INTRODUCTION

Sepsis Definitions
Machine Learning and Electronic Health Records
AIMS
METHODS
Study Design
Study Setting and Population
Study Protocol
Data Set Creation
Imputation
Autoencoder Training
Clustering
RESULTS AND DISCUSSION
Quality of dimensionality reduction and latent representation
Clustering
Assessing clustering propensity
Assessing ideal number of clusters
Partitional Methods
K-means
K-medoids
Hierarchical Methods
Agglomerative clustering with ward linkage
Agglomerative clustering with single and complete linkage
Density-Based Methods
DBSCAN
Making Sense of the Clustering
Limitations and Advantages
CONCLUSIONS
REFERENCES
APPENDIX