The Development of a Physiological Based Sampling Pump for Exposure Assessment
Open Access
- Author:
- Lin, Ming-I
- Graduate Program:
- Industrial Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- October 05, 2009
- Committee Members:
- Andris Frievlads, William A Groves, Dissertation Advisor/Co-Advisor
Andris Freivalds, Committee Chair/Co-Chair
William Arthur Groves, Committee Chair/Co-Chair
John Henry Challis, Committee Member
David Arthur Nembhard, Committee Member
Ling Rothrock, Committee Member - Keywords:
- artificial neural network
sampling pump
ergonomic
ventilation
exposure assessment - Abstract:
- The goal of this dissertation was to develop and evaluate a physiologic sampling pump (PSP) capable of adjusting flow in proportion to minute ventilation rate(VE). Current methods for personal air monitoring involve estimation of time-weighted average(TWA) concentration using a traditional sampling pump (TSP) operating at a fixed flow rate. However, studies have shown that TSP-measured TWA concentration may not correlate well with actual inhaled dose when workload and VE vary considerably. A series of experiments divided into three phases was conducted to develop and evaluate a PSP: (1) development of VE prediction models, (2) verification of prototype PSP performance, and (3) laboratory evaluation of PSP for simulated workplace scenarios. Prediction models based on real-time hysiological and kinematic signals provided by an ambulatory monitoring system (the LifeShirt®) and subject demographic characteristics were developed using data gathered from a series of step-tests completed by nine volunteers. Regression-based VE prediction model performance was satisfactory and comparable to previously published models; however, the best results were obtained using an artificial neural network (ANN) with prediction errors <1% (R2 = 0.88)compared to a reference pneumotachograph. Prediction of VE appears to be a somewhat novel application of ANNs. Subsequent experiments conducted to verify the performance of a prototype PSP programmed with a VE prediction model for flow control showed excellent agreement between actual air sampling rate, subject VE, total sampled volume, and reference measures with errors well below 5% and good reproducibility. Fifteen subjects completed replicate step-tests in an exposure chamber for two simulated workplace exposure scenarios in which m-xylene concentrations were correlated and uncorrelated with VE in order to compare PSP- and TSP-measured TWA concentrations. PSP results were significantly higher than the TSP (PSP/TSP = 1.18) when VE was highly correlated with m-xylene concentration (r = 0.93). Race, gender, age, and BMI did not significantly influence the relationship between PSP and TSP measured concentrations. This is the first laboratory demonstration of the effect of correlation of VE and air concentrations on PSP- and TSP-measured TWA concentrations and results show the potential for significantly improving the accuracy of exposure assessment through the use of a PSP.