To do what I can as a non-clinician to help the fight against COVID-19.
What it does
Forecasts the number of new COVID-19 cases for each county in the US using an LSTM neural network model to model the time series.
How I built it
Predominantly Python and Keras in a notebook environment.
Challenges I ran into
Finding county-level data was more difficult than expected. I was planning to add demographics and social determinants based on county, but had more difficulty finding suitable data sources.
Accomplishments that I'm proud of
Finishing this project as a team of one while also doing my full-time job.
What I learned
Training LSTMs is hard!
What's next for Predicting COVID-19 Case Counts
Improving the model with additional data sources, more time spent in parameter optimization.