What does it mean by #sensitivity vs #sensitivity?

what are false positive rate and false negatives?

How are they calculated?

Here's a simple explanation

[Thread] #medtwitter #tweetorial

SENSITIVITY aka true positive rate (TPR) = number of people who have the disease that tests positive

low sensitivity tests will have high false negative rate (FNR). aka type II errors = number of people who have the disease but tests negative

SPECIFICITY aka true negative rate (TNR) = Number of people who do NOT have the disease that tests negative.

low specificity tests will have high false positive rate (FPR). aka type I errors = number of people without disease who tests positive

A good test rarely misses the thing you are looking for (sensitivity) and rarely mistakes it for something else (specificity)

How are they calculated?

Sensitivity : probability (%) that a sample tests positive given the patient has the disease.

Sensitivity = Number of true positive ÷ (true positives + false negative)

or  = Number of true positives
 ÷ Total number of individuals with the illness

Specificity : probability (%) that a sample tests negative given the patient does NOT have the disease.

Specificity = Number of true negatives ÷ (true negatives + false positives)

or = Number of true negatives
 ÷ Total number of individuals without the illness

eg: you have 150 people with disease and 400 without.
You run the samples, you have;

Those with disease testing positive (TP) = 144

Those without disease testing negative (FP) = 12

Those without disease testing negative (TN) = 388

Those with disease testing negative (FN)= 6

Sensitivity = 144 / (144 + 6) = 0.96

Specificity = 388 / (388 + 12) = 0.9

Therefore the test is 96% sensitive and 97% specific

Thats not all..

There's also PPV and NPV that you need to understand, they are NOT the same as sensitivity & specificity although are related.

positive predictive value (PPV) = probability that you have the disease given your test is positive

negative predictive value (NPV) = probability that you DONT have the disease given your test is negative.

This information is more useful for discussing results with a patient.

eg if the PPV of a test is 92%, then if you tested positive, theres a 92% chance that you've got the disease

So basically, the sensitivity and specificity evaluate the test, whereas the PPV and NPV evaluate the results at hand.

how is it calculated?

PPV = true positive ÷ (true positives + false positives)
or = true positives ÷ samples that tested positive

NPV = true negatives ÷ (true negatives + false negatives)
or = true negatives ÷ samples that tested negative

So, how accurate is a test?

Accuracy: ability to differentiate those with and without the disease correctly.

Accuracy= (TP+TN) ÷ (TP+TN+FP+FN)
or = number correctly predicted ÷ total tested