PREDICTION CLASSIFICATION AND MODELLING USING DECISION TREE WITH ORDERED REGRESSION AND ITS APPLICATION TO SOCIO-BEHAVIORAL FACTORS ASSOCIATED WITH TOOTHBRUSHING FREQUENCY IN CHILDREN

Abstract

Toothbrushing is considered the best self-care behavior for the prevention of oral diseases. Brushing teeth twice a day is considered the social norm, but the development of such habits is dependent on psychosocial, economic, and environmental factors. Recognizing the significance of statistical modeling in medical sciences, this study will use decision trees and ordinal regression to predict frequency of toothbrushing in children. The methodology will be harmonized in the R syntax. The study illustrated the development of the method using 527 observations from WHO oral health questionnaire for children. Before regression analysis, the clinical relevance and significance of each of the 28 variables will be assessed using decision tree analysis and tested for accuracy. The classification obtained will be used as an input for the ordinal regression modeling. According to decision tree analysis, smoking, maternal education, dietary habits, history of toothache and self-rated tooth health contributed significantly to the children’s overall toothbrushing frequency. These six variables were used as input for ordinal regression analysis and the developed syntax was used to assess the goodness of fit for the model. Our proposed method achieves the highest level of forecasting precision possible. The process is an alternate to ordinal regression modelling as the selection of appropriate variables is based on computational analysis, forecasting the importance of the independent variables chosen for the final model. This process demonstrates the possibility of developing prediction models which can then be used to formulate clinical hypothesis and inform future researchers.