Forecasting upper respiratory tract infection burden using high-dimensional time series data and forecast combinations