An Interpretable Hybrid Predictive Model of COVID-19 Cases using Autoregressive Model and LSTM

Abstract

The Coronavirus Disease 2019 (COVID-19) has posed a severe threat to global human health and economic. It is an urgent task to build reliable data-driven prediction models for Covid 19 cases to improve public policy making. However, COVID-19 data shows special transmission characteristics such as significant fluctuations and non-stationarity, which may be difficult to be captured by a single predictive model and poses grand challenges in effective forecasting. In this paper, we proposed a novel Hybrid data-driven model combining Autoregressive model (AR) and long short-term memory neural networks (LSTM). It can be viewed as a new neural network model and the contribution of AR and LSTM is auto tuned in the training procedure. We conduct extensive numerical experiments on data collected from 8 counties of California that display various trends. The numerical results show the Hybrid model' advantages over AR and LSTM by its predictive powers. We show that the Hybrid model achieved 4.195% MAPE, outperformed the AR 5.629% and LSTM 5.070% on average, and provide a discussion on interpretability.

Forecasting Time Series Covid 19 Using Deep Learning Methods: South America Comparative Case Study

“…MAPE metric values ranged between 4% and 10% for the Stacked model, between 3% and 7% for the Bidirectional variant, and between 2% and 3.33% for the Convolutional LSTM model. In relation to the results obtained in (Zhang, Tang, and Yu 2023), by applying a model combining AR and LSTM, a MAPE of 4.173% was achieved. Finally, in (Swaraj et al 2021) a hybrid model composed of ARIMA and NAR was proposed, which obtained a MAPE of 4.7%.…”

“…Given the difficulty of correctly forecasting pandemic behavior, not only traditional Deep Learning models were used in the literature, but also hybrid models were developed to achieve better results. (Zhang, Tang, and Yu 2023) the autoregressive (AR) model is used together with the LSTM, thus taking advantage of the interpretive capacity that AR models can acquire together with the predictive power of LSTM. The effectiveness of the model was proven by applying it in different California counties to predict Covid-19 cases, with the hybrid model presenting a MAPE of 4.173%, outperforming both the AR (5.629%) and LSTM (4.934%) models alone.…”