Challenges and pitfalls of building models for Flu/RSV and respiratory viruses

Aims: Mechanistic viral kinetic (VK) models are increasingly being used to support drug development strategies in the treatment of respiratory virus infection [1]. We overview some of the challenges with these models, using influenza and Respiratory Syncytial Virus (RSV) as representative examples for respiratory viral infections.

Methods: The structure of mechanistic VK models is represented using known biology, typically characterizing the viral life-cycle, drug pharmacodynamics and clinical disease symptoms (Figure 1). Challenges associated with modelling each of these kinetic processes were investigated, including consideration of study design and data limitations.

Figure 1 – Schematic representation of the structural model describing respiratory virus kinetics, pharmacodynamic effects and composite symptom scores.

Results: Parameter identifiability is a challenge during VK model development which is due to study design and measurement techniques. This can be overcome by utilising ‘known’ physiological estimates of RSV and influenza lifespans (1/k + 1/) [2,3]. Improved model stability and individual predictions are obtained by eta correlation of infectivity and virion production. In addition, an “effective” antiviral treatment will often result in a large proportion (>50%) of the viral load data reported as BLQ, which can be appropriately handled with M3 implementation [4]. As the doses administered in challenge studies are often supra-therapeutic, this can present difficulties with EC50 estimation. Standard approaches to model evaluation can be challenging; possible solutions to help identify model robustness and mis-specification will be presented.

Conclusion: Mechanistic VK models can provide valuable insights in understanding the pathophysiology of respiratory virus infection and pharmacodynamic effects. However, several difficulties are presented during model development, and may require careful consideration when translationally simulating to target populations.


1. Canini L, Perelson AS. Viral kinetic modeling: state of the art. J Pharmacokinet Pharmacodyn 2014 41: 431-443.

2. Baccam P, Beauchemin C et al. Kinetics of influenza A virus infection in humans. J Virol 2006 80: 7590-7599.

3. González-Parra G, Dobrovolny HM. Assessing Uncertainty in A2 Respiratory Syncytial Virus Viral Dynamics. Comput Math Methods Med 2015 doi: 10.1155/2015/567589.

4. Beal SL. Ways to fit a PK model with some data below the quantification limit. J Pharmacokinet Pharmacodyn 2001 28: 481-504.