Open Side Menu Go to the Top
Register
How to account for probability in this case of chances of disease How to account for probability in this case of chances of disease

08-08-2020 , 08:03 PM
Quote:
Originally Posted by heehaww
I said that without actually trying it, because I felt like circular logic would be involved. But I stand corrected below:
That's valid and obv way more efficient than what I showed.

However:
Great question, and this shows that my suspicion wasn't completely unfounded. What's wrong is that you have to use 1/13 instead of 5/77. The 1/13 only factors in her sibling being asymptomatic. This way, her being asymptomatic isn't redundantly factored in.

1/13/3 / (1 - 1/13/6) = 2/77

Let's bring both solutions up a level of abstraction for a wider view of what's happening.

First way you showed:
P(wife pos) = P(dad pos & wife asy)*P(wife pos | dad pos & wife asy)
P(dad pos & wife asy) is simply 5/77 because her being asy was already factored into the 5/77. This is key.
P(wife pos | dad pos & wife asy) = 2/5 for the reason you said

2nd way:
P(wife pos | asy) = P(pos & asy) / P(asy)
P(pos & asy) = P(dad pos & dad passed it & wife asy)
P(asy) = 1 - P(dad pos & dad passed it & wife symp)

We can't have overlap between "dad pos" and "wife asy/symp". Setting P(dad pos)=5/77 causes overlap, whereas 1/13 avoids it. With 1/13, the "wife asy" part only builds on the info using new info.
Now I get it. Intuitively, I did have a sense that I was using some kind of overlap which wouldn't be allowed, but I couldn't figure out exactly why. I think I was talking myself into reasoning that when we make the calculations for the last scenario (comparing the dad and her) that we're using her info to solve for him, and then once we have the final "most accurate" answer for him, we can then use that to solve for a more accurate answer for her.

But obviously, we're actually solving for her "most correct" answer at the same time.

Still, one strange thing is that it seems like 40% is too high of a chance of her being positive if we knew that he was positive.

Like her chances at birth are 50%, so it's odd to me that by making it to an age where 1/3 of people by this point are already symptomatic, that she's only earned an extra 10% of "equity" to put her at 60%. I would've thought she'd gain 16.7% equity (making through 1/3 of the 1/2 that she's positive) but obviously I get why that isn't so.
How to account for probability in this case of chances of disease Quote
08-09-2020 , 10:16 AM
Quote:
Like her chances at birth are 50%, so it's odd to me that by making it to an age where 1/3 of people by this point are already symptomatic, that she's only earned an extra 10% of "equity" to put her at 60%.
There appears to be a pattern to how much her equity improves.

Clearing the 1/3 hurdle caused the ratio P(pos | dad pos) : P(neg | dad pos) to equal 2:3

If it were a 2/7 hurdle instead, the ratio would be (5/14)/(1-1/7) = 5/12 = 5:7

So start with n:n and if you clear a k/n hurdle, that brings you to (n-k):n

But changing the 50% would change the pattern.
How to account for probability in this case of chances of disease Quote
08-09-2020 , 03:16 PM
Just wanted to say thanks to all who gave input. I'm not exactly sure if the person we're meeting with on Wednesday is actually a geneticist or just a person who conducts the counseling session, but I am definitely concerned that they're just going to tell us that my wife has a 1/4 chance and won't have done any of the legwork to look at the other factors, even though we already told that department everything on a phone consultation we did a couple weeks ago. I guess we'll see.

I'm also pretty concerned that the stats we went by aren't accurate in regards to symptom onset by age. I did a lot of research as far as trying to find studies and stats available on the internet, but just have no way to verify the accuracy, so going by the 80/20 mark for the dad and the 2/3, 1/3 we were using for my wife and her brother are a bit of faith.

I guess worst case scenario, even if we called the dad a 50/50 but we know for a fact that her older brother tested negative, we can bring the dad down to a 1/3, and then put my wife at 1/6. So that would be without factoring in the dad's age, her age, and her younger brother's age.
So we know she's less than 1/6, but no less than 2/77.
How to account for probability in this case of chances of disease Quote
08-13-2020 , 11:41 PM
Update for those interested:

Had the "counseling" phone session yesterday with the geneticist, and it was pretty frustrating. The geneticist hadn't taken any of the factors into consideration statistically, and when I told him that I'd been researching and working with some statisticians to figure out our more accurate odds and if I could run that by him to see if he could verify any of it, he joked that I was welcome to but he wouldn't be able to make heads or tails of anything I told him mathematically. It's above his pay grade, I suppose. And I asked him if there was anyone there whose job more specifically dealt with that kind of stuff he said not really.

