Predicting HIV-1 Drug Resistance: A Comparison of Three Learning Algorithms

This paper presents an application of learning algorithms to the prediction of HIV-1 phenotypic drug resistance from genotype. The objective of this research consists of two main subjects. The first part is to apply the Support Vector Machine (SVM), the Radial Basis Function Network (the RBF network), and k-Nearest Neighbor (k-NN) to predicting HIV-1 drug resistance. The second part is to study the behavior of each learning algorithms and compare the predictive performance. The results indicate that SVM yields the highest accuracy. The RBF network gives the highest sensitivity whereas k-NN yields the best in specificity.