(Total Views: 591)
Posted On: 05/06/2020 10:04:27 AM
Post# of 154082
Baggo-mh,
There are several probability statistical models. The most used are the normal and the binomial distributions. In cases where the outcome is binary (for example success or failure, or live or death) normally the binomial distribution is used.
From a source:
In our case, we are trying to determine the probability of observing 6 survivals in 10 trials (10 patients). In this case, the probability of success is 1-0.88 and the probability of failure is 0.88 for each one (we assume that each patient’s outcome is independent, independent trials).
Once we have the binomial probability distribution, we ask ourselves: how probable is that the extreme results that we are obtaining (the number of saved people) are significant? , This is the null hypothesis test (meaning we assume the drug does not work and the result are not significant).
To quantify the extreme results we take an extreme of the probability curve and integrate a small area (95%) in the right tail meaning that if this number is small there I large evidence that and alternative hypothesis is correct (the drug works). Unfortunately I have not been able to post graphs in this site, otherwise would share with you the probability distribution curve and values I am obtaining.
Or put in simple terms: how possible it is that after having 100’s of people dying at 88% (in New York) we take 10 patients and only 40% die? If this number works out to be less than 0.05 it means that the possibility is very small (and FDA uses this as the threshold) and the drug works.
There are several probability statistical models. The most used are the normal and the binomial distributions. In cases where the outcome is binary (for example success or failure, or live or death) normally the binomial distribution is used.
From a source:
Quote:
Binomial distribution summarizes the number of trials, or observations when each trial has the same probability of attaining one particular value. The binomial distribution determines the probability of observing a specified number of successful outcomes in a specified number of trials
In our case, we are trying to determine the probability of observing 6 survivals in 10 trials (10 patients). In this case, the probability of success is 1-0.88 and the probability of failure is 0.88 for each one (we assume that each patient’s outcome is independent, independent trials).
Once we have the binomial probability distribution, we ask ourselves: how probable is that the extreme results that we are obtaining (the number of saved people) are significant? , This is the null hypothesis test (meaning we assume the drug does not work and the result are not significant).
To quantify the extreme results we take an extreme of the probability curve and integrate a small area (95%) in the right tail meaning that if this number is small there I large evidence that and alternative hypothesis is correct (the drug works). Unfortunately I have not been able to post graphs in this site, otherwise would share with you the probability distribution curve and values I am obtaining.
Or put in simple terms: how possible it is that after having 100’s of people dying at 88% (in New York) we take 10 patients and only 40% die? If this number works out to be less than 0.05 it means that the possibility is very small (and FDA uses this as the threshold) and the drug works.
(4)
(0)
