Article: Recurrent neural network-based analysis of influencing factors on sports event communication effectiveness from the intelligent media perspective Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.70 pp.1 - 21 Abstract: To address the issues of insufficient real-time performance and accuracy in the communication of sports competitions, this study proposes a dynamic LSTM for sports event communication effectiveness (DLSTM-SEC) model based on dynamic long short-term memory (LSTM) to optimise the communication effect of sports events. The model combines compressed sensing feature selection and the momentum-improved adaptive moment estimation (Adam) algorithm to achieve efficient capture and rapid response to key events. Experimental results show that the DLSTM-SEC model achieves a test accuracy of 91.1% with an average loss value below 0.35. In terms of communication delay, 95% of the latency is controlled within 250 ms, and the 99th-percentile delay is no more than 450 ms. Under abnormal load conditions, the packet loss recovery rate remains above 92%. The results demonstrate that the model has stable real-time communication capability and data adaptability in complex dynamic scenarios, and can support the dynamic optimisation and efficient operation of sports event communication in an intelligent media environment. This study aims to provide a reliable competition communication tool for sports event organisers, intelligent media platforms, and audiences to improve real-time information acquisition and user experience, and realise more efficient event content push and interactive feedback. Inderscience Publishers - linking academia, business and industry through research

Title: Recurrent neural network-based analysis of influencing factors on sports event communication effectiveness from the intelligent media perspective

Authors: Hongwei Zhang; Minghua Jiang

Addresses: School of Sports Management and Communication of Capital University of Physical Education and Sports, Beijing, 100191, China ' Guanghua School of Management of Peking University, Beijing, 100091, China

Abstract: To address the issues of insufficient real-time performance and accuracy in the communication of sports competitions, this study proposes a dynamic LSTM for sports event communication effectiveness (DLSTM-SEC) model based on dynamic long short-term memory (LSTM) to optimise the communication effect of sports events. The model combines compressed sensing feature selection and the momentum-improved adaptive moment estimation (Adam) algorithm to achieve efficient capture and rapid response to key events. Experimental results show that the DLSTM-SEC model achieves a test accuracy of 91.1% with an average loss value below 0.35. In terms of communication delay, 95% of the latency is controlled within 250 ms, and the 99th-percentile delay is no more than 450 ms. Under abnormal load conditions, the packet loss recovery rate remains above 92%. The results demonstrate that the model has stable real-time communication capability and data adaptability in complex dynamic scenarios, and can support the dynamic optimisation and efficient operation of sports event communication in an intelligent media environment. This study aims to provide a reliable competition communication tool for sports event organisers, intelligent media platforms, and audiences to improve real-time information acquisition and user experience, and realise more efficient event content push and interactive feedback.

Keywords: long short-term memory; LSTM; Adam algorithm; sports events; intelligent media communication; interactive feedback.

DOI: 10.1504/IJICT.2026.154392

International Journal of Information and Communication Technology, 2026 Vol.27 No.70, pp.1 - 21

Received: 31 Dec 2025
Accepted: 04 Mar 2026
Published online: 25 Jun 2026 *

Title: Recurrent neural network-based analysis of influencing factors on sports event communication effectiveness from the intelligent media perspective

Keep up-to-date