A Novel Online Multi-task Learning For COVID-19 Multi-Output Spatio-Temporal Prediction

In light of the ongoing COVID-19 pandemic, predicting its trend would significantly impact decision-making. However, this is not a straightforward task due to three main difficulties: temporal autocorrelation, spatial dependency, and concept drift caused by virus mutations lockdown policies. Although machine learning has been extensively used in related work, no previous research successfully addressed all challenges simultaneously. To overcome challenge, we developed novel online multi-task regression algorithm that incorporates chain structure capture ADWIN detector adapt drift, lag time series feature autocorrelation. We conducted several comparative experiments based on number daily confirmed cases 20 areas California affiliated cities. The results from our demonstrate proposed model superior adapting data capturing dependencies across various regions. This leads significant improvement prediction accuracy when compared existing state-of-the-art batch methods, such as N-Beats, DeepAR, TCN, LSTM.