Article: A reinforcement learning - enabled system for personalised sports training plan generation Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.50 pp.72 - 87 Abstract: Personalised sports training has emerged as a critical component in optimising athletic performance and minimising injury risks. Nevertheless, conventional approaches predominantly depend on coaches' subjective expertise, which often falls short in delivering dynamically precise adaptations. In response, this study introduces a reinforcement learning-based framework for generating individualised training regimens. By formulating the training process as a Markov decision process, the system enables an intelligent agent to interact with a simulated training environment, producing optimised training actions derived from real-time user status information. Evaluations conducted on the public FitRec dataset indicate that, relative to conventional baseline techniques, the proposed system yields an average improvement of 15% in predicted performance indicators, while concurrently lowering the incidence of training overload by 30%. These findings highlight the potential of the proposed framework as an effective new paradigm for automated and individualised sports science training. Inderscience Publishers - linking academia, business and industry through research

Title: A reinforcement learning - enabled system for personalised sports training plan generation

Authors: Linli Zhou; Kaili Zhou

Addresses: School of Physical Education, Sichuan Technology and Business University, Meishan, 620000, China ' Chengdu Qingling Qi'an Technology Co., Ltd., Sichuan Technology and Business University, Chengdu, 610051, China

Abstract: Personalised sports training has emerged as a critical component in optimising athletic performance and minimising injury risks. Nevertheless, conventional approaches predominantly depend on coaches' subjective expertise, which often falls short in delivering dynamically precise adaptations. In response, this study introduces a reinforcement learning-based framework for generating individualised training regimens. By formulating the training process as a Markov decision process, the system enables an intelligent agent to interact with a simulated training environment, producing optimised training actions derived from real-time user status information. Evaluations conducted on the public FitRec dataset indicate that, relative to conventional baseline techniques, the proposed system yields an average improvement of 15% in predicted performance indicators, while concurrently lowering the incidence of training overload by 30%. These findings highlight the potential of the proposed framework as an effective new paradigm for automated and individualised sports science training.

Keywords: reinforcement learning; RL; personalised sports training; Markov decision process; MDP; reward function; FitRec dataset.

DOI: 10.1504/IJICT.2025.151069

International Journal of Information and Communication Technology, 2025 Vol.26 No.50, pp.72 - 87

Received: 18 Oct 2025
Accepted: 15 Nov 2025
Published online: 12 Jan 2026 *

Title: A reinforcement learning - enabled system for personalised sports training plan generation

Keep up-to-date