A Novel Method for Quantifying the Inhaled Dose of Air Pollutants Based on Heart Rate, Breathing Rate and Forced Vital Capacity

• Type of Material: Article
• Authors: Robert Greenwald, Emory University; Matthew J. Hayat, Georgia State University; Jerusha Barton, Emory University; Anastasia Lopukhin, Emory University
• Language: English
• Date: 2016-01-25
• Publisher: Public Library of Science
• Publication Version: Final Published Version
• Copyright Statement: © 2016 Greenwald et al
• License: Creative Commons Attribution 4.0 International
• Final Published Version (URL): http://dx.doi.org/10.1371/journal.pone.0147578
• Title of Journal or Parent Work: PLoS ONE
• ISSN: 1932-6203
• Volume: 11
• Issue: 1
• Start Page: e0147578
• End Page: e0147578
• Grant/Funding Information: Funding for this work came from NIEHS grant number K25ES020355.
• Keywords: Dose prediction methods; Asthma; Covariance; Air pollution; Heart rate; Physical activity; Data processing; Sports Medicine
• Research Categories: Biology, Biostatistics; Health Sciences, Public Health; Environmental Sciences

Abstract

To better understand the interaction of physical activity and air pollution exposure, it is important to quantify the change in ventilation rate incurred by activity. In this paper, we describe a method for estimating ventilation using easily-measured variables such as heart rate (HR), breathing rate (fB), and forced vital capacity (FVC). We recruited healthy adolescents to use a treadmill while we continuously measured HR, fB, and the tidal volume (VT) of each breath. Participants began at rest then walked and ran at increasing speed until HR was 160–180 beats per minute followed by a cool down period. The novel feature of this method is that minute ventilation (V˙E) was normalized by FVC. We used general linear mixed models with a random effect for subject and identified nine potential predictor variables that influence either V˙E or FVC. We assessed predictive performance with a five-fold cross-validation procedure. We used a brute force selection process to identify the best performing models based on cross-validation percent error, the Akaike Information Criterion and the p-value of parameter estimates. We found a two-predictor model including HR and fB to have the best predictive performance (V˙E/FVC = -4.247+0.0595HR+0.226fB, mean percent error = 8.1±29%); however, given the ubiquity of HR measurements, a one-predictor model including HR may also be useful (V˙E/FVC = -3.859+0.101HR, mean percent error = 11.3±36%).