THIRD MOLAR MATURITY INDEX IN INDONESIAN JUVENILES: COMPARING LINEAR AND POLYNOMIAL KERNEL PERFORMANCE IN SUPPORT VECTOR REGRESSION FOR DENTAL AGE ESTIMATION

Abstract

Dental age estimation is a branch of forensic odontology that plays a pivotal role in identifying, examining, or determining the legal status of the living and the dead. This research explores the capability of support vector regression to estimate chronological age from the third molar maturity index (I_3M) in Indonesian Juveniles and compares the linear and kernel performance. Two hundred and twenty-two orthopantomo-graphy were measured using I_3M in the lower left third molar and processed using R Studio with Caret extension. The analysis was separated into two groups, group 1 using only I_3M as a predictor, and group 2 using both I_3M and sex. Both groups were analyzed using SVR with the linear and polynomial kernel. The result suggests that using polynomial kernel SVR in group 1 produces the best results, with an R² value of 0.64, RMSE of 1.588 years, and MAE of 1.25 years using degree = 3, c = 0.25. However, the addition of a sex predictor in the model reduces its accuracy when using the polynomial kernel.