Schematic illustration of the 2-stage transcription pipeline. Credit score: Journal of Dental Analysis (2025). DOI: 10.1177/00220345251382452
A brand new examine from King’s Faculty London has revealed that synthetic intelligence (AI) automated speech recognition (ASR) instruments may dramatically enhance how dental professionals document affected person info—saving time and lowering administrative burden.
Nevertheless, whereas transcriptional accuracy of those instruments is excessive, they will battle with extra technical language, and their reliability just isn’t at the moment adequate to help unreviewed use.
The findings had been lately revealed within the Journal of Dental Analysis. Researchers examined 10 totally different ASR techniques to see how effectively they may transcribe spoken orthodontic medical data into written textual content.
The very best-performing system was an experimental pipeline combining OpenAI’s GPT-4o transcription with a big language mannequin for error correction, intently adopted by the Heidi Well being digital scribe and GPT4oTranscribe speech-to-text software programming interface.
Why this issues
Dentists spend vital time typing up medical notes typically through the session, which may scale back face-to-face time with sufferers.
ASR instruments can permit clinicians to dictate their medical notes naturally, liberating them as much as focus extra on direct interplay with the affected person. Essentially the most superior techniques had been quicker and extra correct than guide typing, with as much as 60% in time financial savings.
Key findings
The AI-enhanced experimental pipeline (GPT4oTranscribeCorrected) was discovered to have the bottom error fee, particularly with technical dental phrases. Business techniques like Heidi Well being additionally carried out effectively, however others—resembling Dragon Anyplace—had excessive error charges and even launched clinically vital errors. Importantly, background noise and accent had minimal influence on the very best techniques, making them appropriate for real-world medical settings.
Warning stays, nevertheless. Whereas the expertise is promising, researchers warn that clinically vital errors—resembling misidentifying tooth or therapy plans—can nonetheless happen. They advocate a “human-in-the-loop” method, the place clinicians evaluate and edit transcripts reasonably than counting on them blindly.
Lead writer Ruairi O’Kane stated, “AI speech tools can streamline documentation and improve efficiency, but we must remain vigilant. Even subtle transcription errors can potentially impact patient care.”
What’s subsequent?
The staff suggests future techniques ought to embody confidence indicators to flag unsure phrases and be skilled on bigger, extra numerous dental datasets. In the end, the aim is to assist clinicians grow to be editors of their notes—not simply authors—whereas sustaining security and accuracy.
Extra info:
R. O’Kane et al, Transcription Accuracy of Computerized Speech Recognition for Orthodontic Medical Information, Journal of Dental Analysis (2025). DOI: 10.1177/00220345251382452
Offered by
King’s Faculty London
Quotation:
AI speech instruments may revolutionize dental record-keeping—however warning is required (2025, November 6)
retrieved 6 November 2025
from https://medicalxpress.com/information/2025-11-ai-speech-tools-revolutionize-dental.html
This doc is topic to copyright. Aside from any truthful dealing for the aim of personal examine or analysis, no
half could also be reproduced with out the written permission. The content material is supplied for info functions solely.
