A Software Tool for Estimating Uncertainty of Bayesian Posterior Probability for Disease

Area of Science:

• Medical Diagnostics
• Biostatistics
• Health Informatics

Background:

• Traditional medical diagnosis relies on fixed criteria and thresholds.
• Bayesian inference offers a probabilistic framework for advanced diagnostic analysis.
• Quantifying uncertainty in Bayesian diagnosis is an underexplored but crucial area.

Purpose of the Study:

• To introduce a specialized software tool for quantifying uncertainty in Bayesian diagnosis.
• To facilitate the estimation and graphical representation of uncertainty in posterior probability for disease.
• To address the minimal exploration of uncertainty quantification in this diagnostic field.

Main Methods:

• Developed a freely available software tool utilizing uncertainty propagation techniques.
• Incorporated two primary modules and fifteen submodules for comprehensive analysis.
• Estimated sampling, measurement, and combined uncertainty of posterior disease probability.

Main Results:

• The software tool enables estimation and graphical representation of uncertainty in posterior probability.
• It incorporates key parameters like mean, standard deviation, sample size, and measurement uncertainty.
• Demonstrated practical application using fasting plasma glucose data for diabetes mellitus diagnosis.

Conclusions:

• The developed software tool effectively quantifies uncertainty in Bayesian diagnostic models.
• This advancement supports more robust and reliable disease probability estimations.
• Enhances diagnostic accuracy and patient care through probabilistic analysis.