Breath-by-breath pulmonary O2 uptake kinetics: effect of data processing on confidence in estimating model parameters

To improve the signal-to-noise ratio of breath-by-breath pulmonary O2 uptake (V̇O2p) data, it is common practice to perform multiple step transitions, which are subsequently processed to yield an ensemble-averaged profile. The effect of different data-processing techniques on phase II V̇O2p kinetic parameter estimates (V̇O2p amplitude, time delay and phase II time constant (τV̇O2p)] and model confidence [95% confidence interval (CI95)] was examined. Young (n = 9) and older men (n = 9) performed four step transitions from a 20 W baseline to a work rate corresponding to 90% of their estimated lactate threshold on a cycle ergometer. Breath-by-breath V̇O2p was measured using mass spectrometry and volume turbine. Mono-exponential kinetic modelling of phase II V̇O2p data was performed on data processed using the following techniques: (A) raw data (trials time aligned, breaths of all trials combined and sorted in time); (B) raw data plus interpolation (trials time aligned, combined, sorted and linearly interpolated to second by second); (C) raw data plus interpolation plus 5 s bin averaged; (D) individual trial interpolation plus ensemble averaged [trials time aligned, linearly interpolated to second by second (technique 1; points joined by straight-line segments), ensemble averaged]; (E) 'D' plus 5 s bin averaged; (F) individual trial interpolation plus ensemble averaged [trials time aligned, linearly interpolated to second by second (technique 2; points copied until subsequent point appears), ensemble averaged]; and (G) 'F' plus 5 s bin averaged. All of the model parameters were unaffected by data-processing technique; however, the CI95 for τV̇O2p in condition 'D' (4 s) was lower (P < 0.05) than the CI95 reported for all other conditions (5-10 s). Data-processing technique had no effect on parameter estimates of the phase II V̇O2p response. However, the narrowest interval for CI95 occurred when individual trials were linearly interpolated and ensemble averaged.