Metrum Research Group scientists presented at the 1st annual R/Pharma conference in Boston, MA on August 15-16, 2018. Be sure to check out their presentations.
A physiologically-based mathematical model was developed as a series of ordinary differential equations to describe compositional changes (in fat and fat-free mass, FM & FFM) due to metabolizable energy exchanges in babies from birth to 2 years in low-to-middle income countries.1 The objective of this work was to identify potential biomarkers for future intervention studies, identify when to intervene to protect and/or rescue growth in individuals suffering from malnutrition, and to identify which of these individuals would be more or less likely to respond to a nutritional intervention.
A translation of this model (155 parameters and 26 compartments) using R and the open-source mrgsolve package2 provided an efficient platform for multi-parameter optimization, as required during additional model development and for subsequent simulations. For comparison, a 8.62 seconds simulation with viral and bacterial infections (no interventions) in the R/mrgsolve implementation required 226 seconds in Matlab. Model translation to R also enabled simulations with a Shiny App, allowing users to simulate individual infant phenotypes and infection events and visualize growth and energy levels over time, relative to healthy (WHO) standards.
The model currently also includes a relatively simple implementation of persistent antibiotic therapy with a potential for inclusion of drug exposure-related effects, i.e. - through a pharmacokinetic (PK) model, to describe effects of antiviral or antibiotic therapy. The challenge to this development is the scarcity of available data describing this therapy in malnourished children that would be needed for model calibration. Further development of the model includes linking to other systems models such Mother-fetus energy exchange or PBPK mother-fetus models, to enable simulations of growth beginning at gestation.
References: 1.) Bill & Melinda Gates Foundation Healthy birth, growth and development knowledge integration (HBGDki) project provided access to data and also funded this work. Model was developed by Mike Morimoto and Lyn Powell. 2.) Kyle T Baron (2018). mrgsolve: Simulate from ODE-Based Population PK/PD and Systems Pharmacology Models. R package version 0.8.12. https://CRAN.R-project.org/package=mrgsolve
Despite the explosive growth and adoption of R globally, concerns over how to qualify and administrate R continues to echo in discussions about use in regulated environments. In this talk, I’ll discuss the how to bridge the conceptual tenants of reproducibility, traceability, and accuracy to robust, yet agile, implementations such that and organization can maintain validated systems without imposing the shackles found in traditional validated environments. Furthermore, I will discuss a number of design elements specific to the open-source R ecosystem, such as using packages from CRAN and github, and cover how to embrace these, while responsibly managing risk in enterprise environments.
During the drug development, pharmacometric models are often built to characterize and understand drug efficacy and safety. Simulations based on these models can assist drug development and quantitative decision making. However, computation can be time-consuming and communication with the project team may not be productive.
Shiny applications are developed as a simulation tool which allows rapid real-time simulations based on user-selected inputs and dynamic visualization of the results. It provides an easy access to individuals with no specific background of modeling and simulation. In the talk, I will present some case studies where the shiny application was used to perform simulations and facilitate the communication with decision makers.
Since its foundation in 2004, Metrum Research Group has relied on R as the core technology and central framework for all of the company’s biomedical modeling and simulation (M&S) service activities, spanning more than 475 projects with 150+ different sponsors. Projects include pharmacokinetic-pharmacodynamic modeling, quantitative systems pharmacology models, simulation-based trial design evaluations, disease progression and patient population modeling, model-based meta analysis of competitor data, model-based comparative effectiveness assessments, and data management activities, etc., all within a regulated environment. Analyses were conducted in R or via other software tools which are managed via R scripts, functions, or packages. Key deliverables of M&S projects are routinely provided as R packages or interactive simulation applications, driven by R (and R Shiny). R has also been an essential component of Metrum’s vision for Open Science in biomedical M&S, allowing for accessibility and reproducibility of platform models developed for multiple disease areas.
See all Metrum Research Group publications here.