| Abstract |
Measurement of physical demands of work, and assessment of physical fatigue and the limitations it may have on worker productivity has attracted and occupied work physiologists and industrial engineers for many years. The decrease in performance due to fatigue is widely accepted, but no agreement has been reached in trying to quantify this decrease, or in setting acceptable limits for it. Oxygen uptake measurements during the performance of actual work activities are considered a good measure of the absolute physiologic workload experienced by a person. Expressing measured oxygen uptake as a percentage of maximum oxygen uptake (VO2max), commonly known as relative workload, is recommended by many work physiologists since it provides a subject-specific workload. In addition to accounting for individual differences in physiological capacities among workers, relative workload also enables more accurate assessment of potentials of physical fatigue. The specific aim of this research is to develop a practical and direct method to predict relative workload (expressed as a percentage of maximum oxygen uptake) from in-situ collected sub-maximal oxygen uptake data without the need to determine maximum oxygen uptake. The method is developed by modeling the human cellular utilization system as a stochastic system and on the hypothesis that oxygen uptake data are serially dependent, and that by exploiting this dependence using time series analysis techniques, a regression model between relative workload and a statistical characteristic of collected oxygen uptake data can be developed. In this project, the method developed by Abdelhamid (1999) to predict relative workload from in-situ collected sub-maximal oxygen uptake data was redeveloped for 100 experimental subjects. A relative workload prediction equation (RWP model) was developed with a standard error of prediction of +/- 7 .6% and +/- 0.65 liter min-1 for % VO2max and VO2max, respectively. In an effort to improve the predictive accuracy of the RWP model, a number of factors were considered in constructing a multiple linear regression (MLR) model. The prediction capability was best when the Energy of Green's function, relative heart rate, and body surface area were used as predictors. Using the regression model that combined these predictors, the standard error of prediction for % VO2max and VO2max were +/- 4.9% and +/- 0.53 liter min-1, respectively. All regression models have been validated using non-steady state oxygen uptake and heart rate data measured for 100 validation subjects. In particular, the MLR model provided a robust, efficient and reliable statistical methodology capable of predicting relative workloads from submaximal exercises data collected in-situ. It is expected that the proposed multi-regression model will improve the accuracy of relative workload predictions from sub-maximal oxygen uptake data collected in-situ. This prediction method will be safer for unfit subjects and easier to use in general, compared to maximal testing protocols and restrictive lab requirements. The prediction technique developed in this research will help in better understanding the physical demands for today's workforce doing today's work and will have widespread application in identifying excessively demanding tasks so can be better matched to the abilities of subjects. Specifically, the technique presented will be instrumental for research focusing on: (1) Reliable evaluation of the workload a worker is subjected to, such that engineering or administrative interventions may be better contemplated by management to reduce the workload if needed; (2) Expanding job opportunities for women, older workers, and workers who are partially disabled, by placing in jobs according to their capabilities; (3) Evaluation of the effectiveness of rehabilitation programs for workers who previously suffered overexertion injuries. In general, this research is expected to have widespread application in identifying excessively demanding tasks so they can be better matched to the abilities of subjects. |
