Progress on the CSE diagnostic study. Application of McNemar's test revisited.

Area of Science:

• Medical Diagnostics
• Biostatistics
• Cardiology

Background:

• Comparing diagnostic performance often involves multiple interpretations.
• Traditional McNemar's test has limitations with multiple, equally certain diagnostic statements.
• Accurate statistical reporting is crucial for evaluating diagnostic tools.

Purpose of the Study:

• To present an extension of McNemar's test for comparing diagnostic performance.
• To address limitations in handling multiple, equally certain diagnostic statements.
• To improve the accuracy of sensitivity and specificity statistics.

Main Methods:

• The study extends McNemar's test for analyzing diagnostic statements from ECG interpretations.
• The proposed method assigns partial credit/discredit for correct/incorrect statements when certainty is equal.
• This approach is applied to a pilot database of ECGs with single, documented abnormalities.

Main Results:

• The extended McNemar's test offers a more nuanced evaluation of diagnostic performance.
• It allows for more proper reporting of sensitivity and specificity statistics.
• This method better reflects the accuracy of diagnostic categories with multiple interpretations.

Conclusions:

• The enhanced McNemar's test provides a more statistically sound method for comparing diagnostic performance.
• This statistical advancement is particularly valuable when dealing with multiple, equally certain diagnostic interpretations.
• The improved reporting enhances the evaluation of computer-aided and visual ECG interpretation accuracy.