Author/Editor		Grabec, I; Grošelj, D
Title		Prognosis of periodontal disease healing process
Type		članek
Source		In: Lavrač N, editor. CADAM-95. Zbornik Računalniška analiza medicinskih podatkov; 1995 nov 27-28; Bled. Ljubljana: Inštitut Jožef Štefan,
Publication year		1995
Volume		str. 146-54
Language		eng
Abstract		It is demonstrated how the non-parametric regression, performed by an artificial neural network, can be applied to describe the effect of Widman flap procedure on the periodontal disease in the presence or absence of drug and to forecast the development of the healing process in the future. As an indicator of the periodontal condition the displacement of incisors was measured by an experimental set-up particularly designed for this purpose. Two groups of 11 patients, one with and the other without metronidazole administration, were examined in the period of one year. The measurements revealed a significant influence of the drug presence on the changes of tooth mobility. The data of tooth mobility recorded in dependence of time were further used together with drug presence indicator as components of protoype vectors of a data base. The healing process on a population was modeled empirically using a network of Gaussian neurons having centers of response fields determined by prototype vectors. For the recall of stored information the conditional average estimator was used which corresponds to the non-parametric regression. It is applicable for the estimation of an average behavior of patients under various treatments and for the prediction of the healing process on periodontal tissues in other patients. A comparison of a predicted and the actual healing process permits us to estimate quantitatively if the recovery develops according to expectations.
Descriptors		PERIODONTAL DISEASES TOOTH MOBILITY ANTIBIOTICS PROGNOSIS