Single toxin dose-response models revisited

Area of Science:

• Environmental toxicology
• Quantitative structure-activity relationship (QSAR) modeling

Background:

• Dose-response relationships are fundamental in toxicology for assessing chemical risk.
• Sigmoid curves commonly describe toxin effects, but precise analysis of their phases and statistical modeling requires refinement.

Purpose of the Study:

• To provide a rigorous analysis of sigmoid dose-response relationships for single toxins.
• To define and analyze special points on the dose-response curve, including maximum efficacy and inflection points.
• To introduce and evaluate probabilistic models and statistical estimation methods for toxicity data.

Main Methods:

• Introduction of a toxin efficacy function and definition of four special points on the dose-response curve.
• Probabilistic interpretation and mathematical analysis of Hill, logit, probit, and Weibull models.
• Development of multi-target hit and non-zero natural mortality models.
• Statistical estimation using generalized linear models with binomial distribution for mortality counts.

Main Results:

• The special points define three distinct phases of toxin effect on mortality: low mortality at low/high concentrations and high mortality at intermediate concentrations.
• Probabilistic and mathematical analyses were performed for four standard dose-response models.
• Generalized linear models provided a robust statistical framework for estimating toxicity parameters.

Conclusions:

• The defined special points and phases offer a structured understanding of single toxin dose-response.
• The proposed modeling extensions (multi-target, natural mortality) enhance the applicability of dose-response analysis.
• Statistical estimation using generalized linear models is recommended over nonlinear regression for accurate toxicity assessment.