Machine learning aided multiscale modelling of the HIV-1 infection in the presence of NRTI therapy

The Human Immunodeficiency Virus (HIV) is one of the most common chronic infectious diseases of humans. Increasing the expected lifetime of the patients depends on the use of optimal antiretroviral therapies. The emergence of the drug-resistant strains may decrease the effects of treatments and lead to Acquired Immune Deficiency Syndrome (AIDS) even if the existence of antiretroviral therapy. Investigation of the genotype-phenotype relations is a crucial process to optimize the therapy protocols of the patients. Here we propose a mathematical modelling framework to address the effect of initial strains, initiation timing and adherence levels of nucleotide reverse transcriptase inhibitors (NRTI) on the emergence of a possible AIDS phase. For the first time, we have combined the existing Stanford HIV drug resistance data with a multi-strain within-host ordinary differential equation (ODE) model to track the dynamics of most common NRTI resistant strains. Regardless of the drug choice, the late initiation and poor adherence levels to the NRTI therapy increase the probability of the emergence of the AIDS phase. Overall, the 3TC, D4T-AZT and TDF-D4T drug combinations provide higher success rates. The results are in line with genotype-phenotype data and pharmacokinetic parameters of the NRTI inhibitors, but we show the heavy influence of neighbour viral strains of the initial ones has a considerable effect on the success/failure rates. Improving multiscale models can contribute to understanding the disease progression and treatment options.