Computer-Aided Decoding of Brain Immune Interactions in Gulf War Illness (GWI): A Joint Embedding on Brain Connectomic and ImmunogenomicMarkers
Abstract
We employed a novel classification framework based on a combination of brain connectomics and immuno-geneticapproaches of Gulf War Illness (GWI) to develop novel computer-based diagnostic systems and features. In theprevious project year, we confirmed that veterans with GWI have clear group specific microstructural alteration inlimbic/paralimbic regions compared to GW veterans without illness. This project year, we continued to update subjectdata and confirmed the patterns consistent with our last years findings. We published the result in the scientific journal,and the results are also presented in the VA/DoD virtual conference. In this project year, we continued to work onupgrading the machine learning framework to enhance the classification outcomes. We updated the machine learningframework to use the reinforcement learning method to better define key imaging markers for the classification. Thenew framework provided a significant improvement in classifying veterans with and without GWI. A no-cost extensionwas requested near the end of this project year to add additional subject data, fine-tuning of the new classifier, andfinalizing the software development.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 2020
- Accession Number
- AD1126019
Entities
People
- Bang-Bon Koo
Organizations
- Boston University