Evaluating long-term and high spatiotemporal resolution of wet-bulb globe temperature through land-use based machine learning model.

BACKGROUND: The increase in global temperature and urban warming has led to the exacerbation of heatwaves, which negatively affect human health and cause long-term loss of work productivity. Therefore, a global assessment in temperature variation is essential. OBJECTIVE: This paper is the first of its kind to propose land-use based spatial machine learning (LBSM) models for predicting highly spatial-temporal variations of wet-bulb globe temperature (WBGT), which is a heat stress indicator used to assess thermal comfort in indoor and outdoor environments, specifically for the main island of Taiwan. METHODS: To develop spatiotemporal prediction models for both the working period and noon period, we calculated the WBGT of each weather station from 2001 to 2019 using temperature, humidity, and solar radiation data. These WBGT estimations were then used as the dependent variable for developing the spatiotemporal prediction models. To enhance model performance, we used innovative approaches that combined SHapley Additive exPlanations (SHAP) values for the selection of non-linear variables, along with machine learning algorithms for model development. RESULTS: When incorporating temperature along with other land-use/land cover predictor variables, the performance of LBSM models was excellent, with an R2 value of up to 0.99. The LBSM models explained 98% and 99% of the spatial-temporal variations in WBGT for the working and noon periods, respectively, within the complete models. In the temperature-excluded models, the explained variances were 94% and 96% for the working and noon periods, respectively. IMPACT: WBGT is a common method used by many organizations to access the impact of heat stress on human beings. However, limited studies have mentioned the association between WBGT and health impacts due to the absence of spatiotemporal databases. This study develops a new approach using land-use-based spatial machine learning (LBSM) models to better predict the fine spatial-temporal WBGT levels, with a 50-m × 50-m grid resolution for both working time and noontime. Our proposed methodology could be used in future studies aimed at evaluating the potential long-term loss of work productivity due to the effects of global warming or urban heat island.

Using a land use regression model with machine learning to estimate ground level PM2.5.

Ambient fine particulate matter (PM2.5) has been ranked as the sixth leading risk factor globally for death and disability. Modelling methods based on having access to a limited number of monitor stations are required for capturing PM2.5 spatial and temporal continuous variations with a sufficient resolution. This study utilized a land use regression (LUR) model with machine learning to assess the spatial-temporal variability of PM2.5. Daily average PM2.5 data was collected from 73 fixed air quality monitoring stations that belonged to the Taiwan EPA on the main island of Taiwan. Nearly 280,000 observations from 2006 to 2016 were used for the analysis. Several datasets were collected to determine spatial predictor variables, including the EPA environmental resources dataset, a meteorological dataset, a land-use inventory, a landmark dataset, a digital road network map, a digital terrain model, MODIS Normalized Difference Vegetation Index (NDVI) database, and a power plant distribution dataset. First, conventional LUR and Hybrid Kriging-LUR were utilized to identify the important predictor variables. Then, deep neural network, random forest, and XGBoost algorithms were used to fit the prediction model based on the variables selected by the LUR models. Data splitting, 10-fold cross validation, external data verification, and seasonal-based and county-based validation methods were used to verify the robustness of the developed models. The results demonstrated that the proposed conventional LUR and Hybrid Kriging-LUR models captured 58% and 89% of PM2.5 variations, respectively. When XGBoost algorithm was incorporated, the explanatory power of the models increased to 73% and 94%, respectively. The Hybrid Kriging-LUR with XGBoost algorithm outperformed the other integrated methods. This study demonstrates the value of combining Hybrid Kriging-LUR model and an XGBoost algorithm for estimating the spatial-temporal variability of PM2.5 exposures.

Association between Surrounding Greenness and Mortality: An Ecological Study in Taiwan.

Exposure to surrounding greenness is associated with reduced mortality in Caucasian populations. Little is known however about the relationship between green vegetation and the risk of death in Asian populations. Therefore, we opted to evaluate the association of greenness with mortality in Taiwan. Death information was retrieved from the Taiwan Death Certificate database between 2006 to 2014 (3287 days). Exposure to green vegetation was based on the normalized difference vegetation index (NDVI) collected by the Moderate Resolution Imagine Spectroradiometer (MODIS). A generalized additive mixed model was utilized to assess the association between NDVI exposure and mortality. A total of 1,173,773 deaths were identified from 2006 to 2014. We found one unit increment on NDVI was associated with a reduced mortality due to all-cause (risk ratio [RR] = 0.901; 95% confidence interval = 0.862-0.941), cardiovascular diseases (RR = 0.892; 95% CI = 0.817-0.975), respiratory diseases (RR = 0.721; 95% CI = 0.632-0.824), and lung cancer (RR = 0.871; 95% CI = 0.735-1.032). Using the green land cover as the alternative green index showed the protective relationship on all-cause mortality. Exposure to surrounding greenness was negatively associated with mortality in Taiwan. Further research is needed to uncover the underlying mechanism.

