Prediction Equations for Marathon Performance: A Systematic Review

Abstract

Purpose: Despite the volume of available literature focusing on marathon running and the prediction of performance, no single prediction equations exists that is accurate for all runners of varying experiences and abilities. Indeed the relative merits and utility of the existing equations remain unclear. Thus, the aim of this study was to collate, characterize, compare, and contrast all available marathon prediction equations. Methods: A systematic review was conducted to identify observational research studies outlining any kind of prediction algorithm for marathon performance. Results: Thirty-six studies with 114 equations were identified. Sixty-one equations were based on training and anthropometric variables, whereas 53 equations included variables that required laboratory tests and equipment. The accuracy of these equations was denoted via a variety of metrics; r2 values were provided for 68 equations (r2 = .10–.99), and an SEE was provided for 19 equations (SEE 0.27–27.4 min). Conclusion: Heterogeneity of the data precludes the identification of a single “best” equation. Important variables such as course gradient, sex, and expected weather conditions were often not included, and some widely used equations did not report the r2 value. Runners should therefore be wary of relying on a single equation to predict their performance.

Recommendations for marathon runners: on the application of recommender systems and machine learning to support recreational marathon runners

“…The single best performing predictor (GB-CB) is capable of estimating marathon times that are within 14.34 mins ( < 6% ) of actual race-times. As a baseline reference, this compares favourably with state-of-the-art marathon predictors (Keogh et al 2019), which are associated with an average error of 14.35 mins, based on a set of 19 different prediction formulas, many of which require costly, laboratorybased measures of fitness and ability. That our approach achieved similar prediction performance without the need for laboratory testing speaks to the potential of the proposed approach, and it provides a significant benefit for recreational runners by using their raw training data without the need for laboratory controlled testing.…”

Providing Explainable Race-Time Predictions and Training Plan Recommendations to Marathon Runners

“…This short paper builds on recent research [Berndsen et al 2019;Smyth 2019;Cunningham 2017b, 2018a] on the novel application of recommender systems to marathon running. We describe how to use the raw training data routinely collected by training apps such as Strava and RunKeeper to provide supports for runners during their training.…”

“…Most recreational runners train by following reasonably generic advice that is, at best, tailored to their projected or target finishtime. The recent availability of wearable sensors and mobile fitness applications promises to re-balance this state of affairs by facilitating the provision of tailored training advice Cau et al 2019;Monteiro-Guerra et al 2019;Mulas et al 2013], personalised motivational supports [Boratto et al 2017;Buttussi et al 2006;Hosseinpour and Terlutter 2019;Mulas et al 2011;, sophisticated performance analysis and prediction [Bartolucci and Murphy 2015;, and even in-race guidance [Berndsen et al 2019]. In this work, we build on these ideas by using raw training data to support marathon runners in two important ways: (1) by predicting their projected marathon time at different points in their training; and (2) by providing tailored training recommendations based on their recent training and their current performance goals.…”

TRAP: A Predictive Framework for Trail Running Assessment of Performance

“…Similarly, (Knechtle et al, 2014) regress finish times of half marathons on percent of body fat and running speed during training with linear regression to test and possibly improve the coefficients of these equations proposed in the literature (Knechtle et al, 2011;Rüst et al, 2011;Friedrich et al, 2014). In a review of the literature on equations for marathon prediction times, (Keogh et al, 2019) identify 36 studies with 114 different equations.…”