I mean, I guess it's my bad for assuming that that was part of his job and I've never dealt with a geneticist before but like what the ****, we could be making a potentially life changing decision here, and there's a big difference between there being a 25% chance and a 2% chance.

Regardless, we're getting the test done (which we were pretty darn sure we were going to do anyway) and they send it in the mail and then she swabs her cheek and then we send it back to them. Still could be 3-4 weeks before we know.

Obviously feeling optimistic but it's been tough since the stakes are so high.

Statistically, the geneticist did point out that another reason to feel optimistic about the dad more likely not being a carrier is that the two people who did have it, his father and his sister, had the onset of symptoms in their early 40s and died in their mid 50s, and there does exist a hereditary correlation in family members of age of onset of symptoms and duration from onset til death.

Unfortunately, even if I were to do a really deep dive into the statistics of correlation, I would have zero clue how to apply it mathematically to his odds.

It seems on the surface that there is a reason for this correlation to be good news and bad news. The good news is that had he been a carrier, his odds would be weighted even higher towards developing it in his early 40s, and since he didn't, it lowers the odds that he was actually a carrier.

The bad news is that this weakens the relevance of my wife and her brother having made it to 34 without symptoms. And these factors helped lower the dad's chances, and therefore hers.

Although I suppose that an age correlation is a net good sign, since even though I don't know how high the correlation % is, clearly the higher the correlation, the better. Like if there was a 90% age correlation, that would drastically reduce the dad's odds. Is this an accurate way of looking at it?
How to account for probability in this case of chances of disease Quote
08-15-2020 , 05:54 PM
Really wishing all the best.

Lack of basic probability knowledge is pretty common even among professionals who are supposed to deal with it pretty frequently. However, I think that making the test is the wisest choice (unless its cost is very very big). It's true that the odds are low in a general sense; nonetheless, I guess your wife is way more likely to have HD than a random guy from the street.
How to account for probability in this case of chances of disease Quote
08-16-2020 , 03:36 PM
Quote:
Originally Posted by nickthegeek
Really wishing all the best.

Lack of basic probability knowledge is pretty common even among professionals who are supposed to deal with it pretty frequently. However, I think that making the test is the wisest choice (unless its cost is very very big). It's true that the odds are low in a general sense; nonetheless, I guess your wife is way more likely to have HD than a random guy from the street.
Yeah HD occurs in like 1 in 10,000 people, definitely lower than her 2.6%.

Nick do you know anything about correlation coefficients?

I found a very in depth study about hereditary correlation of siblings and parents.

According to this study, there is a correlation coefficient of .630 plus/minus .039 between siblings who are the offspring of an affected father. Does this mean anything to you?

Would you know how to apply that figure to more accurately assess the odds that my wife's father was a carrier?
How to account for probability in this case of chances of disease Quote
08-18-2020 , 08:06 AM
The correlation coefficient is definitely a relevant information and one that further reduces the chance that wife has it.

By itself, we don't have much information to treat it correctly; however we might make some educated assumptions. First of all, we need to understand why there is such a correlation. Probably, not every HD carrier transmits the disease at the same rate to their offspring. Some might be "weak" transmitters, while other "strongs". If you are a child of a strong transmitter you have a high probability of having it and vice-versa.

Imagine for instance that strongs are sure to transmit and weaks are sure not to transmit. If the proportion of strongs and weaks are 50/50, a random child of a HD carrier will still have a 50% probability to get HD, since the transmitter category is unknown. However, if both a strong and a weak transmitter have both 2 children, both strong's children will have it and both weak's won't have it. So the correlation. In this extreme scenario, correlation between siblings will be 1; if instead there is no strong or weak transmitter, you end up with a 0 correlation.

So we are somewhere in the middle. We make these hypotheses.

