Multivariate Feature Extraction Methods for Biomedical Signals and Health Data Classification

In large health data sets, researchers may want to classify individuals into important groups, such as risk groups, based on multiple important variables from various different data resources. Different groups may warrant different, targeted interventions. Combining different data sources to classify individuals with the highest risks and needs can provide health officials an objective way of targeting scarce health care resources. Since each individual would have variables from multiple data sources, determining the key features that can be used to classify individuals into groups requires specialized statistical techniques. This project aims to develop and investigate new multivariate feature extraction methods and apply these methods in a novel way to biomedical signals and health data. Health authorities and stakeholders in Ontario will be able to use these results to inform health care resource allocation decisions.

Faculty Supervisor:

Dr. Sridhar Krishnan

Student:

Shengkun (Victor) Xie

Partner:

Discipline:

Engineering - computer / electrical

Sector:

Life sciences

University:

Ryerson University

Program: