Evaluation of Machine Learning Models for Detecting Disambiguation on Medical Abbreviations

Abstract

The number of medical abbreviations in the world is due to the increasing number of diseases, technological advances in the medical field, research in the medical field, and the emergence of various drugs. A large number of medical abbreviations often have the same abbreviation but it has a different meaning. The similarity of these medical abbreviations often results in ambiguous abbreviations. The ambiguity of this abbreviation can be reduced by creating a system based on Artificial Intelligent (AI). In this paper, we have compared various models using Naive Bayes, LSTM, Logistic Regression, and SVM to get the best model for medical abbreviations disambiguation. The experimental results indicate that the highest model accuracy is obtained by LSTM model, which is at 97.21%.