- There are two types of transmitter, strong and weak. Their proportion is 50/50.
- The strongs transmit HD with a p>0.5 probability to their offspring.
- The weaks transmit with 1-p probability.

As you can see, because of symmetry, the probability for a child of a HD carrier to have HD is still 50/50 if the "type" is unknown. Keep also in mind that of course the above is not the only way to include this information, but the simplest one.

It's easy to get that the correlation coefficient rho between siblings under the above is (2*p-1)^2. So, if rho=0.63, we have p = 0.8968627, where p is the probability that a strong transmitter transmits (1-p for a weak one).

So, we have these other inputs to our problem.

- Father has a 50/50 probability of being strong/weak if he has it.
- If father is strong, any child has a p= 0.8968627 of having HD.
- If father is weak, any child has a 1-p chance of having HD.

We end up with these results.

- Father is still about 8.15% of being a carrier.
- However, there is just a 0.47% chance that he is strong. With 7.68% he is weak and with 91.9% he is not a carrier.
- Wife's chance of having it drops to just 0.94%

The Netica file to check the results.


Code:
// ~->[DNET-1]->~

bnet tptHD_v2 {
AutoCompile = TRUE;
autoupdate = TRUE;
whenchanged = 1597752690;

visual V1 {
	defdispform = BELIEFBARS;
	nodelabeling = TITLE;
	NodeMaxNumEntries = 50;
	nodefont = font {shape= "Arial"; size= 9;};
	linkfont = font {shape= "Arial"; size= 9;};
	ShowLinkStrengths = 1;
	windowposn = (22, 22, 1126, 482);
	resolution = 72;
	drawingbounds = (1412, 720);
	showpagebreaks = FALSE;
	usegrid = TRUE;
	gridspace = (6, 6);
	NodeSet Node {BuiltIn = 1; Color = 0x00e1e1e1;};
	NodeSet Nature {BuiltIn = 1; Color = 0x00f8eed2;};
	NodeSet Deterministic {BuiltIn = 1; Color = 0x00d3caa6;};
	NodeSet Finding {BuiltIn = 1; Color = 0x00c8c8c8;};
	NodeSet Constant {BuiltIn = 1; Color = 0x00ffffff;};
	NodeSet ConstantValue {BuiltIn = 1; Color = 0x00ffffb4;};
	NodeSet Utility {BuiltIn = 1; Color = 0x00ffbdbd;};
	NodeSet Decision {BuiltIn = 1; Color = 0x00dee8ff;};
	NodeSet Documentation {BuiltIn = 1; Color = 0x00f0fafa;};
	NodeSet Title {BuiltIn = 1; Color = 0x00ffffff;};
	PrinterSetting A {
		margins = (1270, 1270, 1270, 1270);
		};
	};

node Dad {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = ();
	probs = 
		// yes          no           
		  (0.5,         0.5);
	whenchanged = 1597750747;
	belief = (0.08146863, 0.9185314);
	visual V1 {
		center = (492, 66);
		height = 8;
		};
	};

node TransmitterType {
	discrete = TRUE;
	states = (strong, weak, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Dad);
	probs = 
		// strong       weak         no            // Dad 
		  (0.5,         0.5,         0,            // yes 
		   0,           0,           1);           // no  ;
	whenchanged = 1597750591;
	belief = (0.004655889, 0.07681274, 0.9185314);
	visual V1 {
		center = (492, 168);
		height = 1;
		};
	};

node Brother1 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750770;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (288, 300);
		height = 3;
		};
	};

node Wife {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750696;
	belief = (0.009440463, 0.9905595);
	visual V1 {
		center = (492, 294);
		height = 5;
		};
	};

node Brother2 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750717;
	belief = (0.009440463, 0.9905595);
	visual V1 {
		center = (774, 294);
		height = 2;
		};
	};

node Symptoms_55 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Dad);
	probs = 
		// yes          no            // Dad 
		  (0.8,         0.2,          // yes 
		   0,           1);           // no  ;
	numcases = 1;
	whenchanged = 1597750771;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (762, 66);
		height = 7;
		};
	};

node Symptoms_34_w {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Wife);
	probs = 
		// yes          no            // Wife 
		  (0.3333333,   0.6666667,    // yes  
		   0,           1);           // no   ;
	whenchanged = 1597752690;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (492, 402);
		height = 4;
		};
	};

node Symptoms_34 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Brother2);
	probs = 
		// yes          no            // Brother2 
		  (0.3333333,   0.6666667,    // yes      
		   0,           1);           // no       ;
	whenchanged = 1597752674;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (774, 396);
		height = 6;
		};
	};
ElimOrder = (Symptoms_55, Symptoms_34_w, Symptoms_34, Dad, Brother1, Wife, TransmitterType, Brother2);
};

