A Nonmechanistic Parametric Modeling Approach for Benchmark Dose Estimation of Event-Time Data

Area of Science:

• Toxicology
• Biostatistics
• Risk Assessment

Background:

• Traditional risk assessment often relies on end-of-study summary data.
• Complex models can be challenging to estimate and interpret.
• Robust methods are needed for accurate benchmark dose estimation.

Purpose of the Study:

• To propose a novel two-step benchmark dose estimation approach for event-time data.
• To provide a robust and less complex alternative to existing methods.
• To demonstrate the utility and power of the proposed method.

Main Methods:

• A two-step approach involving an event-time model and a dose-response model.
• Step 1: Fit an event-time model to describe event probability over time for each dose.
• Step 2: Fit a dose-response model to estimates (e.g., t50) from Step 1 to derive benchmark dose.

Main Results:

• The proposed method successfully estimates benchmark dose from event-time data.
• Demonstrated application in two distinct examples.
• The time-to-event model analysis showed increased statistical power compared to traditional methods.

Conclusions:

• The novel two-step approach provides a robust method for benchmark dose estimation.
• This method simplifies complex modeling while maintaining accuracy.
• The approach enhances statistical power in risk assessment analyses.