Infectious Disease Epidemic Forecasting Using Online News Time Series Analysis: COVID-19 Case Study

The COVID-19 pandemic has become an important warning for preparedness against the possibility of new infectious diseases emerging in the future. This study aims to investigate the potential of online news to predict infectious disease trends. We used daily online news data in Indonesia related to the COVID-19 pandemic to conduct time series analysis using various machine learning methods, such as Convolutional Neural Network (CNN), Multi-Layer Perceptron (MLP), Recurrent Neural Network (RNN), Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM), AutoRegressive Integrated Moving Average (ARIMA), Error Trend Seasonal (ETS), and Logistic Regression. The findings showed that MLP and CNN models attained the best performance, with an RMSE of 0.15 and MAPE of 24%. We also highlight that the RNN model achieved the best performance in smaller data, with an RMSE of 0.24 and MAPE of 37%. These results reveal the potential of RNN usage in a special case of limited data.