Last edited by nickthegeek; 08-18-2020 at 08:13 AM.
How to account for probability in this case of chances of disease Quote
08-18-2020 , 01:34 PM
Quote:
Originally Posted by nickthegeek
The correlation coefficient is definitely a relevant information and one that further reduces the chance that wife has it.

By itself, we don't have much information to treat it correctly; however we might make some educated assumptions. First of all, we need to understand why there is such a correlation. Probably, not every HD carrier transmits the disease at the same rate to their offspring. Some might be "weak" transmitters, while other "strongs". If you are a child of a strong transmitter you have a high probability of having it and vice-versa.

Imagine for instance that strongs are sure to transmit and weaks are sure not to transmit. If the proportion of strongs and weaks are 50/50, a random child of a HD carrier will still have a 50% probability to get HD, since the transmitter category is unknown. However, if both a strong and a weak transmitter have both 2 children, both strong's children will have it and both weak's won't have it. So the correlation. In this extreme scenario, correlation between siblings will be 1; if instead there is no strong or weak transmitter, you end up with a 0 correlation.

So we are somewhere in the middle. We make these hypotheses.

- There are two types of transmitter, strong and weak. Their proportion is 50/50.
- The strongs transmit HD with a p>0.5 probability to their offspring.
- The weaks transmit with 1-p probability.

As you can see, because of symmetry, the probability for a child of a HD carrier to have HD is still 50/50 if the "type" is unknown. Keep also in mind that of course the above is not the only way to include this information, but the simplest one.

It's easy to get that the correlation coefficient rho between siblings under the above is (2*p-1)^2. So, if rho=0.63, we have p = 0.8968627, where p is the probability that a strong transmitter transmits (1-p for a weak one).

So, we have these other inputs to our problem.

- Father has a 50/50 probability of being strong/weak if he has it.
- If father is strong, any child has a p= 0.8968627 of having HD.
- If father is weak, any child has a 1-p chance of having HD.

We end up with these results.

- Father is still about 8.15% of being a carrier.
- However, there is just a 0.47% chance that he is strong. With 7.68% he is weak and with 91.9% he is not a carrier.
- Wife's chance of having it drops to just 0.94%

The Netica file to check the results.


Code:
// ~->[DNET-1]->~

bnet tptHD_v2 {
AutoCompile = TRUE;
autoupdate = TRUE;
whenchanged = 1597752690;

visual V1 {
	defdispform = BELIEFBARS;
	nodelabeling = TITLE;
	NodeMaxNumEntries = 50;
	nodefont = font {shape= "Arial"; size= 9;};
	linkfont = font {shape= "Arial"; size= 9;};
	ShowLinkStrengths = 1;
	windowposn = (22, 22, 1126, 482);
	resolution = 72;
	drawingbounds = (1412, 720);
	showpagebreaks = FALSE;
	usegrid = TRUE;
	gridspace = (6, 6);
	NodeSet Node {BuiltIn = 1; Color = 0x00e1e1e1;};
	NodeSet Nature {BuiltIn = 1; Color = 0x00f8eed2;};
	NodeSet Deterministic {BuiltIn = 1; Color = 0x00d3caa6;};
	NodeSet Finding {BuiltIn = 1; Color = 0x00c8c8c8;};
	NodeSet Constant {BuiltIn = 1; Color = 0x00ffffff;};
	NodeSet ConstantValue {BuiltIn = 1; Color = 0x00ffffb4;};
	NodeSet Utility {BuiltIn = 1; Color = 0x00ffbdbd;};
	NodeSet Decision {BuiltIn = 1; Color = 0x00dee8ff;};
	NodeSet Documentation {BuiltIn = 1; Color = 0x00f0fafa;};
	NodeSet Title {BuiltIn = 1; Color = 0x00ffffff;};
	PrinterSetting A {
		margins = (1270, 1270, 1270, 1270);
		};
	};

