Image diagnosis models for the oral assessment of older people using convolutional neural networks: A retrospective observational study.
Misato Muramatsu, Masumi Muramatsu, Naoto Takahashi and 6 others
PMID 34935230WHAT IT FOUND
An image recognition model classified six visual oral assessment items with high accuracy except mucous membranes, but it was not tested in routine care, so nurses should not rely on it yet.
Key findings
01The CNN model classified lips with 98.8% accuracy, tongue with 94.3%, saliva with 92.8%, mucous membranes with 78.6%, gingiva with 93.0%, and teeth or dentures with 93.6%.
02Expert agreement was 52.4% for teeth or dentures rating 1 and 52.9% for gingiva rating 3.
03The study did not demonstrate the model's practicality in clinical practice.
STILL TO COME
How it was doneWhat they foundWhat it means for RNs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study only constructed and verified model accuracy and did not demonstrate practicality in clinical practice. Images came from 114 older people at five dental-related facilities, described as higher-risk, so results may not apply to all older adults. The number of images for each rating varied widely across categories. Only six of the eight Oral Assessment Guide items were included; voice and swallow were not assessed. Image quality may have been affected by procedures before photographs, such as rinsing or blowing air, which matter for moisture-related ratings.
The easy way to misread this
Do not conclude that this image recognition model can replace nurse oral assessment. The study built and tested the model on retrospective images, and the paper states that practicality in clinical practice was not demonstrated.