NLME

PBPK

• Component based model building for easy scalability
• Model/component serialization capabilities
• Qualification and validation workflows including global and local sensitivity analysis
• Auto Parallelism and scalability
• Integrated with all Pumas modules
  • E.g. Virtual BE studies with IVIVC+PBPK+BE
• Parameter optimization with Maximum Likelihood and Bayesian Methods

QsP

• Utilize the fast solvers available in DifferentialEquations.jl
• Component based model building for easy scalability
• Virtual population generation
• Qualification and validation workflows including global and local sensitivity analysis
• Model reduction/lumping workflows
• Model/component serialization capabilities
• Auto Parallelism and scalability
• Integrated with all Pumas modules
• Parameter optimization with Maximum Likelihood and Bayesian Methods

Optimal Design

• FO based Fisher Information Matrix (FIM) calculations
• FIM calculations that use the design of elements of the dataset
• Underlying design allows extensibility to cover other FIM approximations such as GLMM base FIM’s
• Seamless integration with NLME modules
• Full suite for OD include D/ED/Adapative - WIP

Machine and Deep Learning Integration

• Seamlessly mix neural networks and differential equations:
  • Learn unknown nonlinear dynamics and covariate-parameter maps directly from data
  • Make interpretable network layers defined by pharmacometric models
  • Train surrogate networks for real-time decision making in deployed apps
• Directly integrate with "big data" sources:
  • Train neural network mixed differential equations with EHR
  • Utilize JuliaDB for datasets unable to fit into memory

IVIVC

• Level A,B,C correlations
• Guided workflow for IVIVC development
• Differential equation and model-based IVIVC
• Seamless integration with NCA/BE/PBPK modules will facilitate integrated research such as virtual bioequivalence trials

Bioequivalence

• All standard design available
  • Parallel
  • Crossover: 2 by 2
  • Crossover: Balaam's design
  • 2S3P2F
  • 4S4P2F
  • Nonparametric
  • Williams design
  • NTI and HVD
• Seamless integration across Pumas to develop model-based BE solutions
• Access to JuliaStats for more complex designs

NCA

• Industry standard non-compartmental analysis toolkit
• NCA functions can be used in all other components of PuMas
  • Deriving auc/cmax from a model prediction and using it to drive an outcome
  • Calculating auc/cmax in IVIVC models

Clinical Trial Simulator

• A flexible toolkit that utilizes all Pumas modules to simulate the most complex of clinical trials at ease
• Core integration with JuliaStats allows evaluation of designs for a specific hypothesis or threshold criteria
• Automatic parallelization at multiple levels to speed up simulation