Back

How well can we forecast the COVID-19 pandemic with curve fitting and recurrent neural networks?

Zhao, Z.; Nehil-Puleo, K.; Zhao, Y.

2020-05-18 health informatics
10.1101/2020.05.14.20102541 medRxiv
Show abstract

Predictions of the COVID-19 pandemic in USA are compared using curve fitting and various recurrent neural networks (RNNs) including the standard long short-term memory (LSTM) RNN and 10 types of slim LSTM RNNs. The curve fitting method predicts the pandemic would end in early summer but the exact date and scale vary with the evolving data used for fitting. All LSTM RNNs result in short-term (8 to 10 days) predictions with comparable accuracies (smaller than 10 %) to curve fitting--they do not show advantage over curve fitting.

Matching journals

The top 9 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.