node Dad {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = ();
	probs = 
		// yes          no           
		  (0.5,         0.5);
	whenchanged = 1597750747;
	belief = (0.08146863, 0.9185314);
	visual V1 {
		center = (492, 66);
		height = 8;
		};
	};

node TransmitterType {
	discrete = TRUE;
	states = (strong, weak, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Dad);
	probs = 
		// strong       weak         no            // Dad 
		  (0.5,         0.5,         0,            // yes 
		   0,           0,           1);           // no  ;
	whenchanged = 1597750591;
	belief = (0.004655889, 0.07681274, 0.9185314);
	visual V1 {
		center = (492, 168);
		height = 1;
		};
	};

node Brother1 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750770;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (288, 300);
		height = 3;
		};
	};

node Wife {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750696;
	belief = (0.009440463, 0.9905595);
	visual V1 {
		center = (492, 294);
		height = 5;
		};
	};

node Brother2 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (TransmitterType);
	probs = 
		// yes          no            // TransmitterType 
		  (0.8968627,   0.1031373,    // strong          
		   0.1031373,   0.8968627,    // weak            
		   0,           1);           // no              ;
	whenchanged = 1597750717;
	belief = (0.009440463, 0.9905595);
	visual V1 {
		center = (774, 294);
		height = 2;
		};
	};

node Symptoms_55 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Dad);
	probs = 
		// yes          no            // Dad 
		  (0.8,         0.2,          // yes 
		   0,           1);           // no  ;
	numcases = 1;
	whenchanged = 1597750771;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (762, 66);
		height = 7;
		};
	};

node Symptoms_34_w {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Wife);
	probs = 
		// yes          no            // Wife 
		  (0.3333333,   0.6666667,    // yes  
		   0,           1);           // no   ;
	whenchanged = 1597752690;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (492, 402);
		height = 4;
		};
	};

node Symptoms_34 {
	discrete = TRUE;
	states = (yes, no);
	kind = NATURE;
	chance = CHANCE;
	parents = (Brother2);
	probs = 
		// yes          no            // Brother2 
		  (0.3333333,   0.6666667,    // yes      
		   0,           1);           // no       ;
	whenchanged = 1597752674;
	evidence = no;
	belief = (0, 1);
	visual V1 {
		center = (774, 396);
		height = 6;
		};
	};
ElimOrder = (Symptoms_55, Symptoms_34_w, Symptoms_34, Dad, Brother1, Wife, TransmitterType, Brother2);
};
Nick I want to thank you so much for doing all that work and responding, but unfortunately the correlation coefficient numbers I quoted were not in reference to testing positive. They were in reference to the correlation of age of onset of symptoms.

This is why it's not so cut and dry, it's saying that if a one sibling developed symptoms in say, their early 40s, the correlation coefficient of their sibling also developing the symptoms in their early 40s is .630 +/-.039.

I'm trying to factor this in to understanding just how much better of a sign it is that my wife's father was not a carrier, given that he made it to 55 without any symptoms, while his sister and his father both developed symptoms in their early 40s.

I'm very sorry for the misunderstanding. My post a few back discussed this, but I'm now realizing that my more recent post which actually listed the math figures did not mention it.

Nick is there any chance I could send you the link to the study where I got these figures, just so there's no chance I'm misinterpreting the stats?
How to account for probability in this case of chances of disease Quote
08-25-2020 , 02:36 PM
I had an idea, but it relied on the distribution of ages of onset to be Normal (in this case, bivariate Normal). That doesn't appear to be the case, or I have conflicting data.

DD, you said age of onset is 80% likely to be from ages 30-50, and 33% likely to be before one turns 34. If the mean is 40, the first % implies a standard deviation of 7.8 years, but that in turn would imply that P(onset before 34) is only 22%. So if both of your %'s are right, it's either not a Normal distribution or the mean is less than 40.

