SciML and Agentic QSP Workflows in Pumas | Tokyo,Japan

Overview

Quantitative Systems Pharmacology (QSP) has long bridged molecular biology and clinical pharmacology, enabling mechanistic modeling of complex disease and drug action. This workshop pushes that bridge forward by integrating QSP with the latest advances in scientific machine learning (SciML), generative modeling, and agentic AI — all within the Pumas ecosystem.

Building on PumasAI's prior workshops on viral dynamics and in-silico clinical trials, this session unifies three threads into a single, hands-on modeling workflow that takes participants from raw data to a complete, reproducible deliverable.

The workshop will be held at Kitasato University, Shirokane Campus, the same venue as the Mathematical Pharmacology Symposium.

Instructors: Dr Vijay Ivaturi and Lorenzo Contento

What you will Learn:

QSP in Pumas

• Constructing mechanistic models of disease biology and drug action in Pumas, with examples drawn from viral dynamics and immunology.

• Simulating in-silico clinical trials to explore dose, regimen, and patient-population effects.

DeepPumas and SciML

• Augmenting mechanistic models with neural network components to learn unknown dynamics and individualize predictions.

• Introducing normalizing flows as a flexible tool for probabilistic modeling of patient populations and parameter distributions.

• Practical guidance on when to reach for hybrid SciML models versus pure mechanistic or pure data-driven approaches.

PumasAide — Agentic Workflows

• A live demonstration of PumasAI's agentic assistant, which automates the end-to-end modeling workflow: data exploration, model building, diagnostics, simulation, and reporting.

• Going from raw dataset to a complete, regulatory-ready report in minutes rather than weeks, with a full audit trail.

Workshop Format

Hands-on, code-forward sessions built around real case studies. Participants will alternate between guided walkthroughs and self-paced exercises, with the PumasAide agent available throughout to accelerate exploration. Materials extend the existing Viral Dynamics and In-Silico Clinical Trials workshops listed below.

Who should Attend

Pharmacometricians, QSP modelers, clinical pharmacologists, and modeling & simulation scientists interested in combining mechanistic modeling with modern AI techniques. Familiarity with PK/PD or QSP modeling is helpful; prior exposure to Pumas is beneficial but not required.

Venue Details
The workshop will be held at Kitasato University Shirokane Campus, Tokyo, the same venue as the Mathematical Pharmacology Symposium.

Room: Bldg. No. 1, 5F, Seminar Room 1507

Tickets

Free registration for the two-day SciML and Agentic QSP Workflows in Pumas workshop. Seats are limited and confirmation will be subject to available capacity.