rmnldwg / lyprox

Web app for exploring patterns and correlations in the lymph node level involvements of head & neck cancer patients.
https://lyprox.org
MIT License
4 stars 4 forks source link

unknowns in combination with OR #19

Closed rmnldwg closed 3 years ago

rmnldwg commented 3 years ago

If a patient has had an FNA, the logical OR introduces a lot of "unknown" labels in the LNLs that were not inspected by the FNA. But "unknown" & "True" for example should always yield "True".

rmnldwg commented 3 years ago

Actually, "unknown" & "True" always yields "True". However, when two diagnoses report "False" (no involvement) and "unknown", then the logical OR in the dashboard displays that as "unknown" too. @unkelbac does that make sense, or should I change it?

rmnldwg commented 3 years ago

Apparently, the numpy functions numpy.any() and numpy.all() are inconsistent. For example:

np.all([None, False])

evaluates to None, while

np.all([False, None])

evaluates to True. I can circumvent this by passing a default array of False values to it. But then we would not observe any None anymore. Not sure if this is desireable...

rmnldwg commented 3 years ago

now FNA shows LNLs that were not checked as healthy. That should be changed