What if I have different information from the Bayes Theorm diseases question?

59 Views Asked by At

I'm studying Bayes' Rule and came across the diseases problem. And wonder if it also works in other circumstances.

Let us say $D$ is the event having the diseases, $T$ is the event testing positive for the diseases and $N$ is the event testing negative . Usually, we will be given with the probability of $P(D)$, $P(T|D)$ and $P(N|\overline{D})$. So we can figure out $P(D|T) = \frac{P(T|D)P(D)}{P(T|D)P(D)+P(T|\overline{D})P(\overline{D})}$.

But What if I know $P(D)$, $P(D|T)$ and $P(\overline{D}|N)$, and trying to figure out $P(T|D)$. Is that possible? So far I can only get $$P(T|D) = \frac{P(\overline{D}|T)[P(\overline{D}|T)P(T))+P(D|T)P(T)]}{P(D)} = \frac{P(\overline{D}|T)[P(T|\overline{D})P(\overline{D})+P(T|D)P(D)]}{P(D)}$$ It seems like if I don't know $P(T)$ or $P(T|D)$ itself I can' find the answer. I wonder if there is a way to figure this out knowing the Bayes' Rule. Thank you.

1

There are 1 best solutions below

0
On BEST ANSWER

IF $T$ and $D$ are independent with $P(D)=P(D\mid T)= 1-P(\overline{D}\mid N)$ then you are stuck. But assuming the test is not completely useless then you can say:

$P(D) = P(D\mid T) P(T) +P(D\mid N)P(N) = P(D\mid T) P(T) +(1-P(\overline D\mid N))(1-P(T))$

so $P(T)=\dfrac{P(D) -1+P(\overline D\mid N)}{ P(D\mid T) -1+P(\overline D\mid N)}$

and $P(T\mid D) = \dfrac{P(D\mid T) P(T)}{P(D)} = \dfrac{P(D\mid T)\,(P(D) -1+P(\overline D\mid N))}{P(D)\, (P(D\mid T) -1+P(\overline D\mid N))}$