Conversation
Notices
-
Embed this notice
remember, a doctor in 2024 is not smart, just a POC that can memorize a few things and show a smug superiority.
-
Embed this notice
When you learn this, you can disarm the mediocre ones quickly, and identify the good ones.
-
Embed this notice
@matty and that is fair enough. But it's a core component of generating a diagnosis.
(Also you should learn it so you understand how medicine works ).
The answer is 2%.
-
Embed this notice
To be fair I have no idea how to calculate that sort of thing either.
-
Embed this notice
@matty The introductory statistics method of bags of marbles is a great way to think about it.
-
Embed this notice
Probability is one of those things that I can never get past the mental block. I don't know why.
-
Embed this notice
@matty Understanding randomness is a very important part of science and technology.
-
Embed this notice
@MeBigbrain @matty @sickburnbro
You split the two groups into "test positive" and "test negative" then, based on rates of false positive/negative, you split those groups into a further two groups
So you have: % positive = % actual positive + % false positive
And: % negative = % actual negative + % false negative
You use the decision tree to calculate total probability of X occurrence, and you use that number as a guide for decision making.
-
Embed this notice
@matty @sickburnbro I feel I have a good grasp of statistics but I don't get this either. 95% seems right to me, 100% of them have a positive diagnosis and then 5% have a false positive.
-
Embed this notice
@pepsi_man @matty @sickburnbro I don't see how that matters. We don't need the 1/1000 number, we've already confirmed the positive diagnosis. It's like if I flip a coin, it comes up heads, and then you say "that only has a 50% chance of being heads." IT'S 100%, WE ALREADY HAVE THE RESULT.
-
Embed this notice
@MeBigbrain @pepsi_man @matty you need it because you don't know if the person has the disease. What you are *attempting* to find out it the probability of the person having the disease, given that they have a positive test result.
-
Embed this notice
@pepsi_man @matty @sickburnbro Why do we need to calculate % negative here? The question states that the patient tested positive, so "% positive" = 100.
-
Embed this notice
@MeBigbrain @matty @sickburnbro
1 out of 1000 will test positive. So. Assume 1000 patients tested. 1 positive (.001%) and 999 (99.999%) negative.
-
Embed this notice
@MeBigbrain @pepsi_man @matty @sickburnbro
-
Embed this notice
@IlDuWuce @monsterislandcolonizer @matty the key is the "assuming you know nothing else"
-
Embed this notice
Ohhhh I get it now. The text is just misleading to me to be very honest. To me it struck as if the 1/1000 patient was tested. I got kind of the same answer doing 0,001/(0,05 . 999), you'll get 2%.
-
Embed this notice
@matty @sickburnbro Explanation
-
Embed this notice
@birdulon @MeBigbrain @matty @pepsi_man in theory yes, in practice, because insurance will deny them.
-
Embed this notice
@birdulon @matty @pepsi_man @sickburnbro Okay I see what you're saying: 5% of people tested generate a false positive result, rather than 5% of positive results are false.
-
Embed this notice
@MeBigbrain @matty @pepsi_man @sickburnbro Yeah, it's why doctors don't like to run exotic tests without probable cause.
-
Embed this notice
@MeBigbrain @matty @pepsi_man @sickburnbro in 1000 people you have 1 true positive and 999*5/100 = 50 false positives
your hypothetical positive-testing patient could be the 1 or among the 50
-
Embed this notice
@MeBigbrain @pepsi_man @matty it sounds like birdulon helped you out though?
-
Embed this notice
@sickburnbro @pepsi_man @matty You and pepsiman are terrible at explaining things. I understand the error but nothing either of you said clarified anything.
-
Embed this notice
Yeah figured as much now. The professor would ask these kind of things and really say that part LOUDLY and then we'd have learned our lesson afterwards.
-
Embed this notice
@IlDuWuce @matty @monsterislandcolonizer its why students used to be taught to read all the instructions before starting the test. Sometimes the end of a question can have a decisive impact on the result, and you can't expect it to be at the front.
-
Embed this notice
@sickburnbro @pepsi_man @matty Yes
-
Embed this notice
@MeBigbrain @pepsi_man @matty what I was trying to highlight was "positive diagnosis". A positive test result and positive diagnosis are not the same thing! And hopefully now it is easy to understand, and can help you when having to deal with healthcare.
-
Embed this notice
@HonkHonkBoom @pepsi_man @matty @sickburnbro OKAY EVERYBODY IN THE THREAD STOP SAYING "FALSE NEGATIVE." THAT WOULD MEAN THAT THEY TESTED NEGATIVE BUT ACTUALLY HAD THE DISEASE. THAT VALUE ISN'T MENTIONED ANYWHERE, AND UNSTATED VARIABLES ARE ASSUMED IRRELEVANT. STOP.
-
Embed this notice
@MeBigbrain @matty @HonkHonkBoom @pepsi_man @sickburnbro I would give extra credit to a student that mentioned "false negative rate omitted, assuming 0%" or acknowledgement to that effect
-
Embed this notice
@MeBigbrain @pepsi_man @matty @sickburnbro no, they didn't say that the 1 out of 1000 that actually had it tested positive, they said he had the disease, if there was an real positive in their sample, 1/1000 prevalence does not guarantee that your group of 1000 has somebody who actually has it, there was no mention of the false negative rate, so, you cannot confidently say 51 positives, it could be 50 false positives and possibly 1 false negative.
