Issue rvandewater/YAIB#144 was raised in the main YAIB repo, highlighting data inconsistencies between Python and R data preprocessing for the mortality task. This difference is caused by the fact that the Python code did not use a base cohort, restricting the observation time too early. This affects all tasks with a maximum length of 24 hours (i.e., mortality24 and kidney_function).
In addition, some further differences were found in the calcuation of maximum gap between consecutive measurements.
Solution
Introduce a base cohort for mortality24 and kidney_function.
Problem
Issue rvandewater/YAIB#144 was raised in the main YAIB repo, highlighting data inconsistencies between Python and R data preprocessing for the mortality task. This difference is caused by the fact that the Python code did not use a base cohort, restricting the observation time too early. This affects all tasks with a maximum length of 24 hours (i.e.,
mortality24
andkidney_function
).In addition, some further differences were found in the calcuation of maximum gap between consecutive measurements.
Solution
Introduce a base cohort for
mortality24
andkidney_function
.