This scoping review looked at how artificial intelligence tools create automated dental reports. Researchers examined 1,265 records from seven different studies that compared AI-generated text to reports written by humans. The studies included various NLP-based models, including GPT variants and fine-tuned large language models.
The analysis found that AI models showed high accuracy for common findings. Readability scores were comparable to those of human-authored reports. When the AI output was adapted to be simpler, patients rated the clarity as improved. Other measures like response latency and output length were also evaluated.
The review noted several limitations, including differences in report types, datasets, languages, and evaluation metrics used across the studies. No safety concerns or adverse events were reported in the included studies. However, the researchers state that standardized evaluation frameworks and larger multilingual datasets are needed before these tools can be used routinely in clinical practice. Readers should understand that this is an early review and not yet ready for standard medical use.