Spatial Modeling of Asthma-Prone Areas Using Remote Sensing and Ensemble Machine Learning Algorithms

In this study, asthma-prone area modeling of Tehran, Iran was provided by employing three ensemble machine learning algorithms (Bootstrap aggregating (Bagging), Adaptive Boosting (AdaBoost), and Stacking). First, a spatial database was created with 872 locations of asthma patients and affecting factors (particulate matter (PM10 and PM2.5), ozone (O3), sulfur dioxide (SO2), carbon monoxide (CO), nitrogen dioxide (NO2), rainfall, wind speed, humidity, temperature, distance to street, traffic volume, and a normalized difference vegetation index (NDVI)). We created four factors using remote sensing (RS) imagery, including air pollution (O3, SO2, CO, and NO2), altitude, and NDVI. All criteria were prepared using a geographic information system (GIS). For modeling and validation, 70% and 30% of the data were used, respectively. The weight of evidence (WOE) model was used to assess the spatial relationship between the dependent and independent data. Finally, three ensemble algorithms were used to perform asthma-prone areas mapping. According to the Gini index, the most influential factors on asthma occurrence were distance to the street, NDVI, and traffic volume. The area under the curve (AUC) of receiver operating characteristic (ROC) values for the AdaBoost, Bagging, and Stacking algorithms was 0.849, 0.82, and 0.785, respectively. According to the findings, the AdaBoost algorithm outperforms the Bagging and Stacking algorithms in spatial modeling of asthma-prone areas.

Commuter PM exposure and estimated life-expectancy loss across multiple transportation modes in Xi'an, China

Commuters are reportedly exposed to severe traffic-related air pollution (TRAP) during their commuting trips. This study was designed and implemented to (1) compare particulate matter (PM) exposure across four common transportation modes; (2) examine and analyze various determining factors; and (3) estimate public health effects caused by commuting exposure to PM. All analyses and calculations were based on the experimental data collected from 13 volunteers, including heart-rate data on 336 commuting trips in four travel modes in Xi'an China. The results indicate highest PM exposure associated with cycling (average PM₁₀, PM_2.5 and PM_1.0 of 114.35, 72.37 and 56.51 μg/m³, respectively), followed by riding transit buses (116.29, 67.60 and 51.12 μg/m³ for the same pollutants, respectively), then taking a taxi (97.61, 58.87 and 45.11 μg/m³), and the lowest exposure onboard subways (55.86, 46.20 and 40.20 μg/m³). A multivariable linear regression model was used to examine major influences on PM concentration variations, with results corroborating significant PM variance across commuting modes, which is also affected by background pollution concentration and relative humidity. Further, years of life expectancy (YLE) loss were estimated using an inhalation dose model together with the life table method: cycling commuters experienced the greatest YLE loss (5.51-6.43 months per capita for the studied age group). During severe pollution periods, substituting other modes (like subway) for cycling could effectively avoid acute exposure. PM_2.5 levels in taxi cabins powered by CNG or methanol were comparatively lower, indicating that implementing alternative energy strategies could effectively lower traffic emissions and population exposure.

Geospatial and temporal variation of prostate cancer incidence

OBJECTIVES

The objective of this study was to evaluate geographical and temporal variations in prostate cancer incidence in Victoria, Australia.

STUDY DESIGN & METHODS

This study analysed 105,349 cases of incident prostate cancer between 1982 and 2016 from the population-based Victorian Cancer Registry. We performed Poisson regression analyses to identify an association between an annual number of prostate cancer counts, prostate-specific antigen (PSA) tests and the elderly male population (≥65) after adjusting for population at risk and years. We also applied Bayesian spatial-temporal models to determine any association with prostate cancer incidence and area-level factors.

RESULTS

The overall trend of the age-standardized prostate cancer incidence was increasing. The highest age-specific incidence was observed among people aged 65-74 years in the pre- and post-PSA periods. Every increase in 1000 PSA tests per 100,000 population, prostate cancer incidence increased by 17% (relative risk [RR] = 1.17, 95% confidence interval [CI] = 1.13-1.22). A 1% increase in the proportion of the male population (≥65) correlated with a 7% increase in prostate cancer cases (RR = 1.07, 95% CI = 1.06-1.10). Compared with early PSA periods, decreasing trends of low-grade cases and growing trends of high- and intermediate-grade cases were observed after a decline in PSA test usage in late PSA periods. Men living in the most socioeconomically advantaged postal areas had a decreased risk of prostate cancer (RR = 0.914, 95% CI = 0.858-0.976).

