CD12 p value. Thanks tech, ohm, borel
OK, so i will try the CD12 p value again with the wolframalpha two binomial distribution test linked at the bottom of this post.
While I am trained as elec engineer, I am somewhat clueless on these combinatorial probablity problems
So maybe this is close. Or not.
ohm's post says the dexamethasone study mortality rate for patients on oxygen but not ventilated was 25%.
Thus perhaps 25% is a useful metric to use in the wolframalpha two binomial districution test.
25% mortality in placebo arm is about 16 deaths.
If 16 deaths in the Ll arm, that is a massive 50% reduction in mortality.
Running the wolfram alpha calculator for 16 deaths in each of placebo and LL arms yields pvalue of 0.02.
So that's pretty good. If mortality is as unfortunately as high as 25% in the placebo arm seems a pretty fair chance to hit the p value of 0.05.
When placebo has 16 deaths, LL can have 18 and still hit 0.05.
One thing I don't understand here is that TechGuru thinks that since the wolframalpha is a two tailed test. I am not sure if the two is because of two normal distributions (one placebo and one LL) or two tails on the plot (one tail on left side of graph, one tail on right side of graph).
Seems to me that "probablity of this outcome by chance" applies whether one or ten tails, but idk. also think wolframalpha would include any tail thingy in their answer, but idk.
borel thinks its probably not the calculator for the CD12 trial.
No doubt the FDA and CYDY have already agreed on not only the matth equations but also the success and fail p value mortality numbers.
Since the CD12 analysis of 195 patients is an "interim analysis", they probably have to hit a more difficult p value.
Just cuz why not, suppose LL reduces mortality by 75%. Then with 16 placebo deaths there are 8 leronlimab deaths.
Run the calculator and that's a p value of 0.0007.
Were this to occur, this is not a home run. this is knocking the ball out of the country.
UCLA EIND death rate I have read on this board was 2/28. (2/28)*130=9.2
so a massive success is not outside the realm of possibility.
OK, so i will try the CD12 p value again with the wolframalpha two binomial distribution test linked at the bottom of this post.
While I am trained as elec engineer, I am somewhat clueless on these combinatorial probablity problems
So maybe this is close. Or not.
ohm's post says the dexamethasone study mortality rate for patients on oxygen but not ventilated was 25%.
Thus perhaps 25% is a useful metric to use in the wolframalpha two binomial districution test.
25% mortality in placebo arm is about 16 deaths.
If 16 deaths in the Ll arm, that is a massive 50% reduction in mortality.
Running the wolfram alpha calculator for 16 deaths in each of placebo and LL arms yields pvalue of 0.02.
So that's pretty good. If mortality is as unfortunately as high as 25% in the placebo arm seems a pretty fair chance to hit the p value of 0.05.
When placebo has 16 deaths, LL can have 18 and still hit 0.05.
One thing I don't understand here is that TechGuru thinks that since the wolframalpha is a two tailed test. I am not sure if the two is because of two normal distributions (one placebo and one LL) or two tails on the plot (one tail on left side of graph, one tail on right side of graph).
Seems to me that "probablity of this outcome by chance" applies whether one or ten tails, but idk. also think wolframalpha would include any tail thingy in their answer, but idk.
borel thinks its probably not the calculator for the CD12 trial.
No doubt the FDA and CYDY have already agreed on not only the matth equations but also the success and fail p value mortality numbers.
Since the CD12 analysis of 195 patients is an "interim analysis", they probably have to hit a more difficult p value.
Just cuz why not, suppose LL reduces mortality by 75%. Then with 16 placebo deaths there are 8 leronlimab deaths.
Run the calculator and that's a p value of 0.0007.
Were this to occur, this is not a home run. this is knocking the ball out of the country.
UCLA EIND death rate I have read on this board was 2/28. (2/28)*130=9.2
so a massive success is not outside the realm of possibility.
5
