Objective:

To develop a hybrid deep learning framework for automated cataract detection that balances diagnostic accuracy with computational efficiency.

Key Findings:

• Achieved 99.10% accuracy on the Eye Cataract Kaggle dataset.
• Obtained 99% precision, 99.21% recall, and a 99.10% F1-score.
• Outperformed traditional machine learning methods (KNN, LR, DT) and deep learning architectures (VGG19, ResNet50, DenseNet201, InceptionV3, EfficientNetB0) by 3-5% in accuracy.

Interpretation:

The hybrid AI approach enhances cataract detection accuracy while maintaining computational efficiency, making it suitable for resource-limited settings.

Limitations:

• Requires external validation across diverse populations and imaging systems before clinical deployment.

Conclusion:

If validated, this hybrid AI system could significantly improve cataract screening and extend its application to grading severity and monitoring progression.