Predicting Stroke from Electronic Health Records

With the spirit of reproducible research, this repository contains all the codes required to produce the results in the manuscript:

C. S. Nwosu*, S. Dev*, P. Bhardwaj, B. Veeravalli and D. John, Predicting Stroke from Electronic Health Records, Proc. 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2019. (* Authors contributed equally.)

Please cite the above paper if you intend to use whole/part of the code. This code is only for academic and research purposes.

Biplot representation of the actual patient attributes projected on the first two principal components.

Code Organization

All codes are written in R.

Code

The script to reproduce all the figures, tables in the paper is main.R. Please run this script in R studio.

Results

We also share the results obtained in our random downsampling experiments. The results obtained with the various benchmarking approaches is found in my_results.csv.