Related Experiment Video
Updated: Jul 4, 2025
05:36
Subjective Refraction Test Using a Smartphone for Vision Screening
Published on: October 18, 2024
702
Implementing and evaluating a fully functional AI-enabled model for chronic eye disease screening in a real clinical
Christos Skevas1, Nicolás Pérez de Olaguer2, Albert Lleó2
1Department of Ophthalmology, University Medical Center Hamburg - Eppendorf, Martinistr. 52, 20249, Hamburg, Germany.
BMC Ophthalmology
|February 2, 2024
Summary
Artificial intelligence (AI) effectively screens for diabetic retinopathy, age-related macular degeneration, and glaucoma in real-world clinical settings. The integrated system demonstrated high accuracy and received positive user feedback, highlighting its potential for improving eye care accessibility.
Area of Science:
- Ophthalmology
- Medical Imaging
- Artificial Intelligence
Background:
- Artificial intelligence (AI) offers potential for cost-effective and accessible eye disease screening.
- AI-based diabetic retinopathy (DR) screening programs have recently gained regulatory approval in several countries.
Purpose of the Study:
- To evaluate the performance, feasibility, and user experience of an integrated AI solution for chronic eye disease screening.
- To assess AI grading accuracy for DR, age-related macular degeneration (AMD), and glaucoma against specialist grading.
- To gather user feedback on a seamless hardware and software system for eye disease screening.
Main Methods:
- A real-world clinical study in Germany evaluated an integrated AI system for DR, AMD, and glaucoma screening.
- The system included AI grading, specialist auditing, and automated patient referral decisions.
- Performance was assessed by comparing AI and specialist grading, analyzing referral accuracy, and collecting user feedback via questionnaires.
Main Results:
- The AI system achieved high sensitivity and specificity for DR, AMD, and glaucoma screening.
- Only 17 out of 231 patients were falsely referred, indicating high overall accuracy.
- Clinical staff reported a seamless workflow and high satisfaction, though improvements for AMD and glaucoma models were suggested.
Conclusions:
- AI-based screening for DR, AMD, and glaucoma is effective in real-world clinical environments.
- The integrated platform received positive usability feedback from screening staff and auditors.
- The auditing function supports efficient expert second opinions, enhancing remote screening potential.
