Particle and microorganism enumeration data: enabling quantitative rigor and judicious interpretation.

Area of Science:

• Microbiology
• Analytical Chemistry
• Statistics

Background:

• Microscopic particle and microorganism quantification often relies on enumeration methods.
• These methods are prone to significant variability stemming from sampling errors and analytical recovery issues.
• Existing statistical methods, like t-distribution, are often unsuitable for analyzing enumeration data due to recovery correction needs and non-normal distribution characteristics.

Purpose of the Study:

• To develop probabilistic models that accurately account for random errors in enumeration data.
• To address variability arising from sampling, non-constant analytical recovery, and counting errors.
• To provide a rigorous framework for analyzing (bio)particle enumeration data, including uncertainty quantification.

Main Methods:

• Development of two probabilistic models to capture enumeration variability.
• Application of Bayes' theorem for uncertainty quantification.
• Utilized numerical integration and Gibbs sampling to derive posterior distributions.

Main Results:

• The developed models successfully quantify uncertainty in particle concentration and log(10)-reduction estimates.
• The approach accounts for sampling error assumptions, non-constant analytical recovery, and counting errors.
• Posterior distributions provide a complete uncertainty measure based on experimental data and recovery variability parameters.

Conclusions:

• The probabilistic modeling approach offers a more accurate and rigorous analysis of (bio)particle enumeration data.
• This method improves the reliability of concentration and reduction estimates by incorporating stochastic processes.
• The framework is adaptable for analyzing both single and replicate enumeration datasets.