AUTHOR=Zrenner Markus , Heyde Christian , Duemler Burkhard , Dykman Solms , Roecker Kai , Eskofier Bjoern M. 

TITLE=Retrospective Analysis of Training and Its Response in Marathon Finishers Based on Fitness App Data

JOURNAL=Frontiers in Physiology

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2021.669884

DOI=10.3389/fphys.2021.669884

ISSN=1664-042X

ABSTRACT=\textit{Objective:} Finishing a marathon requires to first prepare for the 42.2\,km run. Current literature describes which training characteristics are related to marathon performance. However, which training is most effective in terms of a performance improvement remains unclear. \textit{Methods:} We conducted a retrospective analysis of training responses during the 16 weeks of training before a marathon. The analysis was performed on unsupervised fitness app data (Runtastic) from 6771 marathon finishers. We analyzed differences in training volume and intensity between 3 response and 3 marathon performance groups. Training response was quantified by the improvement of the velocity of 10\,km runs $\Delta v_{10}$ between the first and last 4 weeks of the training period. Response and marathon performance groups were classified by the 33.3rd and 66.6th percentile of $\Delta v_{10}$ and the marathon performance time, respectively. \textit{Results:} Subjects allocated in the faster marathon performance group showed systematically higher training volume and higher shares of training at low intensities. Only subjects in the moderate and high response group increased their training velocity continuously along the 16 weeks of training. \textit{Conclusion:} We demonstrate that a combination of maximized training volumes at low intensities, a continuous increase in average running speed up to the aimed marathon velocity and high intensity runs $\leq$5\,\% of the overall training volume was accompanied by an improved 10\,km performance which benefited the marathon performance as well. The study at hand proves that unsupervised workouts recorded with fitness apps can be a valuable data source for future sport science studies.