The supplementary materials are organized as follows.

0_readme.txt: 
	This file.

1_appendix.pdf:
	A single pdf that contains all appendices of the main text. The appendices are organized as follows:
		A. Related Work
		B. Experiments
			B.1 Description of Datasets
			B.2 Supplement Experiment Results
			B.3 Detailed Training Settings
		C. EMFlow
		D. End-to-end Training
		E. Proofs of Theorem 4.1
		F. Extension to AM Metric
		C. Precision and Recall

2_codes:
	A data folder that contains all necessary codes to implement our proposed methods.
		- classifier.py # classifier module
		- imputation.py # imputer module
		- rl.py # reinforcement learning module
		- train.py # training script
		- meta_train.py # a wrapper script for training
		- my_result_writer.py # a helper function for logging
		- flow_models.py # MLP function for emFlow
		- nflow.py # flow model of the imputater

Due to HIPAA regulations, we are not able to publish the Ferritin dataset used in our study. For the accesses of AKI and Sepsis datasets, please refer https://physionet.org/content/mimiciii/1.4/.