Prediction of chronological age based on Demirjian dental age using robust ridge regression method

Introduction: Estimation of age has an important role in legal medicine, endocrine diseases and clinical dentistry. Correspondingly, evaluation of dental development stages is more valuable than tooth erosion. In this research, the modeling of calendar age has been done using new and rich statistical methods. Considerably, it can be considering as a practicable method in medical science that is a combination of some new statistical methods. Materials and Methods: Among the methods used to determine age, the most commonly used method in the world is the modern modified Demirjian’s method based on the calcification of the permanent tooth in panoramic radiography. The study population is consisted of 87 patients who referred to Khatam-ol-Anbia Clinic of Yazd in a simple randomized method during the 12 months of the 2014-2015 year. Using the estimated age of third molar tooth and gender variables, we evaluated the calendar age. In the analysis of regression issues and especially the statistical modeling of many data such as economic data, psychology, social sciences, medical sciences, engineering, etc., we faced with the problem of collinearity among the predictor variables and the presence of remote areas in the data set. The least squares error method in estimation of the parameters of regression model was very sensitive to the outliers. Most of the existing methods for estimating the parameters of these models based on the least squares error approach, affected by the outliers, were yielded to inappropriate estimates, unexpected and high error rates. Robust methods were used to overcome the problem of the outlier observations. It is also recommended the ridge regression to fix the multicollinearity problem. Therefore, in this study, the robust ridge regression estimators will be introduced in the modeling of dependent variables that are less sensitive to the outliers. Results: The mean age of the subjects was 17.21 years old, with a gender difference of 67% female and 33% male. Additionally, in the relationship between the estimated age of 4 teeth lower right (LR), lower left (LL), upper right (UR), upper left (UL) with a correlation coefficient were above 70%. Correlation between upper and lower jaw teeth was about 30% and between the left and right teeth was 60%. The reason of using robust ridge regression model in this study is the existence of outlier data and collinearity between independent variables. Conclusion: The necessity of using advanced statistical methods in medical sciences in the recent research is very important. In order to choose the best model, we need to study the data carefully. In this research, the fitted model for prediction of age based on the robust ridge regression method was more efficient with respect to the other methods.