There is also opportunity for local optimization of this hypothesis by
adding nodes. Since the probability of low LVF is higher than the
probability of it being false, the program explores whether this leads
to a better hypothesis. The highest probability scenario corresponds to
the time intervals in the hypothesis with high LAP corrected by
nitroglycerin. The primary difference is that there are now two causes
for PC and the probability of PC is ,
using the default combination rule. Thus, if those two nodes are added
to the pneumonia hypothesis, the probability becomes 0.0029.
Thus, the HDP is able to rank order the hypotheses and say that the most likely hypothesis is that the nitroglycerin corrected the high LAP. About a third as likely is the hypothesis that the low LVF normalized. About 1.5%as likely is that pneumonia was at least partially the cause for the pulmonary congestion.