Ensemble Classifiers for Predicting HIV-1 Resistance from Three Rule-Based Genotypic Resistance Interpretation...

Area of Science:

• Virology
• Computational Biology
• Clinical Medicine

Background:

• Antiretroviral (ARV) drug resistance is a significant challenge for individuals with HIV.
• Existing rule-based algorithms for inferring HIV-1 susceptibility from genotypic data show discordance, complicating clinical treatment decisions.
• A unified approach is needed to reconcile differing interpretations of genotypic resistance data.

Purpose of the Study:

• To develop and evaluate ensemble classifiers that integrate multiple HIV-1 drug resistance interpretation algorithms.
• To provide a single, consistent HIV resistance profile for improved clinical decision-making.
• To compare the performance of different ensemble strategies for resistance interpretation.

Main Methods:

• Developed ensemble classifiers by integrating three major interpretation algorithms: Agence Nationale de Recherche sur le SIDA (ANRS), Rega, and Stanford HIV Drug Resistance Database (HIVdb).
• Employed three ensemble approaches: stacked generalization, plurality vote, and best-performing algorithm selection.
• Utilized Friedman's test for strategy comparison and evaluated classifier performance using F-measure, sensitivity, and specificity.

Main Results:

• All three ensemble strategies demonstrated comparable performance across selected antiretrovirals.
• Stacked generalization, particularly with a Naïve Bayes learning algorithm, showed a statistically superior F-measure in specific instances.
• Ensemble classifiers successfully generated a single resistance profile from multiple interpretation systems.

Conclusions:

• Ensemble classifiers offer a viable alternative for interpreting HIV-1 genotypic resistance data.
• These integrated approaches can aid clinicians by providing a consolidated resistance profile, thereby supporting more informed treatment decisions.
• The study highlights the potential of ensemble methods to enhance the clinical utility of genotypic resistance testing in HIV management.