State-of-the art concepts and future directions in modelling oxygen consumption and lactate concentration in cycling exercise

Bioenergetic models are used in cycling to estimate the acute physiological response in terms of oxygen consumption (V˙\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{\text{V}}}$$\end{document}O2) and lactate concentration ([La]). First, our aim is to review the bioenergetic modelling literature, presenting historical evolution of concepts, techniques and related limitations. Second, our aim is to discuss how and where new approaches can stem and evolve. This is a narrative review, where different modelling solutions are compared and qualitatively discussed. First, the principal features of the V˙\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{\text{V}}}$$\end{document}O2 and [La] kinetics are presented, and then the models available in the literature are compared in light of what aspects of the physiological responses they can describe. Currently, models can detect most features of V˙\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\dot{\text{V}}}$$\end{document}O2 and [La] kinetics, but no single existing model appears appropriate for every exercising conditions. Limitations hindering the creation of an ultimate model are: the large variability of an exercise, the required mathematical complexity, and lack of reliable physiological data. To overcome these issues, new modelling solutions are being explored in the emerging AI technologies. However, in AI-models, parameters do not have direct physiological meaning and require massive amounts of experimental data for parameter calibration. Despite the great efforts made by model developers and exercise physiologists, universal modelling solutions for the variety of potential exercising conditions remain unavailable. At present, further research is needed to assess the accuracy and predictive power of AI models to move the method forward in our field, as it is being done so in many others.