SESSION D

The importance of exposure-response (ER) analyses in the realm of drug development cannot be overstated. These analyses serve to understand how varying doses or levels of a drug correlate with the therapeutic outcomes, side effects, and overall patient response.

In the conventional landscape, parametric ER time-to-event (TTE) models have been widely adopted. These models, which rely on a fixed form for the hazard function, make assumptions about the shape of the underlying distribution. While they can be straightforward and computationally less demanding, their restrictive nature can sometimes limit the ability to capture the underlying complexity in the data.

A more flexible alternative is the Bayesian semi-parametric approach to TTE modeling. This tutorial introduces this methodology, particularly focusing on models that account for time-varying exposure. One such model discussed in detail is the piecewise exponential model, which divides the timeline into intervals and allows the hazard rate to change at each interval boundary. This allows for a more adaptable fit to the data, especially when there are sudden changes or non-uniform patterns in the exposure levels or response over time.

Key Learning Objectives:

• Introduction to ER Analyses in Drug Development: Emphasize the foundational significance of ER analyses in drug development. Understand its crucial role in optimizing drug dosages and predicting patient responses.
• Parametric ER TTE Models: Dive deep into the world of traditional models, appreciating their strengths, limitations, and their cornerstone position in ER modeling.
• Transition to the Bayesian Semi-Parametric Approach: Explore the enhanced flexibility of semi-parametric models, especially concerning time-varying exposure. Special attention will be given to the piecewise exponential model, highlighting its ability to accommodate dynamic changes in hazard rates across specified time intervals.
• Advanced Topics and Techniques: Delve into the more intricate methods, such as the incorporation of Gaussian processes. Understand their potential in capturing complex data patterns and how they can refine and revolutionize traditional ER analyses.

Attendees of this tutorial will benefit from real-world examples and clear explanations, providing a comprehensive understanding of both traditional and advanced methodologies in the field of ER analyses.