involved in the data collection: a healthy 52-year-old male and a 15-year-old who was previously treated for KC, who developed the condition after increased eye rubbing. Each participant underwent a 10-minute acquisition test, repeated 10 times, to assess eye rubbing activity in an uncontrolled environment. 

 

Four classification methods were used: support vector machines, decision trees, random forest, and XGBoost, which all showed high-quality accuracy results.  The authors concluded that such a device holds promise for detecting eye rubbing through wearable technologies like smartwatches.  However, they acknowledged major limitations, including the small sample size, limited monitored activities, and the inability to differentiate between similar activities such as nose or head scratching.

  

Publication Date: February 2023

 

Reference: Nokas, G., & Kotsilieris, T. (2023). Preventing Keratoconus through Eye Rubbing Activity Detection: A Machine Learning Approach. Electronics [Basel], 12(4), NA.

 

Summary By: Natalie Mezey

17 views

Recent Posts

See All

Can AI ‘see’ the future and predict long-term VA?

Article Title: Machine Learning Models for Predicting Long-Term Visual Acuity in Highly Myopic Eyes Key Takeaway Points: High myopia, being one of the top 3 causes of irreversible blindness in Asia, E

Comments


bottom of page