I think what you’re most interested in is P(B3=stillborn | B1=toterm, B2=stillborn). You’re looking at the joint distribution of trials.
At the most basic level of analysis, I’d want to estimate the cooccurence statistic above. Probably you can’t get exactly that. But there certainly should be P(B1=stillborn), hopefully somewhere P(B2=stillborn2| B1=stillborn), maybe P(B2=stillborn|B1=toterm). With those, I think you can get the maxent estimate for your desired distribution above.
With that, you could start to do the adjustments based on particular causes that can be ruled out (for not having the problem SNP, for example), but I’d think those are more second order effects compared to base rate issues.
My condolences on your loss.
I think what you’re most interested in is P(B3=stillborn | B1=toterm, B2=stillborn). You’re looking at the joint distribution of trials.
At the most basic level of analysis, I’d want to estimate the cooccurence statistic above. Probably you can’t get exactly that. But there certainly should be P(B1=stillborn), hopefully somewhere P(B2=stillborn2| B1=stillborn), maybe P(B2=stillborn|B1=toterm). With those, I think you can get the maxent estimate for your desired distribution above.
With that, you could start to do the adjustments based on particular causes that can be ruled out (for not having the problem SNP, for example), but I’d think those are more second order effects compared to base rate issues.