CONCLUSIONS

Age-specific risk of developing biological prostate cancer, temporal changes in PSA testing and an increasingly elderly population contributed to an increasing trend of prostate cancer incidence. When incidence trends were investigated at a more granular geographic level, socioeconomically advantaged status was associated with decreased prostate cancer risk.

Influence of timeliness and receipt of first treatment on geographic variation in non-small cell lung cancer mortality

Mortality from non-small cell lung cancer (NSCLC) exhibits substantial geographical disparities. However, there is little evidence on whether this variation could be attributed to patients' clinical characteristics and/or socioeconomic inequalities. This study evaluated the independent and relative contribution of the individual- and area-level risk factors on geographic variation in 2-year all-cause mortality among NSCLC patients. In the Hierarchical-related regression approach, we used the Bayesian spatial multilevel logistic regression model to combine individual- and area-level predictors with outcomes while accounting for geographically structured and unstructured correlation. Individual-level data included 3330 NSCLC cases reported to the Victoria Lung Cancer Registry between 2011 and 2016. Area-level data comprised socioeconomic disadvantage, remoteness and pollution data at the postal area level in Victoria, Australia. With the inclusion of significant individual- and area-level risk factors, timely (≤14 days) first definitive treatment (odds ratio [OR] = 0.73, 95% credible interval [Crl] = 0.56-0.94) and multidisciplinary meetings (MDM) (OR = 0.74, 95% Crl = 0.59-0.93) showed an independent association with a lower likelihood of NSCLC 2-year all-cause mortality. Timely and delayed (>14 days) first nondefinitive treatment, no treatment, advanced clinical stage, smoking, poor performance status, public hospital insurance and area-level deprivation were independently associated with a higher likelihood of 2- and 5-year all-cause mortality. NSCLC's 2-year all-cause mortality exhibited substantial geographic variation, mainly associated with timeliness and receipt of first definitive treatment, no treatment followed by patient prognostic factors with some contribution from area-level deprivation, MDM and public hospital insurance. This study highlights NSCLC patients should receive the first definitive treatment within the recommended 14-days from diagnosis.

Preschool asthma: Examining environmental influences using geographic information systems

OBJECTIVE

This study was conducted to explore if environmental factors co-occur in areas with high asthma rates in Head Start (HS) children.

DESIGN

Descriptive.

SAMPLE

Convenience sample of 56 children with asthma enrolled in HS, ages 3-5 years.

MEASUREMENTS

Geographic Information Systems using ArcGIS 10.4 was used to geocode and map aggregated address data at the census tract level through vector map analysis. Location, race, economic status, pollution remediation sites, age of housing, and blood lead levels were assessed for areas with high asthma concentration.

RESULTS

Most children with asthma resided in one census tract, which was 1% of the total service area. Fifty-six percent of housing was built before 1960 with only 10% after 1990, suggesting deteriorating conditions. Pollution remediation sites were found in the vicinity of asthma cases. Elevated lead levels were found in 22% of all HS children; specific values for the children with asthma were not available.

CONCLUSION

Several co-occurring factors were identified. The need for proactive interventions to decrease asthma risk/poor asthma outcomes with HS is evident. GIS locates children with high susceptibility to asthma. This allows public health nurses to target interventions and educate and empower families about environmental exposures and asthma risk factors.

Returning to our roots: The use of geospatial data for nurse-led community research

In the early 20th century, public health nurse, Lillian Wald, addressed the social determinants of health (SDOH) through her work in New York City and her advocacy to improve policy in workplace conditions, education, recreation, and housing. In the early 21st century, addressing the SDOH is a renewed priority and provides nurse researchers with an opportunity to return to our roots. The purpose of this methods paper is to examine how the incorporation of geospatial data and spatial methodologies in community research can enhance the analyses of the complex relationships between social determinants and health. Geospatial technologies, software for mapping and working with geospatial data, statistical methods, and unique considerations are discussed. An exemplar for using geospatial data is presented regarding associations between neighborhood greenspace, neighborhood violence, and children's asthma control. This innovative use of geospatial data illustrates a new frontier in investigating nontraditional connections between the environment and SDOH outcomes.