1. Systematic review of AI in periodontal disease detection.
2. Accuracy for classification varied from 48% to 100%.
3. Convolutional neural networks were predominant.
4. External validation showed variable results.
5. Major limitations included inconsistent reporting and standards.
6. Recommendations for future studies include better designs and quality imaging.
7. Data quality concerns often arose with low-resolution images.
8. Potential exists, but clinical applicability is limited.

Source: International Dental Journal