A semi-automatic tool to quantify training load in endurance athletes

Abstract Funding Acknowledgements Type of funding sources: None. Background Despite the widespread use of wearables in endurance sports allowing the measurements needed for quantification of the training load (TL), studies in sports cardiology consistently use less reliable questionnaires to assess TL.[1] This scant use of activity trackers to determine TL can be explained by the large workload for data analysis. Purpose We aimed to develop a semi-automated post-processing tool that allows quantifying TL, based on heart rate (HR) monitoring by commercial wearables, for future use in clinical and epidemiological studies. We compared the calculated maximal HR automatically derived from HR monitoring during endurance training with the maximal HR obtained during incremental exercise testing in the lab. We describe TL calculated by our tool in young and older endurances athletes and compare it with the TL obtained via questionnaires. Methods Training data was collected in two multicenter studies. Master@Heart is a cohort study of endurance trained middle-aged men between 45 and 70 years.[2] Pro@Heart is a prospective study in young endurance athletes between 16 and 23 years.[3] For the purpose of this substudy, 44 athletes (38 ± 17 years) that had recorded all their training sessions using a chest-worn HR monitor for one year were included. Using an in-house developed pipeline in R (R core Team, Vienna, Austria), maximal HR and TL were calculated from training data. The maximal HR measured in the field (Tangent MaxHR) was then compared with the lab-measured maximal HR (Lab MaxHR). Both Lucia training impulse (LuTRIMP) and Edwards TRIMP (eTRIMP) were compared as measures of TL based on HR-based intensity zones. LuTRIMP, based on individually based exercise intensity zones, and eTRIMP based on the Tangent MaxHR. Participants filled in a standard exercise questionnaire to determine the amount of training sessions and training hours per week. Results Lab MaxHR was lower than Tangent MaxHR (183 ± 12.2bpm vs. 188.3 ± 13.0bpm; Figure 1 A), but correlated strongly (r=0.81, P<0.01; Figure 1 B) with low mean bias (-5.3bpm, LOA±15.4; Figure 1 C). LuTRIMP and eTRIMP correlated very strong (r=0.93, P<0.01). Pro@Heart athletes performed a higher TL (1730 ± 644 vs. 1190 ± 314 AU/week, P<0.01; Figure 2 A), duration (12.2 ± 4.6 hours per week vs. 7.9 ± 2.4, P<0.01; Figure 2 B) and trained more frequent (5.9 ± 1.7 vs. 4.0 ± 1.0 sessions per week, P<0.01; Figure 2 C) than Master@Heart athletes. Neither the reported training sessions nor the training hours in the questionnaires correlated with the measured training sessions (r=0.20) and training hours respectively (r=0.12). Conclusion Our tool allows to semi-automatically quantify different TL parameters in endurance athletes from a wide age range. The use of questionnaires to determine TL is inaccurate. TL calculated using eTRIMP is comparable to TL calculated using LuTRIMP. Since eTRIMP does not require an exercise test, it is more easily implementable in large studies.