A Multivariate Deep Learning Model with Coupled Human Intervention Factors for COVID-19 Forecasting

Artificial intelligence (AI) technology plays a crucial role in infectious disease outbreak prediction and control. Many human interventions can influence the spread of epidemics, including government responses, quarantine, and economic support. However, most previous AI-based models have failed to consider human interventions when predicting the trend of infectious diseases. This study selected four human intervention factors that may affect COVID-19 transmission, examined their relationship to epidemic cases, and developed a multivariate long short-term memory network model (M-LSTM) incorporating human intervention factors. Firstly, we analyzed the correlations and lagged effects between four human factors and epidemic cases in three representative countries, and found that these four factors typically delayed the epidemic case data by approximately 15 days. On this basis, a multivariate epidemic prediction model (M-LSTM) was developed. The model prediction results show that coupling human intervention factors generally improves model performance, but adding certain intervention factors also results in lower performance. Overall, a multivariate deep learning model with coupled variable correlation and lag outperformed other comparative models, and thus validated its effectiveness in predicting infectious diseases.

Adaptive ensemble perception tracking.

Recently, tracking models based on bounding box regression (such as region proposal networks), built on the Siamese network, have attracted much attention. Despite their promising performance, these trackers are less effective in perceiving the target information in the following two aspects. First, existing regression models cannot take a global view of a large-scale target since the effective receptive field of a neuron is too small to cover the target with a large scale. Second, the neurons with a fixed receptive field (RF) size in these models cannot adapt to the scale and aspect ratio changes of the target. In this paper, we propose an adaptive ensemble perception tracking framework to address these issues. Specifically, we first construct a per-pixel prediction model, which predicts the target state at each pixel of the correlated feature. On top of the per-pixel prediction model, we then develop a confidence-guided ensemble prediction mechanism. The ensemble mechanism adaptively fuses the predictions of multiple pixels with the guidance of confidence maps, which enlarges the perception range and enhances the adaptive perception ability at the object-level. In addition, we introduce a receptive field adaption model to enhance the adaptive perception ability at the neuron-level, which adjusts the RF by adaptively integrating the features with different RFs. Extensive experimental results on the VOT2018, VOT2016, UAV123, LaSOT, and TC128 datasets demonstrate that the proposed algorithm performs favorably against the state-of-the-art methods in terms of accuracy and speed.