I interpret the paper's Table 1 to be saying that 40 would be the mean age of your FIL's onset, but that 20 would be the mean age for your wife & her brother. I could easily be reading it wrong.

I imagine there's a way to use the correlation coef without knowing the distribution, but I don't know it, sorry.
How to account for probability in this case of chances of disease Quote
08-26-2020 , 04:28 PM
Quote:
Originally Posted by heehaww
I had an idea, but it relied on the distribution of ages of onset to be Normal (in this case, bivariate Normal). That doesn't appear to be the case, or I have conflicting data.

DD, you said age of onset is 80% likely to be from ages 30-50, and 33% likely to be before one turns 34. If the mean is 40, the first % implies a standard deviation of 7.8 years, but that in turn would imply that P(onset before 34) is only 22%. So if both of your %'s are right, it's either not a Normal distribution or the mean is less than 40.

I interpret the paper's Table 1 to be saying that 40 would be the mean age of your FIL's onset, but that 20 would be the mean age for your wife & her brother. I could easily be reading it wrong.

I imagine there's a way to use the correlation coef without knowing the distribution, but I don't know it, sorry.
No, it's not 80% likely between 30 and 50. It's 80% likely to happen BY 50, so that also includes cases that happen before age 30. What I'd said was that only 20% of cases begin AFTER 50. The break down of cases is:

Under 20 yrs: 10%
20-30- 20%
30-50 - 50%
50+ - 20%

Couldn't find much in the way of distribution within 30-50 years old, it seemed fairly evenly distributed. So my estimate of 33% for my 34 year old wife was based on the 30% of cases under 30, plus a few percentage points.

Considering that her father made it to 55, I suppose he's in even less than the 20th percentile which is good, but also, the stats from under 20 yrs old is actually 7-10%, so I'm adding a little extra equity by assuming 10% there.

So I figured it was a wash, besides there's so much case by case variance anyway.

But all of this is a separate issue from how to then factor in the correlation of onset age among relatives.
How to account for probability in this case of chances of disease Quote
08-26-2020 , 05:51 PM
Quote:
It's 80% likely to happen BY 50
Right my bad.

Quote:
Under 20 yrs: 10%
20-30- 20%
30-50 - 50%
50+ - 20%
So then the distribution may still be Gaussian-like, but with a fatter left tail that's chopped off at 0. I'll take a rough approach that I'm sure isn't the best:

1. Pretend it's perfectly Gaussian, centered at 40 with a standard deviation of 15.
2. Wife & BIL then form a bivariate Gaussian distribution, but shaped by the .63 correlation.
3. Adjust the weights of the <30 and >50 ages to better reflect the real distribution.
4. Use the new distribution to recalculate the chance of reaching a certain age symptom-free.
5. Tweak the previous Netica file accordingly (or redo my calculation from Post #17).

Stay tuned, unless someone chimes in with a less hacky method lol

Edit: oh I just noticed the appendix and it looks like the quoted percentages are compiled from Appendix II Table A. The distribution is bimodal, but the modes are close together so maybe I can still go through with my recipe above...idk there might be no point now, because at this point I'll have zero confidence in whatever final probability I come up with.

Last edited by heehaww; 08-26-2020 at 06:11 PM.
How to account for probability in this case of chances of disease Quote
08-26-2020 , 10:03 PM
I have zero clue about Gaussian, bi-modal, etc. but if you have a game plan, I fully appreciate you running the numbers. (And yes, I recognize they may not even be all that accurate, it's ok)

As far as updates with the sitch, I'm straight up livid that it's been two weeks since our consultation with the geneticist, and we are STILL waiting to receive the test in the mail that he said he'd send out. This also includes two emails that we sent, one a week after and one a few days after that, asking to know where the **** the test was and how come it's taking so long to recieve.

We also had to follow up with them to find out the results of other genetics tests we had performed that were supposed to be posted on her Kaiser account page within a certain amount of time and still haven't been.

We did get the results back when my wife went in yesterday for her next ultrasound, and everything's negative which is a relief, so this Huntington's test is the last thing needed to take care of before we can officially breathe a sigh of relief, but the way the hospital has really dropped the ball on their responsibilities has been pretty infuriating.
How to account for probability in this case of chances of disease Quote

      
m