Here you can find an example analysis on the open FDZ Data:
A very powerful tool of the XOE is the Relative Time Axis !
It consists of
-
defining a reference point to which we would like to compute the time distance to
-
selecting the kind of events we would like to have the time distance to the reference point
For example, we might choose the Diagnosis “Alzheimer’s Disease” and we would like to compute for each diagnosis the time distance to the first Alzheimer’s diagnosis of the corresponding patient, if it exists.
One important thing to note that this kind of analysis is much more performant when data is stored as objects. Having to look for all the relevant information in a relational database is extremely tedious and inefficient!
Why this kind of analysis is important - the temporal relation of events plays a very important role in Causal Discovery! Causes for an effect tend to happen more frequently just before the target and less frequently long before or after it. Analyzing the temporal distribution of events therefore hints at possibly interesting cause-effect relations.
Now the great part about the Relative Time Axis is that it can be defined with floating semantics / additional selection sets . This means that we can treat the “reference point” as a variable, which can be changed by selecting it via another selection set. Per default, in the prepared analysis “Alzheimer’s Disease” is selected, but if you are interested you could always change the selection of the red selection set and therefore change the semantic of the relative time axis. You will have defined a new reference point.
Finally, once a reference point is fixed, we can select some specific kind of diagnosis with the global selections in yellow - and observe how the temporal distribution changes. In particular, potential causes for the effect / reference point will have a peak shortly before the effect.
Simply load the analysis from the home dashboard, read the description texts and just make some selections to observe how the chart responds!
Click to launch the FDZ data model.
User : user
PWD : xplainData