View the Final Code in Jupyter


Computer Science Final Presentation Available Here
~
Math Final Presentation Available Here


Usage: This code was written in Python 3.8.5
Requirements are listed in requirements
Cuda is required to run

Mental illness can be a burden for those whose lives are impacted from it, especially those with serious mental illness. This burden is exasperated when multiple mental illnesses co-occur - a phenomenon known as psychiatric co-morbidity (further referred to as comorbidity). Recently, the field of psychometric analysis, or the analysis of psychological measurement techniquesand theories, has had two major perspectives of how two analyze comorbidity: the latent variable perspective and the network perspective. We extend Cramer et. al.’s (2010) initial hypothesis of the usability of network analysis to latent network path analysis, identifying pathways of comorbidity toidentify causal symptoms of disorders. We will do this in a multi-stage production. First, we will develop networks of individuals in the NCSR building directed graphs based on the onset age of symptoms. Then, we will combine all of the networks for individuals with a comorbid diagnoses using the Deep Graph Library (Wang et al., 2019) (DGL). We then use the combinednetwork to identify latent symptom pathways using Graph Neural Networks using SAGE Graphs with a backend of convolutional neural networks and Long Short-Term Memory (LSTM) Convolutional Neural Networks (RNN).

To get a local copy up and running follow these simple steps.

1. Clone the repo

   git clone https://github.com//025georgialynny/seminar.git

1. Install all requirements

2. Run python3 extract_age_vars.py

3. Run python3 make_indiv_graphs.py

4. Run ./get_dsm_graphs.py

Distributed under the MIT License. See LICENSE for more information.

Project Link: https://github.com/025georgialynny/seminar