Professor/CEO and Co-founder University of Toronto/Intrepid Labs Inc. Toronto, Ontario, Canada
Formulation design for oral dosage forms is still dominated by empirical search over narrow design spaces, which too often yields “the best of what we tried” rather than what is truly optimal. We present an AI-enabled workflow that shifts the objective toward “the best of what is possible.” Instead of relying on heterogeneous literature, whose limited scope and inconsistent measurements constrain model fidelity, we use AI to structure relevant data, learn performance-driving relationships, and guide exploration of broader, more meaningful design spaces. These AI-driven workflows surface higher-performing formulations earlier, focus experimental effort where it matters, reduce API and material use, and provide transparent decision support for CMC teams. We conclude with practical steps for adoption: data readiness, model validation, and clear communication of AI-derived rationale to development teams.
Learning Objectives:
Recognize limits of empirical formulation search in narrow design spaces.
Explain how AI navigates larger design spaces efficiently.
Prioritize informative experiments to reduce API/material use.
Interpret model predictions, uncertainty, and key drivers.
Outline adoption steps: data readiness, model performance/validation, and clear communication.
.jpg "Christine Allen, PhD (she/her/hers) photo")