Hello all,
I have a protocol where the researchers used prevalence to calculate PPV, NPV, and Accuracy (using the second set of formulas below).
I am considering using the first set of formulas instead. Could you please help me how to figure out which formula is more appropriate?
Note: The researchers provided me with four columns (TP, TN, FP, and FN), indicating that these metrics were identified based on their clinical knowledge, and I can calculate specify and sensitivity directly
formula
Accuracy = (TN + TP)/(TN+TP+FN+FP)
PPV=Positive Predicted Value=TP/(TP+FP)
NPV=Negative Predicted Value=TN/(TN+FN)
--------------------------------------------------------------
PPV=p× Sen / [p ×Sen+ (1-p) ×Spe]
NPV= (1-p) ×Spe / [p×(1-Sen)+(1-p) ×Spe]
Accuracy = (sensitivity) (p) + (specificity) (1 - p)
any input appreciated,
Bahar