Module 1 Flashcards
What is a population and what is a sample?
What is a likelihood function?
Gives the probability of the observed outcome for a particular value of the unknown truth; it is the measure of quantitative evidence about that truth
What does evidence do?
Updates your prior beliefs
Explain Bayes’ Theorem.
Bayes’ Theorem allows us to calculate PPV and NPV based on sensitivity, specificity, and prevalence of disease.
Posterior odds = (prior odds) * (likelihood ratio)
The image is just for finding PPV I think but for NPV is similar.
What are odds?
Odds are a way to express the likelihood that an event occurs.
If odds >1, then the top event (H2) is more likely
If odds <1, then the bottom event (H1) is more likely
How is science the ‘search for truth?’
Every scientific experiment brings you closer to the truth. But remember, an experiment doesn’t necessarily reveal the whole truth, it just brings you closer to the truth
What is frequentist probability?
long term relative frequency; Repeat the same experiment over and over and look at the proportion of times the event happens
EX: coin tosses (probability coin lands “hands”), disease prevalence (probability a randomly selected person has the disease)
What are the two types of probability?
Frequentist and bayesian (subjectivist)
What is bayesian probability?
measure of personal belief
EX: sports outcomes (probability Baltimore Orioles win next game), personal risk (probability I contract Covid-19 at grocery store)
When are events considered to be mutually exclusive?
if they cannot occur at the same time
EX: On a coin toss, heads and tails are mutually exclusive
P(A or B) = P(A) + P(B) if A and B are mutually exclusive
When are events considered to be independent?
If knowing whether one occurred tells you nothing about whether the other one occurred.
EX: On two tosses, the result of the first toss is independent of the result of the second toss
P(A and B) = P(a)*P(B) if A and B are independent
What is a joint distribution?
They describe how the outcomes of two experiments behave together. We summarize joint behavior in a two-way contingency table.
What is joint probability?
The probability of two outcomes from two different experiments occurring at the same time
What is a marginal probability?
(used with joint distribution)
The probability of having an outcome in one experiment without caring about the other experiment.
What is relative risk (risk ratio)?
A means of comparing conditional probabilities.
What is conditional probability?
the probability of an outcome in one experiment given the outcome of another experiment
How do we interpret relative risk (risk ratio)? (say your RR is between 0 and 1, between 1 and 2, anf then greater than or equal to 2)
What is sensitivity?
probability a person with disease tests positive (fraction of true positives, a property of the screening test)
What is specificity?
probability a disease-free person tests negative (fraction of true negatives, a property of the screening test)
What is positive predictive value?
probability a person who tests positive has the disease (properties of the disease state)
On the image, that is a very low PPV given the high sens and spec for the test indicating a high positive rate. We interpret as if a fetus tests positive, there’s a low probability that they have PCKD (warrants further diagnostic testing).
What is negative predictive value?
probability a person who tests negative is disease-free (properties of the disease state)
On the image this is a high NPV, so if the fetus tests negative then there’s a high probability that they do not have PCKD.
How do you fill in a 2x2 table given sensitivity, specificity, and prevalence? And then calculate the PPV.
How do you calculate sensitivity?
P(T+|D+)
How do you calculate specificity?
P(D-|T-)
How do you calculate PPV?
P(D+|T+)
How do you calculate NPV?
P(D-|T-)
How do you calculate PPV and NPV with Bayes Theorem?
