Predicting COVID-19 Cases From Atmospheric Parameters Using Machine Learning Approach

Analysis and prediction of infectious diseases based on spatial visualization and machine learning

Infectious diseases are a global public health problem that poses a threat to human society. Since the 1970s, constantly mutated new infectious viruses have been quietly attacking humanity, and at least one new type of infectious disease is discovered every year. Therefore, early warning of infectious diseases will greatly reduce the socio-economic harm of infectious diseases. This study is based on the data of COVID-19 epidemic in China (except Macau and Taiwan Province) from 2020 to 2022. Firstly, we used ArcGIS software to analyze the spatial agglomeration pattern of the number of patients in various regions of China through global spatial autocorrelation analysis, local spatial autocorrelation analysis, center of gravity trajectory migration algorithm and other statistical tools; In addition, the areas with serious COVID-19 epidemic and requiring special attention were screened out. Then, autoregressive integrated moving average model (ARIMA), extreme learning machine (ELM), support vector regression (SVR), wavelet neural network (Wavelet), recurrent neural network (RNN) and long short-term memory (LSTM) were used to predict COVID-19 epidemic data in Guangdong Province, China; And the prediction performance of each model was compared through prediction accuracy indicators. Finally, a multi algorithm fusion learning model based on stacking technology is proposed to address the problem of poor generalization ability of single algorithm models in prediction; Furthermore, radial basis function network (RBF) was used as a two-level meta learner to fuse the above models, and particle swarm optimization (PSO) was used to optimize RBF parameters to reduce generalization error. The experimental results show that the performance of the integrated model is better than that of the single model in the COVID-19 dataset. In order to better apply the stacking model to the prediction of new infectious diseases, we applied the prediction model based on the COVID-19 dataset to the prediction of the number of AIDS and pulmonary tuberculosis (PTB) cases, and verified the wide applicability of our model in the prediction of infectious diseases.

Significance of weather condition, human mobility, and vaccination on global COVID-19 transmission

The transmission growth rate of infectious diseases, particularly COVID-19, has forced governments to take immediate control decisions. Previous studies have shown that human mobility, weather condition, and vaccination are potential factors influencing virus transmission. This study investigates the contribution of weather conditions, namely temperature and precipitation, human mobility, and vaccination to coronavirus transmission. Three machine learning models: random forest (RF), XGBoost, and neural networks, are applied to predict the confirmed cases based on three aforementioned variables. All models' prediction are evaluated via spatial and temporal analysis. The spatial analysis observes the model performance over countries on certain times. The temporal analysis looks at the model prediction of each country during the specified period. The models' prediction results effectively indicate the transmission trend. The RF model performs best with a coefficient of determination of up to 89%. Meanwhile, all models confirm that vaccination is most significantly associated with COVID-19 cases.

Prediction for odor gas generation from domestic waste based on machine learning

Domestic waste is prone to produce a variety of volatile organic compounds (VOCs), which often has unpleasant odors. A key process in treating odor gases is predicting the production of odors from domestic waste. In this study, four factors of domestic waste (weight, wet composition, temperature, and fermentation time) were adopted to be the prediction indicators in the prediction for domestic waste odor gases. Machine learning models (Random Forest, XGBoost, LightGBM) were established using the odor intensity values of 512 odor gases from domestic waste. Based on these data, the regression prediction with supervised machine learning was achieved, in which three different algorithmic models were evaluated for prediction performance. A Random Forest model with a R² value of 0.8958 demonstrated the most accurate prediction of the production of domestic waste odor gas based on our data. Furthermore, the prediction results in the Random Forest model were further discussed based on the microbial fermentation of domestic waste. In addition to enhancing our knowledge of the production of odor from domestic waste, we also explore the application of machine learning to odor pollution in our study.

Association Between Weather Parameters and SARS-CoV-2 Confirmed Cases in Two South African Cities

Several approaches have been used in the race against time to mitigate the spread and impact of COVID-19. In this study, we investigated the role of temperature, relative humidity, and particulate matter in the spread of COVID-19 cases within two densely populated cities of South Africa-Pretoria and Cape Town. The role of different levels of COVID-19 restrictions in the air pollution levels, obtained from the Purple Air Network, of the two cities were also considered. Our results suggest that 26.73% and 43.66% reduction in PM2.5 levels were observed in Cape Town and Pretoria respectively for no lockdown (Level 0) to the strictest lockdown level (Level 5). Furthermore, our results showed a significant relationship between particulate matter and COVID-19 in the two cities. Particulate matter was found to be a good predictor, based on the significance of causality test, of COVID-19 cases in Pretoria with a lag of 7 days and more. This suggests that the effect of particulate matter on the number of cases can be felt after 7 days and beyond in Pretoria.