vpop-calibration 2.3.1

No project description provided

Vpop calibration

Description

A set of Python tools to allow for virtual population calibration, using a non-linear mixed effects (NLME) model approach, combined with surrogate models in order to speed up the simulation of QSP models.

Currently available features

• Surrogate modeling using gaussian processes, implemented using GPyTorch
• Synthetic data generation using ODE models. The current implementation uses scipy.integrate.solve_ivp, parallelized with multiprocessing
• Non-linear mixed effect models:
  • Log-distributed parameters
  • Additive or multiplicative error model
  • Covariates handling
  • Known individual patient descriptors (i.e. covariates with no effect on other descriptors outside of the structural model)
• SAEM: see the dedicated doc for more details

Getting started

• Tutorial: this notebook demonstrates step-by-step how to create and train a surrogate model, using a reference ODE model and a GP surrogate model. It then showcases how to optimize the surrogate model on synthetic data using SAEM
• Other available examples:
  • Data generation using Sobol sequences
  • Data generation using a reference NLME model
  • Training and exporting a GP using synthetic data
  • Running SAEM on a reference ODE model. Note: the current implementation is notably under-optimized for running SAEM directly on an ODE structural model. This is implemented for testing purposes mostly
  • Training a GP with a deep kernel

Support

For any issue or comments, please reach out to [e-mail], or feel free to open an issue in the repo directly.

Authors and acknowledgment

• Paul Lemarre
• Eléonore Dravet
• Adeline Leclerq-Sampson

Roadmap

• NLME:
  • Support additional error models (additive-multiplicative, power, etc...)
  • Support additional covariate models (categorical covariates)
  • Add residual diagnostic methods (weighted residuals computation and visualization)
• Structural models:
  • Integrate with SBML models (Roadrunner)
• Surrogate models:
  • Support additional surrogate models in PyTorch
• Optimizer:
  • Add SVGP for surrogate model optimization

References

• Delyon et al. 99: Bernard Delyon. Marc Lavielle. Eric Moulines. "Convergence of a stochastic approximation version of the EM algorithm." Ann. Statist. 27 (1) 94 - 128, February 1999. https://doi.org/10.1214/aos/1018031103
• Grenier et al. 2018: Grenier, E., Helbert, C., Louvet, V. et al. Population parametrization of costly black box models using iterations between SAEM algorithm and kriging. Comp. Appl. Math. 37, 161–173 (2018). https://doi.org/10.1007/s40314-016-0337-5