Possibility of Anaesthetical Activity Prediction of N-(Pyrrolidinyl)ethyl Esters of Alkoxyphenylcarbamic Acids

Abstract

A back-propagation-type neural network consisting of three layers was used for predicting of surface and infiltration anaesthetical activity of pyrrolidinylethyl esters of 2- and 3-alkoxyphenylcarbamic acids. The prediction of local anaesthesia was based on the knowledge of this property for piperidinylethyl esters, piperidinylpropyl esters and azepanylethyl esters of 2- and 3-alkoxy- phenylcarbamic acids. The structural formulas of compounds pre-processed to the numerical input and RP-HPLC capacity factors (characterizing the lipophility of tested compounds) were used for generating inputs for the neural network. Anaesthetical activities of tested drugs calculated by neural network were in a good agreement with the experimentally measured data. Average error of predicted values ranges from 2% for the surface anaesthesia of pyrrolidinylethyl esters of 2-alkoxy- phenylcarbamic acids to 15% for the infiltration anaesthesia of pyrrolidinylethyl esters of 3-alkoxyphenylcarbamic acids.