Development of a New Predictive Model Enabling the Assessment of the Anaerobic Capacity from a Standard Aerobic Exercise Stress Test

Background: Graded exercise stress test (GXT) is a common and widely accepted method to assess the aerobic capacity and to predict maximal oxygen consumption and cardiorespiratory fitness. The first part of the GXT is aerobic while the last part includes the anaerobic component; meaning, utilization of ATP occurs mostly via anaerobic metabolism. The anaerobic capacity can be determined by performing the Wingate anaerobic test (WAnT). Aim: To predict the anaerobic capacity from a GXT, enabling the use of a single exercise stress test, for healthy and diseased patients. Methods: Designing and developing a new computational simulative predictive model enabling to predict the anaerobic capacity from a GXT using MATLAB software. We collected data from cardiopulmonary exercise stress test (CPET) and from the WAnT, from within the same individuals. We used a greedy heuristic algorithm while Xi denote the aerobic variables that the algorithm add to the regressor at each step, the algorithm choose the variable that provides the best spearman correlation with the prediction (of the anaerobic variables). For each simulation we chose a different set of aerobic variables. Results: A prediction of r=0.8 was observed in the peak power by using 4 different aerobic variables: VCO_2max, max exercise time, max tidal volume, and max breathing frequency. A prediction of r=0.82 was observed the mean power output by using 4 other aerobic variables: VO₂, VCO₂, FeCO₂ and respiratory ratio (VCO₂/VO₂). By using additional aerobic variables such as the slop up to the anaerobic threshold of the VCO2 versus VO2 graph, we could further increase the prediction correlation to 0.91. Conclusion: Prediction of the anaerobic variables is feasible by using advanced statistical prediction tools, hence justifying for further investigation for this relationship in diverse individuals and patients` population.