Related Experiment Video
Updated: Jun 29, 2025
07:08
Using Optical Coherence Tomography and Optokinetic Response As Structural and Functional Visual System Readouts in Mice and Rats
Published on: January 10, 2019
10.0K
Transparency in Artificial Intelligence Reporting in Ophthalmology-A Scoping Review
Dinah Chen1, Alexi Geevarghese1, Samuel Lee2
1Department of Ophthalmology, NYU Langone Health, New York, New York.
Ophthalmology Science
|April 9, 2024
Summary
Reporting on artificial intelligence (AI) in ophthalmology shows significant variability in model development and validation details. Increased transparency is crucial for critical appraisal and equitable clinical implementation of AI tools.
Area of Science:
- Ophthalmology
- Medical Artificial Intelligence
- Health Informatics
Background:
- Artificial intelligence (AI) is increasingly used in ophthalmology for disease classification.
- Standardized reporting of AI model development and validation is lacking.
- Transparency in AI reporting is essential for clinical implementation and bias mitigation.
Purpose of the Study:
- To conduct a scoping review of AI reporting in ophthalmology literature.
- To characterize the transparency in reporting of AI model development and prospective validation.
- To identify gaps in reporting guidelines for AI in ophthalmology.
Main Methods:
- Scoping review of studies published up to January 2022.
- Searches conducted in PubMed, Embase, Web of Science, and CINAHL.
- Evaluation of studies for reporting parameters from CONSORT-AI, MI-CLAIM, and model cards, focusing on dataset details, demographics, and prospective validation outcomes.
Main Results:
- Thirty-seven prospective validation studies involving 27 unique AI models were reviewed.
- Reporting of training dataset annotation and data distribution was variable.
- Demographic reporting (age, gender, race, ethnicity) was inconsistent, particularly for training data, hindering assessment of fairness and generalizability.
Conclusions:
- Current reporting of AI model development and validation in ophthalmology lacks transparency.
- There is a critical need for standardized reporting to ensure appropriate and equitable clinical use of AI tools.
- Enhanced transparency will facilitate critical appraisal and minimize bias in AI implementation.
