5 Key Takeaways

• 1

  AI models demonstrate high accuracy in diagnosing inherited retinal diseases, particularly retinitis pigmentosa, with up to 99.9% accuracy.

• 2

  The systematic review included 22 studies, analyzing 5,412 articles, and highlighted the effectiveness of deep learning models using various imaging modalities.

• 3

  Retinitis pigmentosa showed a pooled sensitivity of 94% and specificity of 99%, while Stargardt disease had sensitivity of 96% and specificity of 99%.

• 4

  Widefield imaging significantly improved specificity for retinitis pigmentosa detection, emphasizing the importance of peripheral retinal assessment.

• 5

  The authors concluded that while AI has great potential in retinal disease diagnosis, further research is needed to enhance model robustness across clinical settings.