Interior_Banner_Events

Predicting the Spread of Covid-19 with Neural Networks

Isaac Boyd, Kieran Ringle, David Hedges, Bradley M. Whitaker, Montana State University Electrical and Computer Engineering Department

In late 2019 SARS-CoV-2 (COVID-19) was exposed to the public. Since the preliminary Chinese outbreak, the virus has spread to the level of a worldwide pandemic. Attempts to contain the virus and mitigate its effects have been hampered by its easy transmissibility. Therefore, it has become necessary for government policymakers as well as healthcare organization policymakers to know where COVID-19 will spread to next so they can prepare for hospitalizations and try to avoid outbreaks. In recent years, Convolutional Neural Networks (CNNs)  have been successful in tracking diseases. A study on Tuberculosis in 2019 showed that by using a CNN to analyze image data the team could diagnose a patient with tuberculosis with an accuracy of 96.63%. The benefit of using this machine learning technique is that a large array of data, static and time-varying, can be composed into a more usable and holistic model. In this research, we will develop a model using CNNs to track and forecast the regional spread of COVID-19. In our model, we will strive to utilize static variables through Keras’s functional application programming interface. This will allow us to take a multivariable input and convert our time-varying data into a model biased by the static traits of individual regions. We also plan to implement a binary classifier to account for skewed data in rural areas where daily case counts are often near zero and infection rates are lower. This model will be used to predict hospitalizations as well as future outbreaks on a county by county basis for Montana, Wyoming, South Dakota, North Dakota, and Idaho.




Additional Abstract Information

Presenters: Isaac Boyd, Kieran Ringel

Institution: Montana State University

Type: Poster

Subject: Computer Science

Status: Approved


Time and Location

Session: Poster 5
Date/Time: Tue 12:30pm-1:30pm
Session Number: 4036