Evaluating diagnostic performance of clinical tests by spreadsheet modeling. Bayesian analysis using Ri/Cj ratio as a...

Area of Science:

• Medical Diagnostics
• Biostatistics
• Clinical Epidemiology

Background:

• Evaluating the performance of clinical diagnostic tests is crucial for accurate patient care.
• Existing methods may not comprehensively address scenarios with multiple test results and clinical states.
• A unified framework for diagnostic test evaluation is needed.

Purpose of the Study:

• To present a general spreadsheet model for evaluating diagnostic performance of clinical tests.
• To introduce the Ri/Cj ratio as a core analytical tool.
• To demonstrate the model's applicability to various matrix configurations and Bayesian analyses.

Main Methods:

• Developed a general spreadsheet model representing test results as an r x c matrix.
• Defined the Ri/Cj ratio: number of subjects with result Ri in state Cj divided by total subjects in state Cj.
• Identified and analyzed special cases: 2 x c, r x 2, and 2 x 2 matrices.

Main Results:

• The Ri/Cj ratio provides a method for graphic analysis of multiple test results across multiple clinical states.
• The model accommodates various combinations of test results and clinical states.
• The Ri/Cj ratio serves as a general approach for Bayesian analysis, encompassing likelihood ratios, ROC analysis, sensitivity, and specificity.

Conclusions:

• The proposed spreadsheet model and Ri/Cj ratio offer a versatile tool for diagnostic test performance evaluation.
• This approach facilitates a more comprehensive understanding of test accuracy in complex clinical scenarios.
• The model provides a unified framework for various established diagnostic performance metrics.