eICU_DataLoader
	Uses the original .csv files (currently patient.csv, medication.csv, diagnosis.csv, pastHistory.csv, lab.csv, apacheApsVar.csv)
	and builds one dataframe that contains all of the data we want to extract but needs a lot of processing and cleaning
	The consolidated dataframe is saved to memory.

DataProcessor
	Loads the previously built conslidated dataframe and takes care of all of the processing to get to a useable feature set.
	medication.csv, diagnosis.csv and lab.csv are integrated according to an offset threshold, such that only entries with
	offset < offset_threshold are included.
	While processing, the script goes through each hospital for each of these tables and saves respective outputs to memory.
	This has made the runtime much more manageable and the preprocessing more transparent, since the intermediate steps can be 
	easily inspected, however it does produce a considerable amount of files if running on the entire set, just FYI.
	Finally, a fully processed and useable feature matrix is saved to memory which will be used further

DataManager
	This class loads the previously processed and loaded data and prepares it for training.
	It has implemented two sampling strategies (as discussed in the meetings), normalises the features, chooses the n most important
	features through a RandomForest permutation analysis and has funmctions built in to feed the data to the model.

ICD10code_transformer
	This class takes the pre-loaded files and sorts the ICD10 codes into their respective parts.
	(As discussed with Annika)

All other functions and classes in data_management should be helper functions and similar.
Let me know if there are questions regarding these.