Fig. 8

MAE of the district-level case number predictions of SEIR, MRPMC, LSTM-T, Informer, and our method in New York City, San Francisco, Toronto, London, Berlin, and Tokyo. a The MAE of one-week-ahead (7 days) predictions. b The MAE of two-weeks-ahead (14 days) predictions. Lower MAE values indicate better prediction performance. MAE: mean absolute error; SEIR: the Susceptible-Exposed-Infectious-Removed model [58]; MRPMC: Mortality Risk Prediction Model for COVID-19 [60]; LSTM-T: a long short-term memory network deep transfer learning model for COVID-19 case trend prediction [31]; Informer: a state-of-the-art deep learning model for time series prediction [59]; and Our method: a deep transfer learning model with adapted TransCodes