Prediction of Multiple Recession in Lower Anterior Using Artificial Intelligence

Abstract

Background and Aim: Soft tissue recession is the movement of the gingival edge apical to a tooth's cemento-enamel junction (CEJ) or dental implant platform. This study used intraoral frontal photos to test machine-learning methods for lower anterior tooth recession detection and multiple recession.
Materials and Methods: Orange was employed with squeeze net embedding for gingival recession images. We trained and tested logistic regression and naïve bayes algorithms on intraoral frontal images of diverse lower anterior recession kinds to predict and classify the image embeddings. Accuracy was measured via a confusion matrix and roc curve.
Results: Squeezenet-embedded machine learning systems accurately classified recession, predicting and classifying lower anterior tooth recession. The accuracy of naïve bayes and logistical regression is 96% and 100%, respectively, with class accuracy of 95% and 100%.
Conclusion: Predicting and classifying multiple lower anterior recessions using AI advances clinical practice. It has no bias or negative examination error compared to human examination. It predicts early recession better than humans.