Accuracy of Machine-learning Model Using Prediction One^Ⓡ to Predict Increase in Number of Decayed, Missing, Filled Teeth (DMF Teeth) after 5 Years

Abstract

 In recent years, there has been an increase in the use of artificial intelligence to predict the onset of disease using machine-learning models. In our study, we evaluated the accuracy of a machine-learning model using Prediction One^Ⓡ to predict an increase in the number of decayed, missing, and filled teeth (DMF teeth) after 5 years in adults. We enrolled 284 participants who underwent dental checkups at Asahi University Hospital in 2016 and 2021. Data on the 284 dental checkups were divided into training (200 participants) and validation (84 participants) data. A machine-learning model using Prediction One^Ⓡ Version 1.3 was created based on dental checkup training data in 2016 and number of DMF teeth in 2016 and 2021. Then, we examined the accuracy of the machine-learning model using the validation data. At the results, we showed a positive correlation between the predicted increase in the number of DMF teeth after 5 years by the machine-learning model and actual increase after 5 years, with a correlation coefficient value of 0.802. Furthermore, sensitivity and specificity between the presence or absence of an increase in the number of DMF teeth predicted by the machine-learning model and presence or absence of an actual increase in the true number showed values of 1.00 and 0.77, respectively. These results suggest that the machine-learning model created by Prediction One^Ⓡ can predict an increase in the number of DMF teeth after 5 years with high-level accuracy in adults.