Propagation of measurement errors in oxygen consumption algorithms.

Abstract

Oxygen consumption is often determined by combining the results of several different measurements. Because each of these measurements contains an error or uncertainty, the resultant determination of the oxygen consumption will also contain some combination of these measurement errors. We have performed a sensitivity analysis on several algorithms which are used to determine oxygen consumption. The analysis shows how the measurement errors combine in each algorithm and determine the final error in the oxygen consumption value. In one algorithm, 1 per cent measurement errors can produce a 32 per cent error in the determination of oxygen consumption whereas in another algorithm, under the same conditions, the 1 per cent measurement errors can cause an overall error as low as 2 per cent. The analysis shows how small improvements in measurement accuracy can yield large improvements in the accuracy of the oxygen consumption value and circumstances in which large improvements in measurement accuracy will produce only small increases in the accuracy of the oxygen consumption determination.