How neural networks and deep learning technology can help predict neurological disorders

The scientific community has long believed that the body’s immune and brain systems operate independent from each other.

While limited research efforts have provided little reason to dispute this claim, a recent discovery by the University of Virginia’s Center for Brain Immunology and Glia (BIG) of a physical connection between the immune and neurological systems has drastically changed this notion. The BIG team has discovered a previously unknown lymphatic system made up of T-cells in the vessels surrounding the spinal cord, showing a clear link between the immune system and brain.

“There has historically been a limited scope in differentiating cell type and function in biological datasets,” said MSDS student Caitlin Dreisbach. “However, major advances in genomic sequencing now allow researchers and scientists to move beyond [prior understandings] of the complexity of big data.”

Capstone posterMSDS students Morgan Wall, Ali Zaidi and Dreisbach are working with Prof. Abigail Flower (Data Science Institute) to discover how immune cells in the brain can be used to address emerging questions about cell type and function. The research team is looking at different types of cells in both the cortex and hippocampus, two major various regions of the brain. Their goal is to better identify the functions of these cells, ultimately to find better ways to address neurological disorders such as Alzheimer’s and Parkinson’s Disease.

“Revolutionary processes in technology have opened new biological questions,” said Dreisbach, “they allow us to overcome fundamental challenges inherent in the genomic sciences.”

This project focuses on classifying the different types of cells by gene expression data, ultimately resulting in the potential ability to identify immunological themes in the brain.

“We are trying to use data science to overcome key issues in large, publicly available datasets of gene expression,” said Dreisbach. “The central objective is to develop and apply analytic methods in neuroimmunology.”

A key strength of this project is the use of existing data from the published literature and a strong multidisciplinary research team, including data science, neuroscience, computational biology, nursing, and genomics.

Specifically, Dreisbach, Wall, and Zaidi are using data science methods to reduce the complexity and bulk of the dataset, with the goal of creating a neural network that will identify distinguishing gene sets within each cell type. Their research is broadening the understanding of how the immune system impacts neurological disorders, such as Alzheimer’s disease, opening the door for an entirely new approach in predicting adverse neurological outcomes.  

“This research is important because of future applications of these methods,” Dreisbach said. “It can help to expound upon immune-cell specific functions and translational research on specific neurological diseases.”


M. Wall, C. Dreisbach, A. Zaidi, A. Flower, C. Overall. "Using Autoencoders and Text Mining to Characterize Single Cell Populations in the Hippocampus and Cortex." 2018 Systems and Information Engineering Design Symposium (SIEDS), Charlottesville, VA, 2018.


Learn about other Capstone Projects

Completed in:
2018
Project type:
Partner:

University of Virginia Brain Institute