Comparison of Classifier Fusion Methods for Predicting Response to Anti HIV-1 Therapy

Abstract

BackgroundAnalysis of the viral genome for drug resistance mutations is state-of-the-art for guiding treatment selection for human immunodeficiency virus type 1 (HIV-1)-infected patients. These mutations alter the structure of viral target proteins and reduce or in the worst case completely inhibit the effect of antiretroviral compounds while maintaining the ability for effective replication. Modern anti-HIV-1 regimens comprise multiple drugs in order to prevent or at least delay the development of resistance mutations. However, commonly used HIV-1 genotype interpretation systems provide only classifications for single drugs. The EuResist initiative has collected data from about 18,500 patients to train three classifiers for predicting response to combination antiretroviral therapy, given the viral genotype and further information. In this work we compare different classifier fusion methods for combining the individual classifiers.Principal FindingsThe individual classifiers yielded similar performance, and all the combination approaches considered performed equally well. The gain in performance due to combining methods did not reach statistical significance compared to the single best individual classifier on the complete training set. However, on smaller training set sizes (200 to 1,600 instances compared to 2,700) the combination significantly outperformed the individual classifiers (p<0.01; paired one-sided Wilcoxon test). Together with a consistent reduction of the standard deviation compared to the individual prediction engines this shows a more robust behavior of the combined system. Moreover, using the combined system we were able to identify a class of therapy courses that led to a consistent underestimation (about 0.05 AUC) of the system performance. Discovery of these therapy courses is a further hint for the robustness of the combined system.ConclusionThe combined EuResist prediction engine is freely available at http://engine.euresist.org.

HIV infection and psychiatric illness

“…96 Computer programs for predicting response to combination antiretroviral therapy, given the viral genotype and further information, have lately been developed. 97 Since the process of reverse transcription lacks a proofreading mechanism, mutations leading to drug resistance are generated frequently. 97 Efavirenz has well documented acute CNS side effects, and have been reported in up to 50% of the patients.…”

“…97 Since the process of reverse transcription lacks a proofreading mechanism, mutations leading to drug resistance are generated frequently. 97 Efavirenz has well documented acute CNS side effects, and have been reported in up to 50% of the patients. Such side effects include dizziness, light-headedness, somnolence, insomnia, unusual dreams, amnesia, agitation, derealisation, depersonalization, hallucinations and euphoria, and typically begin after the first dose and usually resolve within 2-4 weeks of therapy or somewhat later.…”

HIV-GRADE: A Publicly Available, Rules-Based Drug Resistance Interpretation Algorithm Integrating Bioinformatic Knowledge

“…However, this assumption is supported by observations like the successful treatment of patients with the dual protease inhibitor combination-boosted lopinavir and saquinavir [27] , where the resensitizing effects of the mutation L76V seem to be relevant [28] . Although bioinformatic-based prediction tools like the EuResist engine [29,30] have shown a superior performance when predicting the success of combinations of antiretroviral drugs, as compared to human experts [31] or rules-based algorithms [32] , they are still not widely accepted. Rules-based systems or systems that predict the phenotypic resistance factor for each drug, like geno2pheno [resistance] (virtual phenotype), are most frequently used.…”

Error Corrective Fusion of Classifier Scores for Spoken Language Recognition

“…The final output is solely given by the most reliable classifier of the fusion system. A modified version of LAW was also presented [37] in which the output is a weighted mean of reliability estimate of each subsystem which is the fraction of correctly classified samples in the local region surrounding the test segment reported by KNN classifier. The output of the fusion system is then defined as,…”