INTRODUCTION: Interpolation methods circumvent poor time resolution of breath-by-breath oxygen uptake (VO₂) kinetics at exercise onset. We report an interpolation-free approach to the improvement of poor time resolution in the analysis of VO₂ kinetics. METHODS: Noiseless and noisy (10% Gaussian noise) synthetic data were generated by Monte Carlo method from pre-selected parameters (Exact Parameters). Each data set comprised 10 (VO₂)-on transitions with noisy breath distribution within a physiological range. Transitions were superposed (no interpolation, None), then analysed by bi-exponential model. Fitted model parameters were compared with those from interpolation methods (average transition after Linear or Step 1-s interpolations), applied on the same data. Experimental data during cycling were also analysed. The 95% confidence interval around a line of parameters' equality was computed to analyse agreement between exact parameters and corresponding parameters of fitted functions. RESULTS: The line of parameters' equality stayed within confidence intervals for noiseless synthetic parameters with None, unlike Step and Linear, indicating that None reproduced Exact Parameters. Noise addition reduced differences among pre-treatment procedures. Experimental data provided lower phase I time constants with None than with Step. CONCLUSION: In conclusion, None revealed better precision and accuracy than Step and Linear, especially when phenomena characterized by time constants of <30 s are to be analysed. Therefore, we endorse the utilization of None to improve the quality of breath-by-breath [Formula: see text] data during exercise transients, especially when a double exponential model is applied and phase I is accounted for.
INTRODUCTION: Interpolation methods circumvent poor time resolution of breath-by-breath oxygen uptake (VO₂) kinetics at exercise onset. We report an interpolation-free approach to the improvement of poor time resolution in the analysis of VO₂ kinetics. METHODS: Noiseless and noisy (10% Gaussian noise) synthetic data were generated by Monte Carlo method from pre-selected parameters (Exact Parameters). Each data set comprised 10 (VO₂)-on transitions with noisy breath distribution within a physiological range. Transitions were superposed (no interpolation, None), then analysed by bi-exponential model. Fitted model parameters were compared with those from interpolation methods (average transition after Linear or Step 1-s interpolations), applied on the same data. Experimental data during cycling were also analysed. The 95% confidence interval around a line of parameters' equality was computed to analyse agreement between exact parameters and corresponding parameters of fitted functions. RESULTS: The line of parameters' equality stayed within confidence intervals for noiseless synthetic parameters with None, unlike Step and Linear, indicating that None reproduced Exact Parameters. Noise addition reduced differences among pre-treatment procedures. Experimental data provided lower phase I time constants with None than with Step. CONCLUSION: In conclusion, None revealed better precision and accuracy than Step and Linear, especially when phenomena characterized by time constants of <30 s are to be analysed. Therefore, we endorse the utilization of None to improve the quality of breath-by-breath [Formula: see text] data during exercise transients, especially when a double exponential model is applied and phase I is accounted for.
Authors: A Aliverti; B Kayser; M Cautero; R L Dellacà; P E di Prampero; C Capelli Journal: Respir Physiol Neurobiol Date: 2009-08-26 Impact factor: 1.931
Authors: Florent Baty; Christian Ritz; Signe Marie Jensen; Lukas Kern; Michael Tamm; Martin Hugo Brutsche Journal: PLoS One Date: 2017-11-08 Impact factor: 3.240