Predicting Oxidative Potential of Vehicular Air Pollution in Metropolitan Atlanta Using Multivariate Regression Modeling Pubblico

Yang, Eric (2016)

Permanent URL: https://etd.library.emory.edu/concern/etds/x346d470z?locale=it
Published

Abstract

Background: Oxidative potential (OP) has been considered to be the new, more novel method of evaluating air pollution exposure, because it takes into account various aspects of air pollution, such as particle size, chemical components, and meteorology. The method for evaluating oxidative potential in this study is with DTT oxidation (OPDTT), using different components of air quality as predictors, such as black carbon, organic carbon, and ozone.

Aims: 1) The first aim of this study is to create a predictive model for OPDTT of air pollution in vehicular traffic of metropolitan Atlanta. The second aim is to use that model to estimate oxidative potential of particle pollution in another study. The third aim is to conduct preliminary analysis of the estimated oxidative potential data on predicting health response.

Methods: All Data was provided by the Atlanta Commuter Studies (ACE-1 and ACE-2). ACE-2 was used to create the predictive model, because it contained OPDTT data. The model was used to create OPDTT estimates for ACE-1. Those estimates were analyzed with changes and percent changes in exhaled nitric oxide (eNO) and forced expiratory volume in 1 second (FEV1) to evaluate the associations between OPDTT and corresponding health response.

Results: The final predictive model for oxidative potential is:

OPDTT = 7.76 - 2.45WSOC + 0.19WSOC2 + 0.019BC - 0.10NOISE + 0.033WSOC*NOISE - 0.0026WSOC2*NOISE.

The adjusted R2 of 0.75 was one of the highest. The predictive model's predictor p-values were all statistically significant except for BC (p=0.34). The ACE-1 OPDTT estimates (0.60±0.14 nmol/minute/m3) created from the model did not appear to have a significant association with change in eNO and FEV1.

Conclusion: In conclusion, the predictive model has reliable fit statistics. However, further analyses involving longitudinal data should be done in regards to evaluating the relationship between OPDTT and health response.

Table of Contents

INTRODUCTION…………………………………………………………………..1

METHODS……………………………………………………………………….….4

RESULTS…………………………………………………………………………....8

DISCUSSION…………………………………………………………………......10

CONCLUSION…………………………………………………………………….13

REFERENCES……………………………………………………………………..14

Table 1: Correlation analysis between OPDTT and all the potential predictors……20

Table 2: Statistics of the Final Predictive Model………………………………………...21

Table 3: Descriptive statistics for the ACE-1 predicted OPDTT estimates attained from the ACE-2 final predictive model…22

Figure 1: Graph of OPDTT and WSOC………………………………………...23

Figure 2: Graph of OPDTT and BC……………………………………………..24

Figure 3: Graph of OPDTT and Noise………………………………………….25

Figure 4: Graph of OPDTT and PAH……………………………………………26

Figure 5: Graph of OPDTT and PNC…………………………………………...27

Figure 6: Graph of OPDTT and O3……………………………………………..28

Figure 7: Graph of OPDTT and WSOC Fit with Linear Spline…………....29

Figure 8: Graph of OPDTT and WSOC Fit with Quadratic Trend………..30

Figure 9: Box Plots of Change in eNO Stratified by OPDTT Tertile Categories…….31

Figure 10: Box Plots of Percent Change in eNO Stratified by OPDTT Tertile Categories…...32

Figure 11: Box Plots of Change in FEV1 Stratified by OPDTT Tertile Categories………….....33

Figure 12: Box Plots of Percent Change in FEV1 Stratified by OPDTT Tertile Categories…..34

About this Master's Thesis

Rights statement
  • Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.
School
Department
Degree
Submission
Language
  • English
Research Field
Parola chiave
Committee Chair / Thesis Advisor
Committee Members
Partnering Agencies
Ultima modifica

Primary PDF

Supplemental Files