A0991
Title: Semiparametric mixed effect state space model with prior information for state-level COVID-19 prediction
Authors: Mengying You - Shanghai University of International Business and Economics (China) [presenting]
Abstract: A state-space approach is introduced to model COVID-19 new case dynamics at the U.S. state level. The model incorporates prior information about curve shape through general smoothing splines. By utilizing a mixed-effects model, information between similarly shaped states is shared, and extra variability is allowed through random effect functions. An efficient state-space formulation of the model enables fast computation and excellent prediction of future observations. Compared to the cubic spline time series model using state-level COVID-19 data, the approach demonstrates a superior ability to extract common patterns and provide more reliable forecasts with smaller prediction intervals.
