Author: Vikram Sinha, PhD on May 30, 2019 
As the use of quantitative systems pharmacology (QSP) models is seeing widespread models – the publication by Kirouac et al., 2019 raises the issue of reproducibility. The authors conducted a survey of published models (see Table 1, Kirouac et al., 2019) and found that after evaluating 12 models only 4 were executable, in that figures from the associated manuscript could be reproduced. The authors point to the diversity of modeling platforms that are in use and diversity of “code” files. While this journal requires model code be submitted as part of any manuscript – the diversity in the modeling platforms and submitted code is a reality. The authors propose a few ideas to enable model sharing going forward, including annotated, standardized model scripts and files.
To see continued use and acceptability of QSP models, reproducibility and ease of access of previously established models are critical. To a major extent, the more established approaches of population-based approaches and physiologically-based PK models have standardized the “code” with these models; this has resulted in consistent training and broader use in their application. Indeed, as QSP models become more public – the debate should reside with the implications of the biology rather than questions about the output due to an inability to reproduce models.
By nature, the development and use of QSP models are highly technical and practitioners of QSP study extremely complex biological systems where data sources are varied, and models tend to be assumption rich. While descriptions of methods and data sources are included in published papers, these sections often lack enough detail. As PSP does, journals should include policies on data availability and encourage the use of reusable model code as part of manuscript submissions. If practitioners can agree to standards for the submission of model code, wider uptake and impact would be accelerated.
Image by Kirouac, et al. CPT Pharmacometrics Syst. Pharmacol., doi.org/10.1002/psp4.12390, is licensed under CC BY-NC-ND 4.0. ©2019 The authors.
The comment feature is locked by administrator.