The Role of Artificial Intelligence in Sports Analytics: A Systematic Review and Meta-Analysis of Performance Trends
Роль искусственного интеллекта в спортивной аналитике: систематический обзор и метаанализ тенденций эффективности
2025-06-27
SCID: 54.1/z8wjmm8x
Discuss with AI
computer visiondeep learninginjury preventionmachine learningsports performance analysis
Figures from the paper
Abstract (AI)
This systematic review and meta-analysis investigates the application of artificial intelligence (AI) in sports performance analysis. Sixteen peer-reviewed studies spanning 13 distinct sports disciplines were included, employing a variety of AI techniques—from classical machine learning algorithms to advanced deep learning and computer vision models. Methods applied encompassed Convolutional Neural Networks (CNNs), Long Short-Term Memory (LSTM) networks, reinforcement learning, and predictive modeling architectures. The pooled average classification accuracy was 87.78% (95% CI: 82.66–92.90), although substantial heterogeneity was observed across studies (I2 = 93.75%). Computer vision and deep learning-based approaches were associated with higher performance metrics in several studies, particularly in movement-intensive sports such as tennis and basketball. Nevertheless, several challenges were identified, including lack of standardization in model evaluation, limited algorithmic transparency, and difficulties in generalizing findings from controlled laboratory environments to real-world competitive settings. The results underscore the promising role of AI in optimizing training protocols, supporting tactical decisions, and enhancing injury prevention strategies. Further research is warranted to address the ethical, methodological, and practical considerations surrounding the deployment of AI in sports contexts.
Key Findings
1
A systematic review synthesized 16 peer-reviewed studies across 13 sports disciplines applying classical machine learning, deep learning, computer vision, and reinforcement learning.
2
AI shows promise for optimizing training protocols, supporting tactical decisions, and enhancing injury prevention strategies in sports.
3
Computer vision and deep learning approaches achieved higher performance metrics in several studies, especially for movement-intensive sports such as tennis and basketball.
4
Evidence is constrained by inconsistent evaluation standards, limited algorithmic transparency, and uncertain generalizability from laboratory settings to competitive environments.
5
The pooled average AI classification accuracy was 87.78% (95% CI: 82.66–92.90), with substantial between-study heterogeneity (I² = 93.75%).
Research Object
AI applications in sports performance analysis across 13 sports disciplines
Research Subject
Classification accuracy, performance trends, and practical effectiveness of AI-based analysis methods, including their generalizability and evaluation challenges
Publication Details
Publication Date
2025-06-27
Journal
Publisher
ISSN
Cited by
119
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai3
The PRISMA 2020 statement: an updated guideline for reporting systematic reviews2021
AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both2017
Review on Wearable Technology in Sports: Concepts, Challenges and Opportunities2023