Related Experiment Videos
Local influence analysis when interfacing toxicokinetic and proportional hazard models
Nicola Sartori1, Karl Thomaseth, Alberto Salvan
1Department of Statistics, Ca' Foscari of Venezia, Venice, Italy. sartori@unive.it
Statistics in Medicine
|July 27, 2004
Summary
This study analyzed dioxin exposure and cancer risk using a toxicokinetic model. Results showed no significant local influence on risk estimates from model factors, but identified key influential parameters.
Area of Science:
- Environmental Epidemiology
- Toxicology
- Biostatistics
Background:
- Dioxin exposure is linked to cancer mortality.
- Accurate dose-response analysis requires reliable toxicokinetic modeling.
- Understanding model sensitivity is crucial for risk assessment.
Purpose of the Study:
- To perform a local influence analysis on a dioxin dose-response model.
- To assess the impact of perturbations in toxicokinetic model factors on cancer risk estimates.
- To identify influential factors within the toxicokinetic model.
Main Methods:
- Utilized a proportional hazard model with time-dependent dioxin dose.
- Employed a toxicokinetic model to estimate absorbed dioxin dose.
- Conducted local influence analysis on toxicokinetic model parameters and constants.
Main Results:
- No significant local influence was detected on the overall cancer risk estimates.
- Specific factors within the toxicokinetic model were identified as more influential than others.
- The analysis provides insights into the sensitivity of the risk assessment to model inputs.
Conclusions:
- The current risk estimates for dioxin exposure are robust to minor perturbations in the toxicokinetic model.
- Identifying influential factors allows for targeted refinement of future toxicokinetic models.
- This approach enhances the reliability of environmental health risk assessments.