Forecasting air quality time series using deep learning

Deep learning for predicting the occurrence of cardiopulmonary diseases in Nanjing, China

The efficiency of disease prevention and medical care service necessitated the prediction of incidence. However, predictive accuracy and power were largely impeded in a complex system including multiple environmental stressors and health outcome of which the occurrence might be episodic and irregular in time. In this study, we established four different deep learning (DL) models to capture inherent long-term dependencies in sequences and potential complex relationships among constituents by initiating with the original input into a representation at a higher abstract level. We collected 504,555 and 786,324 hospital outpatient visits of grouped categories of respiratory (RESD) and circulatory system disease (CCD), respectively, in Nanjing from 2013 through 2018. The matched observations in time-series that might pose risk to cardiopulmonary health involved conventional air pollutants concentrations and metrological conditions. The results showed that a well-trained network architecture built upon long short-term memory block and a working day enhancer achieved optimal performance by three quantitative statistics, i.e., 0.879 and 0.902 of Nash-Sutcliffe efficiency, 0.921% and 0.667% of percent bias, and 0.347 and 0.312 of root mean square error-standard deviation ratio for RESD and CCD hospital visits, respectively. We observed the non-linear association of nitrogen dioxide and ambient air temperature with CCD hospital visits. Furthermore, these two environmental stressors were identified as the most sensitive predictive variables, and exerted synergetic effect for two health outcomes, particular in winter season. Our study indicated that high-quality surveillance data of atmospheric environments could provide novel opportunity for anticipating temporal trend of cardiopulmonary health outcomes based on DL model.

Prophet forecasting model: a machine learning approach to predict the concentration of air pollutants (PM2.5, PM10, O3, NO2, SO2, CO) in Seoul, South Korea

Amidst recent industrialization in South Korea, Seoul has experienced high levels of air pollution, an issue that is magnified due to a lack of effective air pollution prediction techniques. In this study, the Prophet forecasting model (PFM) was used to predict both short-term and long-term air pollution in Seoul. The air pollutants forecasted in this study were PM2.5, PM10, O3, NO2, SO2, and CO, air pollutants responsible for numerous health conditions upon long-term exposure. Current chemical models to predict air pollution require complex source lists making them difficult to use. Machine learning models have also been implemented however their requirement of meteorological parameters render the models ineffective as additional models and infrastructure need to be in place to model meteorology. To address this, a model needs to be created that can accurately predict pollution based on time. A dataset containing three years worth of hourly air quality measurements in Seoul was sourced from the Seoul Open Data Plaza. To optimize the model, PFM has the following parameters: model type, changepoints, seasonality, holidays, and error. Cross validation was performed on the 2017-18 data; then, the model predicted 2019 values. To compare the predicted and actual values and determine the accuracy of the model, the statistical indicators: mean squared error (MSE), mean absolute error (MAE), root mean squared error (RMSE), and coverage were used. PFM predicted PM2.5 and PM10 with a MAE value of 12.6 µg/m3 and 19.6 µg/m3, respectively. PFM also predicted SO2 and CO with a MAE value of 0.00124 ppm and 0.207 ppm, respectively. PFM's prediction of PM2.5 and PM10 had a MAE approximately 2 times and 4 times less, respectively, than comparable models. PFM's prediction of SO2and CO had a MAE approximately five times and 50 times less, respectively, than comparable models. In most cases, PFM's ability to accurately forecast the concentration of air pollutants in Seoul up to one year in advance outperformed similar models proposed in literature. This study addresses the limitations of the prior two PFM studies by expanding the modelled air pollutants from three pollutants to six pollutants while increasing the prediction time from 3 days to 1 year. This is also the first research to use PFM in Seoul, Korea. To achieve more accurate results, a larger air pollution dataset needs to be implemented with PFM. In the future, PFM should be used to predict and model air pollution in other regions, especially those without advanced infrastructure to model meteorology alongside air pollution. In Seoul, Seoul's government can use PFM to accurately predict air pollution concentrations and plan accordingly.

Binary Time Series Classification with Bayesian Convolutional Neural Networks When Monitoring for Marine Gas Discharges

The world’s oceans are under stress from climate change, acidification and other human activities, and the UN has declared 2021–2030 as the decade for marine science. To monitor the marine waters, with the purpose of detecting discharges of tracers from unknown locations, large areas will need to be covered with limited resources. To increase the detectability of marine gas seepage we propose a deep probabilistic learning algorithm, a Bayesian Convolutional Neural Network (BCNN), to classify time series of measurements. The BCNN will classify time series to belong to a leak/no-leak situation, including classification uncertainty. The latter is important for decision makers who must decide to initiate costly confirmation surveys and, hence, would like to avoid false positives. Results from a transport model are used for the learning process of the BCNN and the task is to distinguish the signal from a leak hidden within the natural variability. We show that the BCNN classifies time series arising from leaks with high accuracy and estimates its associated uncertainty. We combine the output of the BCNN model, the posterior predictive distribution, with a Bayesian decision rule showcasing how the framework can be used in practice to make optimal decisions based on a given cost function.

Soft Computing Applications in Air Quality Modeling: Past, Present, and Future

Air quality models simulate the atmospheric environment systems and provide increased domain knowledge and reliable forecasting. They provide early warnings to the population and reduce the number of measuring stations. Due to the complexity and non-linear behavior associated with air quality data, soft computing models became popular in air quality modeling (AQM). This study critically investigates, analyses, and summarizes the existing soft computing modeling approaches. Among the many soft computing techniques in AQM, this article reviews and discusses artificial neural network (ANN), support vector machine (SVM), evolutionary ANN and SVM, the fuzzy logic model, neuro-fuzzy systems, the deep learning model, ensemble, and other hybrid models. Besides, it sheds light on employed input variables, data processing approaches, and targeted objective functions during modeling. It was observed that many advanced, reliable, and self-organized soft computing models like functional network, genetic programming, type-2 fuzzy logic, genetic fuzzy, genetic neuro-fuzzy, and case-based reasoning are rarely explored in AQM. Therefore, the partially explored and unexplored soft computing techniques can be appropriate choices for research in the field of air quality modeling. The discussion in this paper will help to determine the suitability and appropriateness of a particular model for a specific modeling context.

Transductive LSTM for time-series prediction: An application to weather forecasting

Long Short-Term Memory (LSTM) has shown significant performance on many real-world applications due to its ability to capture long-term dependencies. In this paper, we utilize LSTM to obtain a data-driven forecasting model for an application of weather forecasting. Moreover, we propose Transductive LSTM (T-LSTM) which exploits the local information in time-series prediction. In transductive learning, the samples in the test point vicinity are considered to have higher impact on fitting the model. In this study, a quadratic cost function is considered for the regression problem. Localizing the objective function is done by considering a weighted quadratic cost function at which point the samples in the neighborhood of the test point have larger weights. We investigate two weighting schemes based on the cosine similarity between the training samples and the test point. In order to assess the performance of the proposed method in different weather conditions, the experiments are conducted on two different time periods of a year. The results show that T-LSTM results in better performance in the prediction task.