Binomial distribution sample confidence intervals estimation for positive and negative likelihood ratio medical key...

Area of Science:

• Medical Statistics
• Diagnostic Test Evaluation
• Biostatistics

Background:

• Confidence intervals for likelihood ratios (LRs) are crucial for interpreting diagnostic test results.
• Traditional methods using normal distribution approximation of binomial distribution have known anomalies, particularly in limit cases.
• Improved estimation quality is needed for accurate clinical decision-making based on LRs.

Purpose of the Study:

• To develop and evaluate novel methods for computing confidence intervals of likelihood ratios.
• To compare the performance of these new methods against the classical approach (RPWald).
• To identify the most effective method for accurate LR confidence interval estimation.

Main Methods:

• Four new methods (RPAC, RPAC0, RPAC1, RPAC2) were developed using an exact probability calculation algorithm.
• Methods were implemented in PHP and compared using five confidence interval assessment criteria.
• Experiments were conducted across various sample sizes and binomial variable ranges.

Main Results:

• The proposed RPAC2 method demonstrated the best overall performance in computing confidence intervals for both positive and negative likelihood ratios.
• New methods showed improved accuracy compared to the classical RPWald method, especially in challenging limit cases.
• Experimental validation confirmed the superiority of the RPAC2 method across diverse data scenarios.

Conclusions:

• The RPAC2 method offers a significant improvement for calculating confidence intervals of likelihood ratios.
• This advancement can lead to more reliable diagnostic test evaluations and clinical interpretations.
• The study provides a more robust computational tool for biostatistical analysis in medical research.