A likelihood approach to estimating sensitivity and specificity for binocular data: application in ophthalmology

Area of Science:

• Ophthalmology
• Biostatistics
• Medical Diagnostics

Background:

• Ophthalmology frequently involves evaluating paired eyes, leading to correlated data.
• Standard statistical methods for diagnostic tests often ignore this eye correlation, potentially causing inaccurate sensitivity and specificity estimates.
• Accurate performance metrics are crucial for reliable diagnostic procedures.

Purpose of the Study:

• To develop and present a statistical methodology that accounts for the correlation between paired eyes in diagnostic evaluations.
• To improve the accuracy of sensitivity and specificity estimation for binocular data.
• To provide a robust framework for analyzing paired binary outcomes in ophthalmology.

Main Methods:

• A likelihood-based model was developed for binocular or paired binary outcomes.
• Maximum likelihood estimation was used for model parameter estimation.
• The method was extended to multi-test and multi-disease scenarios and validated with a simulation study.

Main Results:

• The proposed method effectively accounts for eye correlation, yielding more accurate estimates of sensitivity and specificity.
• Simulation studies demonstrated the efficiency of the proposed estimation method.
• The methodology was successfully applied to analyze diabetic retinopathy data.

Conclusions:

• The developed likelihood-based method provides a statistically sound approach for analyzing correlated binocular data in diagnostic testing.
• This method enhances the reliability of sensitivity and specificity estimates in ophthalmological studies.
• The approach is adaptable to complex scenarios, including multiple tests and diseases.