Watch on demand

Ask any pharmacometrician where their time actually goes, and the honest answer is rarely “solving the scientific question.” More often, it’s wrestling with syntax, sequencing workflow steps correctly, and getting different tools to talk to each other. This overhead sits between a modeler and the question they’re trying to answer, and it’s a tax nearly every practitioner pays, regardless of experience level.

This on demand session tackles that problem directly, exploring what happens when an AI assistant becomes a genuine modeling collaborator rather than just another tool to learn.

An AI assistant that understands pharmacometrics

Imagine describing your model in ordinary language, the way you’d explain it to a colleague, and having that description translated automatically into working, reproducible R code, executed on the same trusted Phoenix NLME engine and RsNLME toolset your team already relies on. That’s the premise behind the MCP Server for Certara RsNLME, which sits at the center of this recording.

Rather than replacing the modeler’s expertise, the assistant handles the translation layer: taking your intent, understanding the standard NLME workflow, and producing code that’s ready to run, not a rough draft that still needs heavy editing.

Speed isn’t the whole story, precision matters too

Getting to a first model faster is valuable, but it only solves half the problem. The other half is thoroughness: how completely have you actually searched the space of plausible models? That’s where pyDarwin comes in, applying machine learning driven model search to explore possibilities a manual forward addition and backward elimination process might miss or take far longer to reach.

To show this isn’t just theoretical, the presenters walk through a real comparative study using a published population PK dataset (17 DMAG). A manually recoded literature model was benchmarked against both the MCP assisted approach and pyDarwin’s automated search, all evaluated under identical selection criteria within Certara NLME. 结果就是：both automated approaches meaningfully outperformed the manual build, producing lower −2LL values and stronger QPC predictive check scores.

The takeaway isn’t that automation replaces the pharmacometrician. It’s that combining the MCP server’s speed with pyDarwin’s search thoroughness can get teams to better fitting, more predictive models faster, while the scientist retains full control over criteria, diagnostics, and the final judgment call.

What you’ll see in this recording

Certara’s team of technologists and pharmacometricians walk through this end to end, including:

• How plain language descriptions become validated, runnable NLME code
• A live demonstration using real data and modeling problems
• How pyDarwin’s search process sharpens the final model
• A side by side look at manual, machine learning, and AI assisted model building approaches
• Where human judgment remains essential, even as more of the process becomes automated
• A look at what’s next for Model Designer and the broader Phoenix Cloud pharmacometrics roadmap

Meet the speakers

Samer Mouksassi, PharmD, PhD, FCP, FISoP, Senior Director and Application Scientist at Certara, has held several roles at the company since 2007. He has helped biopharma clients apply model based drug development across more than 100 studies and has contributed to over 60 peer reviewed publications in pharmacometrics.

Mark Sale, MD, Vice President at Certara, has more than thirty years of experience in population pharmacokinetic and pharmacodynamic modeling across both academia and industry. In his current role, he oversees population PK/PD analyses and strategic consulting on drug development, and contributes to Certara’s software development group supporting AI and machine learning systems in pharmacometrics. He previously served as global director of Modeling and Simulation at GlaxoSmithKline and held faculty appointments at Georgetown, Indiana University School of Medicine, Mercer University School of Pharmacy, and the State University of New York at Buffalo.

Arjen Bos, Principal Product Manager at Certara, brings more than 20 years of experience in the life sciences industry. Since joining Certara two years ago, he has worked across a variety of Phoenix Cloud modules.

James Craig, MS, Staff Software Engineer at Certara, has 15 years of experience building scientific and pharmacometric software in collaborative, cross functional settings. His background spans statistical computing, modeling workflows, and domain specific languages for nonlinear mixed effects modeling, including PML for Phoenix NLME and NMTRAN for NONMEM, alongside enterprise and cloud native application development. He has co authored nine scientific publications in leading journals.

Who this is for

This recording is built for Clinical Pharmacologists, Quantitative Clinical Pharmacologists, Pharmacometricians, Pharmacokineticists, and PK Analysts, anyone who spends meaningful time in the modeling trenches and wants to see whether AI assistance can meaningfully change that experience.