IPM Proj
Storyboard Text
- Hey! You won't believe it! I just got tesed and I'm a Super!
- What??? No way what do you mean tested?
- I asked my doctor to give me this "Basic Super Test" and it came back positive!
- That can't be right. I've known you my entire life and your only super power is overeating...
- Well, the test was positive.. why shouldn't I believe it?
- Let's look at some statistics!...
- Specificity is the measure of how often a test will be negative for a person who does not have the tested condition. A high specificity helps rule in the condition when the test is positive. Specificity = true negatives/ (true negative + false positive).
- Specificity is the measure of how often a test will be negative for a person who does not have the tested condition. A high specificity helps rule in the condition when the test is positive. Specificity = true negatives/ (true negative + false positive).
- Looks like your test had a low specificity of 20%. That is not a reliable value and it is likely a false positive.
- ....That's depressing.
- Guess What?! I did my research and I found an even better test! The test has high specificity! This one says that I'm so super that I'm BARELY EVEN HUMAN!
- Two friends living in a super-powered world!
- Whatttt?
-
- Sensitivity?! You didn't mention that before.
- This test has low sensitivity. It's also not a reliable test. Your result is likely a false negative.
-
- Sensitivity is as important as specificity. Sensitivity is the measure of how often a test will be positive for a person who has the tested condition. A high sensitivity helps rule out the condition when the test is negative. Sensitivity = true positive / (true positive + false negative).
- Sooo...I have to find ANOTHER test now?
-
- Why don't we look for a test that has a good balance of specificity and sensitivity! We can get a good idea of their reliability by investigating the test's PPV & NPV!
- Wait... PPV and NPV? What are those?
-
- PPV is the positive predictive value of a test, it is the probability that people who take a test and get a positive result are actually "true positives". NPV is the negative predictive value of a test. This states the probability that people who take the test and get a negative result are actually "true negatives".
- How do you test for those?!?
-
- To test for PPV use this equation: PPV = true positives/ (true positive + false positives) To test for NPV use this equation: NPV= true negative/ (true negative + false negatives)
- The next day...
- What about this test? The "Modest Hero 2.0"
- An ROC curve plots the sensitivity vs 1-specificity to see that there’s a good sensitivity to specificity ratio for the exam. If we lower the cutoff point too much, it increases the sensitivity and NPV, but at the same time it decreases the specificity and PPV. On the other hand, if we increase the cutoff point, it increases the specificity and PPV, but will decrease the test’s sensitivity and NPV.
-
- A ROC Curve will help us find the perfect balance of specificity and sensitivity. And you're in luck, the "Modest Hero 2.0" looks great!!!
- AWESOMESAUCE!! I'm gonna ask my doctor for that one!
-
-
- So how'd it go?
- Well, you were right after all. I'm not a Super. I have no powers...
-
- Hey, cheer up! You are already pretty awesome. You don't need powers to be great!
- Aww shucks! You know what? Your super-power is being the WORLD'S BEST FRIEND!
- One week later...
- This looks pretty good! Both the PPV and NPV are 95%. We can even look at the ROC Curve to make sure this time that this test works!
-
-
- A few weeks later, some disappointing news arrives...
-
-
- If you say so...
- FIN
-
-
-
-
-
Over 30 Million Storyboards Created