Dynamic multidrug therapies for hiv: optimal and sti control approaches.

By B. Adams, H. Banks, H. Kwon, H. Tran. 2004

244 citations recorded by the source corpus.

We formulate a dynamic mathematical model that describes the interaction of the immune system with the human immunodeficiency virus (HIV) and that permits drug "cocktail " therapies. We derive HIV therapeutic strategies by formulating and analyzing an optimal control problem using two types of dynamic treatments representing reverse transcriptase (RT) in hibitors and protease inhibitors (PIs). Continuous optimal therapies are found by solving the corresponding optimality systems. In addition, using ideas from dynamic programming, we formulate and derive suboptimal structured treatment interruptions (STI)in antiviral therapy that include drug-free periods of immune-mediated control of HIV. Our numerical results support a scenario in which STI therapies can lead to long-term control of HIV by the immune response system after discontinuation of therapy.

Subfields: Hepatology, Immunology, Virology

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Hepatology

In Immunology

In Virology

Citations in the corpus

Reinforcement Learning in Healthcare: A Survey (2019, methodology): “Simulations show that the basin of attraction of the healthy steady-state is rather small compared to that of the nonhealthy steady state [141].”
ON IDENTIFIABILITY OF NONLINEAR ODE MODELS AND APPLICATIONS IN VIRAL DYNAMICS (2011, background): “In biomedical research, the dynamic system could be a virus, such as HIV, infecting a human and the system input could be antiretroviral therapy; how to control or stabilize the virus via intentionally designed intervention strategies is still an interesting and challenging research topic [1, 22, 55”
Clinical data based optimal STI strategies for HIV: a reinforcement learning approach (2006, methodology): “Furthermore, we consider an optimal control problem with infinite time horizon and discounted costs while in (Adams et al., 2004) a finite horizon and undiscounted costs are considered.”
Transfer Learning Across Patient Variations with Hidden Parameter Markov Decision Processes (2016, methodology): “[11] leverage the mathematical representation of how a patient responds to HIV treatments [1] in developing an RL approach to ﬁnd effective treatment policies using methods introduced in [10].”
PAC Optimal Exploration in Continuous Space Markov Decision Processes (2013, background): “The simulated HIV treatment problem (Adams et al. 2004), introduced to reinforcement learning by Ernst et al. (2006), is a six dimensional continuous state space, two dimensional discrete action space problem, modeled after clinical data.”
Model-based Reinforcement Learning for Semi-Markov Decision Processes with Neural ODEs (2020, methodology): “However, we still observe that the Latent-ODE and ODE-RNN develop a high-performing policy much faster than the model-free baseline over 1500 episodes in the HIV environment.”
Reinforcement learning design for cancer clinical trials (2009, unknown): “They consider discounted instantaneous costs (which is a continuous function directly associated with actions) as their reward function: the rationale behind this comes from a validated and identified HIV model [37].”
A Reinforcement Learning Design for HIV Clinical Trials (2014, methodology): “Combination therapy is based on the premise that all virions are not homogeneous (Adams et al., 2004).”
Robust and Efficient Transfer Learning with Hidden Parameter Markov Decision Processes (2017, methodology): “(Ernst et al. 2006) leverage the mathematical representation of how a patient responds to HIV treatments (Adams et al. 2004) in developing an RL approach to ﬁnd effective treatment policies using ﬁtted-Q iteration.”