-
Embed this notice
@Escoffier @AmonMaritza to your body, little. Because when fasting your body will use the same metabolic pathway to generate ketones to fuel your brain.
-
Embed this notice
I read about a guy that lost a crazy amount of weight on the water and vitamins diet. that’s probably risky, but body fat is literally there to help you through famine.
-
Embed this notice
@AmonMaritza @sickburnbro An interesting question is, to your body, what is the difference between a keto diet and the kind of extended fasting you're talking about?
-
Embed this notice
@sickburnbro I had a lot of interesting conversations with doctors while I was busy losing 300 lbs doing keto.
-
Embed this notice
did they say it shouldn’t be possible?
-
Embed this notice
@AmonMaritza @sickburnbro They said all kinds of things all of which were hilariously wrong. They said that keto would hurt me. It was an unhealthy diet. That I needed to keep eating goyslop.
And a lot of these happened when I was down 200 lbs so I had already lost a considerable amount of weight.
-
Embed this notice
@Escoffier @AmonMaritza this is also why intermittent fasting works.
-
Embed this notice
@MeBigbrain @matty @pepsi_man @petra @sickburnbro
I still don’t understand the Monty Hall problem.
Makes no fucking sense
-
Embed this notice
@Jonny @MeBigbrain @matty @pepsi_man @sickburnbro It only made "sense" to me when I diagrammed out all the possible outcomes before seeing the goat and then the new set of possible outcomes after seeing the goat. I could *see* why the result was what it was.
Then I could write it up in the usual statistical/probability formalism.
But it would never make sense to me without the pictures. I could know the result from moving the math symbols around but I would never have understood it intuitively.
-
Embed this notice
@sickburnbro @MeBigbrain @pepsi_man @matty It's similar to the Monty Hall problem with the goats.
Opening a door and observing the new information changes the set of possible outcomes. But it's tempting to ignore that, which makes for the confusion. It even confused Erdős .
-
Embed this notice
@petra @sickburnbro @pepsi_man @matty Goddamn, the Monty Hall problem. I feel like I'm in that situation again. I didn't get that one immediately either, everyone in my social circle insisted that it was correct but none of them could explain it. That makes me think that they're just repeating what someone else is saying without actually understanding it themselves, while I don't accept something until I'm capable of explaining it.
-
Embed this notice
@Jonny @petra @matty @pepsi_man @sickburnbro I think it's designed to feel like a trick and that gets people. Like if you pick wrong the first time, a scammer would just jump on that and you would lose. Trying to pursuade you to switch feels like you got it right and they're trying to talk you out of it, even though they always do it regardless of if someone picked right the first time or not.
-
Embed this notice
@petra @matty @MeBigbrain @pepsi_man @sickburnbro
Not me.
After the goat shows up, every cell in my brain still says 50-50
-
Embed this notice
@MeBigbrain @Jonny @petra @matty @pepsi_man yes. and the ones that are tricky like this are probably why people tried to formalize the stuff to begin with.
-
Embed this notice
@Jonny @MeBigbrain @matty @pepsi_man @petra @sickburnbro You don't understand the scenario, or the solution?
The explanation is straightforward, just counterintutive to most people's understanding of probability. Probability does not represent a discrete fact about the world, it represents a measure of your knowledge about something's outcome.
If you pick 1 door out of the three, you have a 1/3 chance of getting it right.
If you picked it correctly, then you win, you go home, and there's no problem to solve.
If you did not pick it correctly -- the 2/3 chance -- then you are shown a non-winning door, and asked if you want to change it.
Now, if you are in this situation, which you would be in 2/3 times, the host will now give you a choice to switch to one of the other ones.
Now, you KNOW that the 1 door you picked is wrong. You aren't picking one door out of three, you are picking one door out of two. So your second choice is not to find 1 winner out of 3, it's a choice to find one winner out of the 2 remaining.
Which means that because you now have more information, the chance of either of the remaining two being winners is 50% -- because you've eliminated one of the choices that was wrong.
If the game started by the host simply saying "Pick one of these three doors, the prize can be in any of them -- oh, but we're going to throw out one of the wrong ones, and you can only choose between two, one of which is the winner" -- it would be obvious that it's a 50/50 choice without the third door.
The intuition is that each door has an inherent 1/3 chance that's a property of the door. That's not, that's just a property of your knowledge about the door. Note that the host does not think each door has a 1/3 chance, he knows which door has a 100% chance and which doors have a 0% chance.
-
Embed this notice
@MeBigbrain @sickburnbro @matty
:shrug_yui:
Stats is hard bc it's taught incorrectly. It's not really math, it's just Set Theory with math as a descriptor.
-
Embed this notice
@pepsi_man @matty @sickburnbro I find that most of my problem with statistics is misunderstanding terms, the math is usually easy.
Bread up, Bro
Matty
pepsi_man
Coon Appreciater
Bloodytailspike
Il D~uWu~ce
Birdulon
HonkHonkBoom
AmonMaritza
Escoffier Gab refugee
Jonny
petra
『g r e y』 :uv: ☦️
All GNU social JP content and data are available under the