A BAYESIAN DECISION FRAMEWORK FOR OPTIMIZING SEQUENTIAL COMBINATION ANTIRETROVIRAL THERAPY IN PEOPLE WITH HIV

Area of Science:

• * Computational biology and bioinformatics
• * Precision medicine and personalized healthcare
• * Statistical modeling and data science

Background:

• * Combination antiretroviral therapy (cART) is crucial for managing HIV but associated with adverse effects.
• * Optimizing cART for individual people with HIV (PWH) is essential for long-term health and reducing comorbidities.
• * Precision medicine approaches are needed to tailor cART regimens effectively.

Purpose of the Study:

• * To develop a data-driven framework for optimizing personalized cART assignments.
• * To address the high-dimensional combinatorial challenge in estimating cART effects.
• * To reduce the risk of comorbidities while maintaining viral suppression in PWH.

Main Methods:

• * A two-step Bayesian decision framework was developed for sequential cART optimization.
• * A multivariate Gaussian process model was used for longitudinal individual observations.
• * An uncertainty-penalized policy optimization was employed using quantified uncertainty.

Main Results:

• * The proposed method demonstrated clinical utility in assisting treatment decisions for PWH.
• * The framework effectively balances viral suppression and comorbidity risk reduction.
• * Application to the Women's Interagency HIV Study data validated the approach.

Conclusions:

• * The Bayesian decision framework offers a novel approach to personalized cART optimization.
• * This method aids physicians in making informed treatment choices for PWH.
• * It contributes to improving long-term health outcomes for people living with HIV through precision medicine.