ABSTRACT
Every year millions of people, from all walks of life, spend months training to run a traditional marathon. For some it is about becoming fit enough to complete the gruelling 26.2 mile (42.2 km) distance. For others, it is about improving their fitness, to achieve a new personal-best finish-time. In this paper, we argue that the complexities of training for a marathon, combined with the availability of real-time activity data, provide a unique and worthwhile opportunity for machine learning and for recommender systems techniques to support runners as they train, race, and recover. We present a number of case studies-a mix of original research plus some recent results-to highlight what can be achieved using the type of activity data that is routinely collected by the current generation of mobile fitness apps, smart watches, and wearable sensors.
ABSTRACT
Completing a marathon usually requires at least 12-16 weeks of consistent training, but busy lifestyles, illness or injury, and motivational issues can all conspire to disrupt training. This study aims to investigate the frequency and performance cost of training disruptions, especially among recreational runners. Using more than 15 million activities, from 300,000 recreational runners who completed marathons during 2014-2017, we identified periods of varying durations up to 16 weeks before the marathon where runners experienced a complete cessation of training (so-called training disruptions). We identified runners who had completed multiple marathons including: (i) at least one disrupted marathon with a long training disruption of ≥ 7 days; and (ii) at least one undisrupted marathon with no training disruptions. Next, we calculated the performance cost of long training disruptions as the percentage difference between these disrupted and undisrupted marathon times, comparing the frequency and cost of training disruptions according to the sex, age, and ability of runner, and whether the disruptions occurred early or late in training. Over 50% of runners experienced short training disruptions up to and including 6 days, but longer disruptions were found to be increasingly less frequent among those who made it to race-day. Runners who experience longer training disruptions ( ≥ 7 days) suffer a finish-time cost of 5-8% compared to when the same runners experienced only short training disruptions (<7 days). While we found little difference (<5%) in the likelihood of disruptions-when comparing runners based on sex, age, ability, and the timing of a disruption-we did find significant differences in the the cost of disruptions (10-15%) among these groups. Two sample t -tests indicate that long training disruptions lead to a greater finish-time cost for males (5%) than females (3.5%). Faster runners also experience a greater finish-time cost (5.4%) than slower runners (2.6%). And, when disruptions occur late in training (close to race-day), they are associated with a greater finish-time cost (5.2%) than similar disruptions occurring earlier in training (4.4%). By parameterising and quantifying the cost of training disruptions, this work can help runners and coaches to better understand the relationship between training consistency and marathon performance. This has the potential to help them to better evaluate disruption risk during training and to plan for race-day more appropriately when disruptions do occur.
