You have looked at predicting the seriousness of adverse
events. To explore additional exercises, complete the following steps:
-
You might
want to look at the types of adverse events that occur. Try the following:
-
See if you can use the COSTART
analysis to predict the clusters that you obtained from analyzing
the SYMPTOM_TEXT variable. You can do this with the Cluster node.
To combine variables together, you might want to try a Decision Tree
node.
-
The original data contains other
variables, such as medications and lab tests. You know that the type
of adverse event is affected by drug interactions. Using the original
data, see if you can text mine the medications field to roll up variables
for the medications that patients are currently taking. Then use these
variables to try to predict the clusters that you obtained for the
SYMPTOM_TEXT variable.
-
If you have
access to a MedDRA program, run the text through that and perform
the same tasks with the MedDRA results that you did with the COSTART
terms in this book.
Now that
you have completed this example, you are ready to learn about the
three new nodes in Text Miner 4.2: the Text Parsing,
Text Filter, and Text Topic nodes.