Welcome to Pumas

Pumas (PharmaceUtical Modeling And Simulation) is a suite of tools to perform quantitative analytics of various kinds across the horizontal of pharmaceutical drug development. The purpose of this framework is to bring efficient implementations of all aspects of the analytics in this domain under one cohesive package. Pumas 2.0 currently includes:

• Non-compartmental Analysis
• Specification of Nonlinear Mixed Effects (NLME) Models
• Simulation of NLME model using differential equations or analytical solutions
• Deep control over the differential equation solvers for high efficiency
• Estimation of NLME parameters via Maximum Likelihood, Expectation Maximization and Bayesian methods
• Parallelization capabilities for both simulation and estimation
• Mixed analytical and numerical problems
• Simulation and estimation diagnostics for model post-processing
• Interactive model exploration and diagnostics tools through webapps
• Automated report generation for models and non-compartmental analysis
• Global and local sensitivity analysis routines for multi-scale models
• Bioequivalence analysis
• Optimal design of experiments

Additional features are under development, with the central goal being a complete clinical trial simulation engine which combines efficiency with a standardized workflow, consistent nomenclature, and automated report generation. All of this takes place in the high level interactive Julia programming language and integrates with the other packages in the Julia ecosystem for a robust user experience.

License

Pumas is covered by the Julia Computing EULA. Pumas is a proprietary product developed by Pumas-AI, Inc. It is available free of cost for educational and research institutes. For commercial use, please contact sales@pumas.ai or to know more visit Pumas for Enterprises

Getting Started: Accessing Pumas

Starting version 2.0, Pumas is available exclusively via the Pumas for Enterprises or Pumas for Academia platform powered by JuliaHub. Once you register for access, the getting started video series will guide you through the platform and help you get started.

One can start using Pumas by invoking it from the REPL as below.

using Pumas

To start understanding the package in more detail, please checkout the tutorials at the start of this manual. We highly suggest that all new users start with the Introduction to Pumas tutorial! If you find any example where there seems to be an error, please open an issue in the discussion forum listed below.

If you have questions about usage, please join the official Pumas Discourse and take part in the discussion there.

Annotated Table Of Contents

Below is an annotated table of contents with summaries to help guide you to the appropriate page. The materials shown here are links to the same materials in the sidebar. Additionally, you may use the search bar provided on the left to directly find the manual pages with the appropriate terms.

Tutorials

These tutorials give an "example first" approach to learning Pumas and establish the standardized nomenclature for the package. Additionally, ways of interfacing with the rest of the Julia ecosystem for visualization and statistics are demonstrated. Thus we highly recommend new users check out these tutorials before continuing into the manual. More tutorials can be found at https://tutorials.pumas.ai/

• Introduction to Pumas
• Working Example
• Using the Model Macro
• Building a Subject
• Running a Simulation
• Handling the SimulatedObservations
• Conclusion

Basics

The basics are the core principles of using Pumas. An overview introduces the user to the basic design tenants, and manual pages proceed to give details on the central functions and types used throughout Pumas.

• Overview of Pumas
• The Pumas NLME Workflow
• The Pumas NCA Module Workflow
• The Pumas Bioequivalence Workflow
• The Pumas Optimal Design Workflow
• Defining NLME models in Pumas
• The @model macro interface
• The PumasModel Function-Based Interface
• The @emmodel macro interface
• Model Representation
• Using the latexify output
• Configurations
• Stitching things together
• A note about future releases
• Dosage Regimens, Subjects, and Populations
• Dosage Regimen Terminology
• The Subject Constructor
• PumasNDF
• Simulation of Pumas Models
• The simobs Function
• Handling Simulated Returns
• Visualizing Simulated Returns
• Different usage patterns of simobs
• Estimating Parameters of Pumas Models
• Defining Data for Estimation
• Maximum Likelihood Estimation
• Bayesian Estimation
• Maximum A Posteriori (MAP) Estimation
• Estimating Parameters using SAEM
• Frequently Asked Questions (FAQ)

Model Components

This section of the documentation goes into more detail on the model components, specifying the possible domain types, dosage control parameters (DCP), and the various differential equation types for specifying problems with analytical solutions and alternative differential equations such as delay differential equations (DDEs), stochastic differential equations (SDEs), etc.

• PumasModel-Domains
• Matching Parameter Types and Domains
• RealDomain and VectorDomain
• Positive-Definite Matrix Domains
• Distributional Domains
• Constrained
• PumasEMModel-Domains
• Declaring parameters: $@param$ and $@random$ blocks
• Formulas with covariates
• Dose Control Parameters (DCP)
• Input Definition
• Analytical Solutions and Differential Equations
• Analytical Solutions
• Differential Equations DEProblem
• PumasModel-Error models
• Gaussian models
• Other continuous models
• Models for discrete data
• Models for incomplete data
• Summary of Error Models
• PumasEMModel-Error models
• Gaussian models
• 0-Dispersion-Parameter models
• 1-Dispersion-Parameter models
• Summary of Error Models

Analysis

This section of the documentation defines the analysis tooling. Essential tools such as diagnostics, plotting, report generation, and sensitivity analysis are discussed in detail in this portion.

• Inference
• Extracting standard errors (stderror)
• Simulation and Estimation Diagnostics
• The inspect Function
• Global Sensitivity Analysis
• Visual Predictive Check (VPC)
• Overview
• Plotting
• Publication-Quality and Interactivity
• Keywords
• Pagination
• Functions
• Customizing Plots
• Interactive Web Applications
• Exploration of Estimates
• Stopping Applications
• Listing Models
• Fitted Model Comparisons
• PDF Reports
• NCA
• Pumas

Non Compartmental Analysis

This section of the documentation covers the use of Non Compartmental Analysis tooling available with Pumas and independently.

• Non-Compartmental Analysis (NCA)
• Read Source Data
• Create a NCAPopulation
• Exploratory Data Analysis
• Run Analysis
• Generate Report
• Data Format and parsing for NCA
• Pumas-NCA data format - PumasNCADF
• read_nca
• Examples
• Specification of Groups
• Handling Missing and BLQ Data
• Missing data handling
• BLQ handling
• The NCA Functions
• Analysis and Reporting
• run_nca
• summarize
• report

Bioequivalence

Pumas Development Team

Please visit www.pumas.ai to know more about our team and capabilities.

Citing Pumas

If you use Pumas within scientific reports, please cite Pumas via the following paper:

Rackauckas, Chris, et al. "Accelerated predictive healthcare analytics with pumas, a high performance pharmaceutical modeling and simulation platform." (2020).

or in Bibtex form:

@article{rackauckas2020accelerated,
  title={Accelerated predictive healthcare analytics with pumas, a high performance pharmaceutical modeling and simulation platform},
  author={Rackauckas, Chris and Ma, Yingbo and Noack, Andreas and Dixit, Vaibhav and Mogensen, Patrick Kofod and Byrne, Simon and Maddhashiya, Shubham and Santiago Calder{\'o}n, Jos{\'e} Bayo{\'a}n and Nyberg, Joakim and Gobburu, Jogarao VS and others},
  year={2020}
}