Statistics
Hemant Kulkarni, Amit Amin, Manju Mamtani, Manik Amin
Lata Medical Research Foundation, Nagpur, India
Interface
Andrew G. Dean, EpiInformatics.com, and Roger Mir

This module helps evaluate the performance of a candidate diagnostic test or screening procedure against a known reference or gold standard. It is assumed that a reference standard (the true diagnosis) exists. The module lets you choose the type of diagnostic test based on the outcome it reports. Then, the module uses pertinent statistics to evaluate the performance of the diagnostic test.

For two or more exposure levels, the program calculates Sensitivity, Specificity, Positive and Negative Predictive Values, Diagnostic Accuracy, Likelihood Ratios, Diagnostic Odds, Cohen's Kappa, Entropy Reduction, and a Bias Index.

If more than two levels of results are specified, a plot (ROC curve) showing proportions of true positives versus false positives is produced.