Pharmacometric Models

Using Ordinary Differential Equation (ODE) solvers we encode how the drug diffuses through different part of the body.

These models are often used in preclinical and Phase I trials in order to establish safe dosing regimens. While safety is the main concern, we strive to detect signs of early efficacy.

Main Challenge

Small data and non-linear system dynamics

Our Unique Solution

We are able to infer both the population and individual level parameters and present it to non-specialists in an easily interpretable form.

Joint Survival Models

Linking the model for the hazard with submodels for individual biomarkers. Joint models for comorbidities.

In Phase II and Phase III these models inform the risk of an event by incorporating individual patients’ biomarker evolution, which may be governed by a complex non-linear process.

Main Challenge

Non linear proces dynamics, predicting for the study arm and for the individual patient.

Our Unique Solution

Continuing to incorporate machanistic and semi-mechanistic models at this stage to help guide efficacy.

Genomic Models

Retain proper uncertainties and produce more accurate predictions than popular tools like Lasso and PCA.

This is the wide data of the big data spectrum. Traditional machine learning approaches including clustering and dimensionality reduction are unlikely to uncover meaningful clinical relationships.

Main Challenge

Too many possible combinations frequently guide the researchers to confuse signal with noise.

Our Unique Solution

By using better priors and including candidate features in our generative models allows us to test their efficacy and construct proper uncertainty intervals.

Personalized Brain Network Models

In collaboration with the leading neuroscientist from the University of Marseille, we are modeling the brain dynamics of epilectic seizures with stochastic differential equations.

Neuroscientists have spent years learning about the processes that govern signal propagation in the brain including the conditions that results in pathologies including epilepsy. Even though these systems are still not completely understood, enough progress have been made to develop a model for the human brain.

Main Challenge

Dynamic non-linear models are computationally difficult to fit, even with the most advanced inference algorithms.

Our Unique Solution

By using simulations, reparametrizations, and the ability to quickly fit hundreds of model variations, we are able to sufficiently approximate the data generating process.

Want to learn more?

Get in Touch!