Frontiers in Physiology (Jun 2024)

Identification of doping suspicions through artificial intelligence-powered analysis on athlete’s performance passport in female weightlifting

  • Hyunji Ryoo,
  • Samuel Cho,
  • Taehan Oh,
  • YuSik Kim,
  • Sang-Hoon Suh

DOI
https://doi.org/10.3389/fphys.2024.1344340
Journal volume & issue
Vol. 15

Abstract

Read online

IntroductionDoping remains a persistent concern in sports, compromising fair competition. The Athlete Biological Passport (ABP) has been a standard anti-doping measure, but confounding factors challenge its effectiveness. Our study introduces an artificial intelligence-driven approach for identifying potential doping suspicious, utilizing the Athlete’s Performance Passport (APP), which integrates both demographic profiles and performance data, among elite female weightlifters.MethodsAnalyzing publicly available performance data in female weightlifting from 1998 to 2020, along with demographic information, encompassing 17,058 entities, we categorized weightlifters by age, body weight (BW) class, and performance levels. Documented anti-doping rule violations (ADRVs) cases were also retained. We employed AI-powered algorithms, including XGBoost, Multilayer Perceptron (MLP), and an Ensemble model, which integrates XGBoost and MLP, to identify doping suspicions based on the dataset we obtained.ResultsOur findings suggest a potential doping inclination in female weightlifters in their mid-twenties, and the sanctioned prevalence was the highest in the top 1% performance level and then decreased thereafter. Performance profiles and sanction trends across age groups and BW classes reveal consistently superior performances in sanctioned cases. The Ensemble model showcased impressive predictive performance, achieving a 53.8% prediction rate among the weightlifters sanctioned in the 2008, 2012, and 2016 Olympics. This demonstrated the practical application of the Athlete’s Performance Passport (APP) in identifying potential doping suspicions.DiscussionOur study pioneers an AI-driven APP approach in anti-doping, offering a proactive and efficient methodology. The APP, coupled with advanced AI algorithms, holds promise in revolutionizing the efficiency and objectivity of doping tests, providing a novel avenue for enhancing anti-doping measures in elite female weightlifting and potentially extending to diverse sports. We also address the limitation of a constrained set of APPs, advocating for the development of a more accessible and enriched APP system for robust anti-doping practices.

Keywords