Effects of surrounding environment on incidence of end stage renal disease.

The rising prevalence and incidence of end-stage renal disease (ESRD) have been noted around the world. However, no study has been conducted to examine the effect of surrounding environment on incidence of ESRD. This study assessed the associations of exposure to PM2.5 level and surrounding green spaces, separately, with incidence of ESRD in Taiwan. Demographic and clinical data used in this study was retrieved from the National Health Insurance Research Database from 2003 to 2012. PM2.5 data collected from the Environmental Protection Administration of Taiwan and a hybrid land-use regression model was used to approximate long-term exposure to PM2.5. Percentage of exposure to surrounding green spaces was used to determine individual exposure level. Cox proportional hazards models with a generalized estimating equation were applied to investigate the effect of surrounding environment on incidence of ESRD. The results showed significant positive association between exposure to PM2.5 level and incidence of ESRD; but inverse association between exposure to surrounding green spaces and incidence of ESRD (adjusted hazard ratio (AHR) = 1.08, 95% CI: 1.00-1.15 for exposure to PM2.5 level; AHR = 0.90, 95%CI: 0.84-0.98 for surrounding green spaces). Together, the findings from this study have added suggestive evidence on the adverse effect of exposure to PM2.5 level and the beneficial effect of exposure to surrounding green spaces on the incidence of ESRD in a general population in Taiwan.

The Effect of Surrounding Greenness on Type 2 Diabetes Mellitus: A Nationwide Population-Based Cohort in Taiwan.

This study determines whether surrounding greenness is associated with the incidence of type 2 diabetes Mellitus (T2DM) in Taiwan. A retrospective cohort study determines the relationship between surrounding greenness and the incidence of T2DM during the study period of 2001-2012 using data from the National Health Insurance Research Database. The satellite-derived normalized difference vegetation index (NDVI) from the global MODIS database in the NASA Earth Observing System is used to assess greenness. Cox proportional hazard models are used to determine the relationship between exposure to surrounding greenness and the incidence of T2DM, with adjustment for potential confounders. A total of 429,504 subjects, including 40,479 subjects who developed T2DM, were identified during the study period. There is an inverse relationship between exposure to surrounding greenness and the incidence of T2DM after adjustment for individual-level covariates, comorbidities, and the region-level covariates (adjusted HR = 0.81, 95% CI: 0.79-0.82). For the general population of Taiwan, greater exposure to surrounding greenness is associated with a lower incidence of T2DM.

Spatial Clustering of Dengue Fever Incidence and Its Association with Surrounding Greenness.

With more than 58,000 cases reported by the country's Centers for Disease Control, the dengue outbreaks from 2014 to 2015 seriously impacted the southern part of Taiwan. This study aims to assess the spatial autocorrelation of the dengue fever (DF) outbreak in southern Taiwan in 2014 and 2015, and to further understand the effects of green space (such as forests, farms, grass, and parks) allocation on DF. In this study, two different greenness indexes were used. The first green metric, the normalized difference vegetation index (NDVI), was provided by the long-term NASA MODIS satellite NDVI database, which quantifies and represents the overall vegetation greenness. The latest 2013 land use survey GIS database completed by the National Land Surveying and Mapping Center was obtained to access another green metric, green land use in Taiwan. We first used Spearman's rho to find out the relationship between DF and green space, and then three spatial autocorrelation methods, including Global Moran's I, high/low clustering, and Hot Spot were employed to assess the spatial autocorrelation of DF outbreak. In considering the impact of social and environmental factors in DF, we used generalized linear mixed models (GLMM) to further clarify the relationship between different types of green land use and dengue cases. Results of spatial autocorrelation analysis showed a high aggregation of dengue epidemic in southern Taiwan, and the metropolitan areas were the main hotspots. Results of correlation analysis and GLMM showed a positive correlation between parks and dengue fever, and the other five green space metrics and land types revealed a negative association with DF. Our findings may be an important asset for improving surveillance and control interventions for dengue.