Integrated PK/PD Modeling and Machine Learning to Support CNS Compound Selection

A CNS drug program needed to choose among 29 compounds for first-in-human studies, complicated by indirect pharmacology involving a rapidly-turning-over biomarker, a slowly developing efficacy marker, and heterogeneous preclinical data across multiple study designs. Certara implemented a machine-learning-enabled nonlinear mixed-effects modeling strategy using Phoenix NLME to analyze all studies simultaneously, building a unified mechanistic framework that separated drug-specific parameters from physiological system parameters. The analysis showed the most in-vitro-potent compound was not necessarily the optimal first-in-human candidate; a compound nicknamed the "ugly duckling" was ultimately selected for having the most favorable overall pharmacological profile despite not being the most potent, demonstrating "objective, system level decision making." No specific efficacy percentages or response rates are given.

The customer is anonymised in Certara’s own case study — not withheld by us.

Certara is a biosimulation software company (Simcyp, Phoenix) whose PK/PD and QSP models predict drug behavior in the body; the FDA itself licenses Certara's software to review sponsor submissions.

This sits in the Build stage of the AI use-case lifecycle: Named examples from the foundation-model labs and platforms pharma teams use to actually construct, train and generate the model or application.