Ask Experts Questions for FREE Help !
Ask
    oxitt's Avatar
    oxitt Posts: 1, Reputation: 1
    New Member
     
    #1

    Sep 19, 2011, 12:08 PM
    Probability calculation based on Bayes Rule
    I have two independent tests, A and B where A+ denotes the test outcome is positive, A- denotes the test outcome is negative. Random variable x denotes that the patient has AIDS. X+ denotes the patient has AIDS, x- denotes the patient doesn't have AIDS.

    We have the following probabilities:
    P(A-|x-) = 0.81
    P(A-|x+) = 0.20
    P(A+|x-) = 0.19
    P(A+|x+) = 0.80

    P(B-|x-) = 0.88
    P(B-|x+) = 0.03
    P(B+|x-) = 0.12
    P(B+|x+) = 0.97

    And prior probability P(x+) = 0.003

    How can I calculate P(x+|A+, B+)? I know how to calculate it for P(x+|A+) but now there are two different tests and I am confused about the calculation.
    jcaron2's Avatar
    jcaron2 Posts: 986, Reputation: 204
    Senior Member
     
    #2

    Sep 19, 2011, 08:51 PM
    If you write out Bayes rule for this case you get the following:

    P(x+|A+, B+) = P(A+, B+|x+)P(x+)/P(A+, B+)

    Since A and B are independent, that should mean

    P(A+, B+|x+) = P(A+|x+)P(B+|x+)

    and

    P(A+, B+) = P(A+)P(B+)

    To compute P(A+), we use the marginal probability:

    P(A+) = P(A+|x+)P(x+) + P(A+|x-)P(x-)

    and likewise for P(B+), where P(x-) = 1 - P(x+)

    So putting it all together, we get

    P(x+|A+, B+) = P(A+|x+)P(B+|x+)P(x+)/[(P(A+|x+)P(x+) + P(A+|x-)(1-P(x+))(P(B+|x+)P(x+) + P(B+|x-)(1-P(x+)))]

    P(x+|A+, B+) = (0.8)(0.97)(0.003)/[((0.80)(0.003) + (0.19)(0.997))((0.97)(0.003) + (0.12)(0.997))]

    P(x+|A+, B+) = (0.002328)/[(0.19183)(0.12255)] = 0.099

    That says that there's only about a 10% chance of x+ even if both tests come back positive. That seems like a really inaccurate pair of tests, but based on the numbers you gave, there's around a 2.3% chance of getting two false positives, whereas only around 0.3% of test-takers are actually positive based on prior probability. So a ~90% false positive rate seems like a reasonable answer.


Not your question? Ask your question View similar questions

 

Question Tools Search this Question
Search this Question:

Advanced Search

Add your answer here.


Check out some similar questions!

Geometric probability, conditional probability with area of square [ 2 Answers ]

Chose a point at random in a square with sides 0<x<1 and 0<y<1. Let X be the x coordinate and Y be the y coordinate of the point chosen. Find the conditional probability P(y<1/2 / y>x).

Functional-based and activity based costing [ 3 Answers ]

Criado, My assignment for this week is way above my head, the textbook is not of much help and I need some guidance as I go along. Below is the assignment. I hope you can help me. pmonica Trinity Clinic has identified three activities for daily maternity care: occupancy and feeding, nursing...

Odds calculation in statistics given probability [ 1 Answers ]

Hello, I have a question that I am having difficulty understanding. The question is If given the probability of being hired is .20, calculate the odds that you will be hired? My thought at first was to calculate the odds by using the probability of .20 and the complement .80 (chance of not...

Painting Cabinets - Oil based or Water based? [ 1 Answers ]

Going to be painting my oak kitchen cabinets white. I heard that oil based paint would be best, but don't want to deal with the extra mess. Is there anything wrong with using a good acrylic enamel?


View more questions Search