Integrating genomics and high-frequency physiologic data for sepsis detection

In this project we intend to integrate publicly available -omic and clinical datasets using natural language processing techniques. Combining genomics data with physiologic read-outs may be effective in creating robust machine learning and data analysis pipeline. The example microarray gene expression data can be downloaded from GEO (https://www.synapse.org/#!Synapse:syn5612563) and physiologic data from eICU (https://eicu-crd.mit.edu/). The idea is to map phenotypic terms to causal genes (for sepsis) and follow the SIRS timeline to form the integrated data set. After that robust machine learning models can be formed using the integrated data and compared with already existing models.