Prediction Of Carboxylic Acid Toxicity Using Machine Learning Model

Abstract

Carboxylic acids are organic compounds characterized by the presence of a carboxyl functional group capable of donating a proton and forming carboxylate ions in aqueous solutions. The carboxylic acid has widely been used in in manufacturing and medical applications. The rapid growth in carboxylic acid has established a need to predict its toxicity. The purpose of this paper to build predictive toxicity of carboxylic acid models by using five molecular descriptors (refractive index, The octanol/water partition coefficient (log P), acid dissociation constant (pKa), density, and dipole moment) through Machine Learning algorithms. The accuracy of the Machine Learning algorithm was determined by using three different types of models which are Decision Tree, Random Forest and k-Nearest Neighbour (k-NN). Among the machine learning algorithms used, we have determined that the decision tree is the best model for predicting the toxicity of carboxylic acid. This finding demonstrates that the decision tree model exhibits an acceptable level of performance in predicting toxicity within the field of toxicology.