Open vanh17 opened 3 years ago
Hi, I wasn't a part of this piece of work, but I will message Matt to make sure he sees it.
Cheers, Ash
@mattshardlow, another follow-up with the open issue.
I looked at the current MIMIC version here. This one is the most current one and it was released in 2016, which I suspect it was the one used in your paper. However, not all MIMIC .csv files have 11 columns or more. Did it give you any errors during the project or there might be a sublist of the MIMIC dataset that you passed onto the processing script at run-time?
Hi @mattshardlow and @ash-williams,
According to the paper, the code here used MIMIC-III version 1.4, however, when I used the same MIMIC-III version with this java code, it threw out-of-bound errors at two places:
I tried to put in check system in place and ignore lines/files that do not satisfy the two conditions above, i.e. need to have at least 12 elements (index starts at 0) and ignore empty lines. However, all it returned to me were texts from PRESCRIPTION.csv and these texts were just numbers.
Also, when I looked at the code here, the MIMIC-III files are expected to be in .txt format but the downloaded 1.4 version contains only .csv files. Because of this, I was just wondering if you used another preprocessor to convert .csv files into .txt format.
Thank you, @mattshardlow and @ash-williams!