Mining multi-granular multivariate medical measurements

<p>This thesis is motivated by the need to predict the mortality of patients in the Intensive Care Unit. The heart of this problem revolves around being able to accurately classify multivariate, multi-granular time series patient data. The approach ultimately taken in this thesis involves using Z-Score normalization to make variables comparable, Single Value Decomposition to reduce the number of features, and a Support Vector Machine to classify patient tuples. This approach proves to outperform other classification models such as k-Nearest Neighbor and demonstrates that SVM is a viable model for this project. The hope is that going forward other work can build off of this research and one day make an impact in the medical community.</p>

Name thesis:degree_name

Master of Science (MS) in Computer Science

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Year

2014

Author dc:creator

• Sykes, Conrad

• Computer Sciences

Statement dc:rights

• Access is available to all users

Repository record dc:identifier

https://dc.ewu.edu/theses/196

OAI identifier oai:identifier

oai:dc.ewu.edu:theses-1195