A machine learning exploration of Human Connectome data

Author

Katrina Prantzalos

Date Approved

2018

Degree Type

Open Access Senior Honors Thesis

Department or School

Mathematics

First Advisor

Dr. Andrew Ross

Abstract

The intent of this project is to statistically examine the relationships between psychological traits, demographics, and physiological information provided by various brain scans. The Human Connectome Project dataset includes brain imaging data, demographics, and psychological data, from 1,206 individuals. Using this data, we performed exploratory data analysis, including an investigation of the distinctions in connectome data across participants with anxiety and/or depression as opposed to those without. Our aim was to predict the two mental disorders from given data via machine learning techniques such as Random Forest and Boosted Trees methods. Unfortunately, we were unable to reasonably predict anxiety or depression from the connectome data through any of the attempted methods.

Comments

R or RStudio is necessary to run the R file.

Recommended Citation

Prantzalos, Katrina, "A machine learning exploration of Human Connectome data" (2018). Senior Honors Theses and Projects. 632.
https://commons.emich.edu/honors/632