Planning a sports training program using Adaptive Particle Swarm Optimization with emphasis on physiological constraints

BMC Research Notes Kumyaito et al. BMC Res Notes (2018) 11:9 https://doi.org/10.1186/s13104-017-3120-9 Open Access RESEARCH NOTE Planning a sports training program using Adaptive Particle Swarm Optimization with emphasis on physiological constraints Nattapon Kumyaito1, Preecha Yupapin3,4* and Kreangsak Tamee1,2* Abstract Objective: An effective training plan is an important factor in sports training to enhance athletic performance. A poorly considered training plan may result in injury to the athlete, and overtraining. Good training plans normally require expert input, which may have a cost too great for many athletes, particularly amateur athletes. The objectives of this research were to create a practical cycling training plan that substantially improves athletic performance while satisfying essential physiological constraints. Adaptive Particle Swarm Optimization using ɛ-constraint methods were used to formulate such a plan and simulate the likely performance outcomes. The physiological constraints considered in this study were monotony, chronic training load ramp rate and daily training impulse. Results: A comparison of results from our simulations against a training plan from British Cycling, which we used as our standard, showed that our training plan outperformed the benchmark in terms of both athletic performance and satisfying all physiological constraints. Keywords: Sports training plan, Training-performance interaction models, Physiological constraints, Particle Swarm Optimization Introduction Sports training is a process intended to improve athletic performance by means of developing both the physical and mental conditions of the athlete. Sports training can, however, have an opposite and detrimental effect to that intended. A positive result would be an improvement of physical fitness while a negative result would be an increase in fatigue. Training-performance interaction models [1, 2] have defined the relationship between sports training programs and the intended results. Successful athletes utilize a training-performance interaction model to plan wisely in advance of training and rest at appropriate intervals to maximize physical fitness improvement while minimizing the chronic fatigue. While high performance is the ultimate goal of an athlete, *Correspondence: ; 1 Department of Computer Science and Information Technology, Faculty of Science, Naresuan University, Phitsanulok 65000, Thailand 4 Faculty of Electrical & Electronics Engineering, Ton Duc Thang University, District 7, Ho Chi Minh City 700000, Vietnam Full list of author information is available at the end of the article detrimental outcomes affecting athletic performance are likely when physiological constraints are not considered, resulting in the risk of overtraining. These physical constraints include training monotony [3], and chronic training load (CTL) ramp rate [4]. While there are a number of apps available [5, 6] their cost is substantial, but there is little evidence to suggest that their use enables a substantial rise of athletic performance. Some research has been undertaken on the generation of sports training plans simply, quickly and efficiently [7–11]. However, while these ‘simple’ training plans have been shown to improve athletic performance, the systems are somewhat impractical with no mechanisms to handle or manipulate necessary constraints. Most of this related work discusses the search for a global optimal solution to scheduling sports training programs. According to [12], particle swarm optimization (PSO) is the most prevalent swarm intelligence-based optimization algorithm. This algorithm has significant advantages over previous optimization schemes [13] and © The Author(s) 2018. This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/ publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Kumyaito et al. BMC Res Notes (2018) 11:9 has been successfully extended to constrained optimization [14]. However, there is little previous research to be found that applies PSO to the construction of optimal training programs. In [15] a modified form of PSO that applies ɛ constraint methods, referred to as adaptive PSO, was used to generate an optimal cycling training plan using simulated athlete data. The result is a cycling training plan purposed to enhance athletic performance by taking into account physiological constraints such as training monotony [3], CTL ramp rate [4] and daily training load. The latter physiological constraint was derived from the British Cycling’s training plan for inclusion in our research. Main text Problem formulation This section defines a training plan for a simulated athlete. An 8-week training plan is considered as appropriate preparation for endurance sports [16]. The training plan consists of 56 training sessions each of which introduces a daily training goal by means of average heart rate (HR) in units of beats per minute (bpm) and activity duration (D) in minutes (min). The boundary of the HR data was extracted from the simulated athlete, 35 years old male who has 51 bpm at resting state, 165 bpm as FTHR and 189 bpm as maximum HR. The lower bound is resting HR and the upper bound is the maximum HR. For ease of use, the training plan should be personalized to the individual athlete. Functional Threshold Heart Rate (FTHR) is considered as a factor that reflects the current level of the athlete’s physical fitness. The Coggan’s training zone [17] corresponding to the simulated athlete’s FTHR is adopted. The boundary of training duration was observed from the training behavior of national athletes which ranged between 30 min and 5 h. The classified heart rate training zones and duration of training zones are illustrated in Table 1. Particles encoding The PSO technique begins by randomly initiating the number of potential training plans as a collection of particles or a swarm. Each particle is encoded into a 112-dimension array from a given training plan of 56 sessions. Each training session has HR and duration data. Thus, at iteration r, the ith solution that includes M training sessions can be expressed as   r r r r r r r r , Di,M , . . . , HRi,M , Di,3 HRi,3 , Di,2 , HRi,2 , Di,1 Tir = HRi,1 A full codification of a particle can then be written as HR1 D1 HR2 D2 HR3 D3 . . . HR56 D56 Page 2 of 6 Table 1 Heart rate zone and duration zone HR zone HR (bpm) HR (% of FTHR) Duration zone Duration (min) 0 51–81 30.91–49.09 0 30 1 82–112 49.7–67.88 1 60 2 113–124 68.4 (...truncated)