Optimum model for Prediction of Dental Anomaly Patterns with Deleterious Oral Habits among School Going Children-A Machine Learning Approach

Abstract

This study is to find the association between dental anomaly patterns(DAP) with deleterious oral habits among school going children of age 07 -13 years, in National Capital Region and predicts the most informative etiological criteria to the DAP development through machine learning. Openbite, spacing and tongue thrust has positive correlation, p<0.05 while crowding and spacing has negative correlation with gender, p<0. 05. The nail biting in age groups 7-9, 10-11 and 12-13 years (22.7,25.and 20.3%) is highest and tongue thrust (20.5,19.6 and 22.1%) associated with these groups respectively. Open bite neural network has Accuracy, AUC and Gini are 95.0%, 0.684 and 0.368 respectively which is better than other classifiers in training data and in test sample is 94.3% accuracy. Cross bite Neural network has Accuracy, AUC and Gini are 94.8%, 0.697 and 0.394 respectively which is better than other classifiers in training data and for test sample is 94.8% accuracy. The detection and management of dental anomalies patterns in early age can avoid potential orthodontic and esthetic problems in a child.